Add a link to my WIP novel, Curse of the Hollowtides

This page is already moved out of this repo, so the links to it just cause trouble.

.gitignore

@@ -1,2 +1,5 @@
-/Gemfile.lock
+__pycache__/
-_site/
+/output/
+/dat-version/
+/ipfs-version/
+/gergelypolonkaieu_site.egg-info/

.hyde.el

@@ -1,2 +0,0 @@
-(setq hyde/git/remote "origin"
-      hyde/git/remote-branch "master")

404.md

@@ -1,11 +0,0 @@
----
-layout: page
-title: Not Found
-permalink: /404.html
----
-
-The page you are looking for is not here. Maybe it was but I have removed it. Most likely it was intentionally. If you think I made a mistake, please tell me.
-
-{% if page.url contains '/akarmi' %}
-If you are looking for the pictures that used to be here, you should definitely contact me. For reasons.
-{% endif %}

CNAME

@@ -1 +0,0 @@
-gergely.polonkai.eu

Gemfile

@@ -1,3 +0,0 @@
-source 'https://rubygems.org'
-
-gem 'github-pages'

Makefile

@@ -0,0 +1,74 @@
+PY?=python3
+PELICAN?=pelican
+PELICANOPTS=
+
+BASEDIR=$(CURDIR)
+INPUTDIR=$(BASEDIR)/content
+OUTPUTDIR=$(BASEDIR)/output
+CONFFILE=$(BASEDIR)/pelicanconf.py
+PUBLISHCONF=$(BASEDIR)/publishconf.py
+
+DEBUG ?= 0
+ifeq ($(DEBUG), 1)
+	PELICANOPTS += -D
+endif
+
+RELATIVE ?= 0
+ifeq ($(RELATIVE), 1)
+	PELICANOPTS += --relative-urls
+endif
+
+help:
+	@echo 'Makefile for a pelican Web site                                           '
+	@echo '                                                                          '
+	@echo 'Usage:                                                                    '
+	@echo '   make html                           (re)generate the web site          '
+	@echo '   make clean                          remove the generated files         '
+	@echo '   make regenerate                     regenerate files upon modification '
+	@echo '   make publish                        generate using production settings '
+	@echo '   make serve [PORT=8000]              serve site at http://localhost:8000'
+	@echo '   make serve-global [SERVER=0.0.0.0]  serve (as root) to $(SERVER):80    '
+	@echo '   make devserver [PORT=8000]          serve and regenerate together      '
+	@echo '   make ssh_upload                     upload the web site via SSH        '
+	@echo '   make rsync_upload                   upload the web site via rsync+ssh  '
+	@echo '                                                                          '
+	@echo 'Set the DEBUG variable to 1 to enable debugging, e.g. make DEBUG=1 html   '
+	@echo 'Set the RELATIVE variable to 1 to enable relative urls                    '
+	@echo '                                                                          '
+
+html:
+	$(PELICAN) $(INPUTDIR) -o $(OUTPUTDIR) -s $(CONFFILE) $(PELICANOPTS)
+
+clean:
+	[ ! -d $(OUTPUTDIR) ] || rm -rf $(OUTPUTDIR)
+
+regenerate:
+	$(PELICAN) -r $(INPUTDIR) -o $(OUTPUTDIR) -s $(CONFFILE) $(PELICANOPTS)
+
+serve:
+ifdef PORT
+	$(PELICAN) -l $(INPUTDIR) -o $(OUTPUTDIR) -s $(CONFFILE) $(PELICANOPTS) -p $(PORT)
+else
+	$(PELICAN) -l $(INPUTDIR) -o $(OUTPUTDIR) -s $(CONFFILE) $(PELICANOPTS)
+endif
+
+serve-global:
+ifdef SERVER
+	$(PELICAN) -l $(INPUTDIR) -o $(OUTPUTDIR) -s $(CONFFILE) $(PELICANOPTS) -p $(PORT) -b $(SERVER)
+else
+	$(PELICAN) -l $(INPUTDIR) -o $(OUTPUTDIR) -s $(CONFFILE) $(PELICANOPTS) -p $(PORT) -b 0.0.0.0
+endif
+
+
+devserver:
+ifdef PORT
+	$(PELICAN) -lr $(INPUTDIR) -o $(OUTPUTDIR) -s $(CONFFILE) $(PELICANOPTS) -p $(PORT)
+else
+	$(PELICAN) -lr $(INPUTDIR) -o $(OUTPUTDIR) -s $(CONFFILE) $(PELICANOPTS)
+endif
+
+publish:
+	$(PELICAN) $(INPUTDIR) -o $(OUTPUTDIR) -s $(PUBLISHCONF) $(PELICANOPTS)
+
+
+.PHONY: html help clean regenerate serve serve-global devserver publish

README.md

@@ -1,2 +1,18 @@
-gergelypolonkai.github.io
+# Website of Gergely Polonkai
-=========================
+
+## Publishing to the web
+
+```
+poetry run pelican content
+rsync --archive --verbose --human-readable --delete output/ /data/gergely.polonkai.eu/html
+restorecon -vr /data/gergely.polonkai.eu/html
+```
+
+## Publishing to IPFS
+
+```
+poetry run pelican -s ipfspublish.py -o ipfs-version content
+cd ipfs-version
+ipfs add -r .
+ipfs name publish --key=gergely.polonkai.eu /ipfs/XXX
+```

_config.yml

@@ -1,19 +0,0 @@
-# Site settings
-title: Gergely Polonkai
-email: gergely@polonkai.eu
-description: "developer, systems engineer and administrator"
-baseurl: ""
-url: "http://gergely.polonkai.eu"
-timezone: Europe/Budapest
-name: Gergely Polonkai
-paginate: 10
-paginate_path: "/blog/page/:num"
-exclude: ['README.md', 'Gemfile', 'Gemfile.lock', 'CNAME', ".hyde.el"]
-include: ['.well-known']
-plugins:
-  - jekyll-gist
-  - jekyll-paginate
-
-# Build settings
-markdown: kramdown
-permalink: pretty

_data/contacts.yaml

@@ -1,42 +0,0 @@
-- text: E-mail
-  link: mailto:gergely@polonkai.eu
-  image: email.png
-- text: Stack Exchange
-  link: http://stackexchange.com/users/1369500/gergelypolonkai
-  image: stackexchange.png
-- text: LinkedIn
-  link: http://www.linkedin.com/in/gergelypolonkai
-  image: linkedin.png
-- text: Skype
-  link: skype:gergely.polonkai
-  image: skype.png
-- text: Facebook
-  link: http://facebook.com/Polesz
-  image: facebook.png
-- text: Google+
-  link: https://plus.google.com/+GergelyPolonkai/about
-  image: google_plus.png
-- text: Twitter
-  link: http://twitter.com/GergelyPolonkai
-  image: twitter.png
-- text: Tumblr
-  link: http://gergelypolonkai.tumblr.com
-  image: tumblr.png
-- text: deviantArt
-  link: http://gergelypolonkai.deviantart.com
-  image: deviantart.png
-- text: Hashnode
-  link: https://hashnode.com/@gergelypolonkai
-  image: hashnode.png
-- text: Keybase
-  link: https://keybase.io/gergelypolonkai
-  image: keybase.png
-- text: Liberapay
-  link: https://liberapay.com/gergelypolonkai
-  image: liberapay.png
-- text: Mastodon
-  link: https://social.polonkai.eu/@gergely
-  image: mastodon.png
-- text: Pay me a coffee
-  link: https://paypal.me/GergelyPolonkai/250
-  image: paypal.png

_drafts/gtk-actionable-in-action.md

@@ -1,15 +0,0 @@
----
-layout: post
-title: "GtkActionable in action"
-author:
-  name: "Gergely Polonkai"
-  email: "gergely@polonkai.eu"
----
-
-I have seen several people (including myself) struggling with
-disabling/enabling menu items, toolbar buttons and similar UI
-interfaces based on different conditions. It gets even worse if there
-are multiple representations of the same action in the same
-application, e.g. a menu item and a toolbar button exists for the same
-action. But with GTK+ 3.4, we have GtkAction, which is exactly for
-this kind of situations.

_drafts/measuring-coverage-with-codecov-for-libtool-projects.md

@@ -1,17 +0,0 @@
----
-layout: post
-title: "Measuring code coverage with codecov for libtool projects"
-author:
-  name: "Gergely Polonkai"
-  email: "gergely@polonkai.eu"
----
-
-I have recently found [codecov][https://codecov.io/]; they offer free
-services for public GitHub projects. As I have recently started writing
-tests for my SWE-GLib project, I decided to give it a go. Things are not
-this easy if you use GNU Autotools and libtool, though…
-
-The problem here is that these tools generate output under `src/.libs/`
-(given that your sources are under `src/`) and `gcov` has hard times
-finding the coverage data files. Well, at least in the codecov
-environment, it works fine on my machine.

_drafts/writing-a-gnome-shell-extension.md

@@ -1,326 +0,0 @@
----
-layout: post
-title: "Writing a GNOME Shell extension"
----
-
-I could not find a good tutorial on how to write a GNOME Shell
-extension. There is a so called step by step
-[instruction list](https://wiki.gnome.org/Projects/GnomeShell/Extensions/StepByStepTutorial)
-on how to do it, but it has its flaws, including grammar and clearance.
-As I wanted to create an extension for my SWE GLib library to display
-the current position of some planets, I dug into existing (and working)
-extensions’ source code and made up something. Comments welcome!
-
----
-
-GNOME Shell extensions are written in JavaScript and are interpreted
-by [GJS](https://wiki.gnome.org/action/show/Projects/Gjs). Using
-introspected libraries from JavaScript is not a problem for me (see
-SWE GLib’s
-[Javascript example](https://github.com/gergelypolonkai/swe-glib/blob/master/examples/basic.js);
-it’s not beautiful, but it’s working), but wrapping your head around
-the Shell’s concept can take some time.
-
-The Shell is a Clutter stage, and all the buttons (including the
-top-right “Activities” button) are actors on this stage. You can add
-practically anything to the Shell panel that you can add to a Clutter
-stage.
-
-The other thing to remember is the lifecycle of a Shell
-extension. After calling `init()`, there are two ways forward: you
-either use a so called extension controller, or plain old JavaScript
-functions `enable()` and `disable()`; I will go on with the former
-method for reasons discussed later.
-
-If you are fine with the `enable()`/`disable()` function version, you
-can ease your job with the following command:
-
-```
-gnome-shell-extension-tool --create-extension
-```
-
-This will ask you a few parameters and create the necessary files for
-you. On what these parameters should look like, please come with me to
-the next section.
-
-## Placement and naming
-
-Extensions reside under `$HOME/.local/share/gnome-shell/extensions`,
-where each of them have its own directory. The directory name has to be
-unique, of course; to achieve this, they are usually the same as the
-UUID of the extension.
-
-The UUID is a string of alphanumeric characters, with some extras added.
-Generally, it should match this regular expression:
-`^[-a-zA-Z0-9@._]+$`. The convention is to use the form
-`extension-name@author-id`, e.g. `Planets@gergely.polonkai.eu`. Please
-see
-[this link](https://wiki.gnome.org/Projects/GnomeShell/Extensions/UUIDGuidelines)
-for some more information about this.
-
-## Anatomy of an extension
-
-Extensions consist of two main parts, `metadata.json` and
-`extension.js`.
-
-The `metadata.json` file contains compatibility information and, well,
-some meta data:
-
-```json
-{
-    "shell-version": ["3.18"],
-    "uuid": "planets@gergely.polonkai.eu",
-    "name": "Planets",
-    "description": "Display current planet positions"
-}
-```
-
-Here, `shell-version` must contain all versions of GNOME Shell that is
-known to load and display your extension correctly. You can insert minor
-versions here, like I did, or exact version numbers, like `3.18.1`.
-
-In the `extension.js` file, which contains the actual extension code,
-the only thing you actually need is an `init()` function:
-
-```javascript
-function init(extensionMeta) {
-    // Do whatever it takes to initialize your extension, like
-    // initializing the translations. However, never do any widget
-    // magic here yet.
-
-    // Then return the controller object
-    return new ExtensionController(extensionMeta);
-}
-```
-
-## Extension controller
-
-So far so good, but what is this extension controller thing? It is an
-object which is capable of managing your GNOME Shell extension. Whenever
-the extension is loaded, its `enable()` method is called; when the
-extension is unloaded, you guessed it, the `disable()` method gets
-called.
-
-```javascript
-function ExtensionController(extensionMeta) {
-    return {
-        extensionMeta: extensionMeta,
-        extension: null,
-
-        enable: function() {
-            this.extension = new PlanetsExtension(this.extensionMeta);
-
-            Main.panel.addToStatusArea("planets",
-                                       this.extension,
-                                       0, "right");
-        },
-
-        disable: function() {
-            this.extension.actor.destroy();
-            this.extension.destroy();
-
-            this.extension = null;
-        }
-    }
-}
-```
-
-This controller will create a new instance of the `PlanetsExtension`
-class and add it to the panel’s right side when loaded. Upon
-unloading, the extension’s actor gets destroyed (which, as you will
-see later, gets created behind the scenes, not directly by us),
-together with the extension itself. Also, for safety measures, the
-extension is set to `null`.
-
-## The extension
-
-The extension is a bit more tricky, as, for convenience reasons, it
-should extend an existing panel widget type.
-
-```javascript
-function PlanetsExtension(extensionMeta) {
-    this._init(extensionMeta);
-}
-
-PlanetsExtension.prototype = {
-    __proto__ = PanelMenu.Button.prototype,
-
-    _init: function(extensionMeta) {
-        PanelMenu.Button.prototype._init.call(this, 0.0);
-
-        this.extensionMeta = extensionMeta;
-
-        this.panelContainer = new St.BoxLayout({style_class: 'panel-box'});
-        this.actor.add_actor(this.panelContainer);
-        this.actor.add_style_class_name('panel-status-button');
-
-        this.panelLabel = new St.Label({
-            text: 'Loading',
-            y_align: Clutter.ActorAlign.CENTER
-        });
-
-        this.panelContainer.add(this.panelLabel);
-    }
-};
-```
-
-Here we extend the Button class of panelMenu, so we will be able to do
-some action upon activate.
-
-The only parameter passed to the parent’s `_init()` function is
-`menuAlignment`, with the value `0.0`, which is used to position the
-menu arrow. (_Note: I cannot find any documentation on this, but it
-seems that with the value `0.0`, a menu arrow is not added._)
-
-The extension class in its current form is capable of creating the
-actual panel button displaying the text “Loading” in its center.
-
-## Loading up the extension
-
-Now with all the necessary import lines added:
-
-```javascript
-// The PanelMenu module that contains Button
-const PanelMenu = imports.ui.panelMenu;
-// The St class that contains lots of UI functions
-const St = imports.gi.St;
-// Clutter, which is used for displaying everything
-const Clutter = imports.gi.Clutter;
-```
-
-As soon as this file is ready, you can restart your Shell (press
-Alt-F2 and enter the command `r`), and load the extension with
-e.g. the GNOME Tweak Tool. You will see the Planets button on the
-right. This little label showing the static text “Planets”, however,
-is pretty boring, so let’s add some action.
-
-## Adding some periodical change
-
-Since the planets’ position continuously change, we should update our
-widget every minute or so. Let’s patch our `_init()` a bit:
-
-```javascript
-this.last_update = 0;
-
-MainLoop.timeout_add(1, Lang.bind(this, function() {
-    this.last_update++;
-    this.panelLabel.set_text("Update_count: " + this.last_update);
-}))
-```
-
-This, of course, needs a new import line for `MainLoop` to become available:
-
-```javascript
-const MainLoop = imports.mainloop;
-const Lang = imports.lang;
-```
-
-Now if you restart your Shell, your brand new extension will increase
-its counter every second. This, however, presents some problems.
-
-SWE GLib queries can sometimes be expensive, both in CPU and disk
-operations, so updating our widget every second may present problems.
-Also, planets don’t go **that** fast. We may update our timeout value
-from `1` to `60` or something, but why don’t just give our user a chance
-to set it?
-
-## Introducing settings
-
-Getting settings from `GSettings` is barely straightforward, especially
-for software installed in a non-GNOME directory (which includes
-extensions). To make our lives easier, I copied over a
-[convenience library](https://github.com/projecthamster/shell-extension/blob/master/convenience.js)
-from the [Hamster project](https://projecthamster.wordpress.com/)’s
-extension, originally written by Giovanni Campagna. The relevant
-function here is `getSettings()`:
-
-```javascript
-/**
- * getSettings:
- * @schema: (optional): the GSettings schema id
- *
- * Builds and return a GSettings schema for @schema, using schema files
- * in extensionsdir/schemas. If @schema is not provided, it is taken from
- * metadata['settings-schema'].
- */
-function getSettings(schema) {
-    let extension = ExtensionUtils.getCurrentExtension();
-
-    schema = schema || extension.metadata['settings-schema'];
-
-    const GioSSS = Gio.SettingsSchemaSource;
-
-    // check if this extension was built with "make zip-file", and thus
-    // has the schema files in a subfolder
-    // otherwise assume that extension has been installed in the
-    // same prefix as gnome-shell (and therefore schemas are available
-    // in the standard folders)
-    let schemaDir = extension.dir.get_child('schemas');
-    let schemaSource;
-    if (schemaDir.query_exists(null))
-        schemaSource = GioSSS.new_from_directory(schemaDir.get_path(),
-                                                 GioSSS.get_default(),
-                                                 false);
-    else
-        schemaSource = GioSSS.get_default();
-
-    let schemaObj = schemaSource.lookup(schema, true);
-    if (!schemaObj)
-        throw new Error('Schema ' + schema + ' could not be found for extension '
-                        + extension.metadata.uuid + '. Please check your installation.');
-
-    return new Gio.Settings({ settings_schema: schemaObj });
-}
-```
-
-You can either incorporate this function into your `extension.js` file,
-or just use `convenience.js` file like I (and the Hamster applet) did
-and import it:
-
-```javascript
-const ExtensionUtils = imports.misc.extensionUtils;
-const Me = ExtensionUtils.getCurrentExtension;
-const Convenience = Me.imports.convenience;
-```
-
-Now let’s create the settings definition. GSettings schema files are XML
-files. We want to add only one settings for now, the refresh interval.
-
-```xml
-<?xml version="1.0" encoding="utf-8"?>
-<schemalist>
-    <schema id="org.gnome.shell.extensions.planets" path="/org/gnome/shell/extensions/planets/">
-        <key name="refresh-interval" type="i">
-            <default>30</default>
-            <summary>Refresh interval of planet data</summary>
-            <description>Interval in seconds. Sets how often the planet positions are recalculated. Setting this too low (e.g. below 30) may raise performance issues.</description>
-        </key>
-    </schema>
-</schemalist>
-```
-you need to compile these settings with
-
-    glib-compile-schemas --strict schemas/
-
-Now let’s utilize this new setting. In the extension’s `_init()`
-function, add the following line:
-
-```javascript
-this._settings = Convenience.getSettings();
-```
-
-And, for `getSettings()` to work correctly, we also need to extend our
-`metadata.json` file:
-
-```json
-    "settings-schema": "planets"
-```
-
-After another restart (please, GNOME guys, add an option to reload
-extensions!), your brand new widget will refresh every 30 seconds.
-
-## Displaying the planet positions
-
-## The settings panel
-
-## Start an application

_drafts/writing-an-sdk.md

@@ -1,67 +0,0 @@
----
-layout:    post
-title:     "Lessens you learn while writing an SDK"
-date:      2016-03-19 12:34:56
-tags:      [development]
-published: false
-author:
-    name: Gergely Polonkai
-    email: gergely@polonkai.eu
----
-
-In the last few months I’ve been working on a GLib based SDK for
-client applications that want to communicate with a Matrix.org
-homeserver.
-
-For whoever doesn’t know it, Matrix is a decentralized network of
-servers (Homeservers). Clients can connect to them via HTTP and send
-messages (events, in Matrix terminology) to each other. They are
-called events because these messages can be pretty much anything from
-instant messages through automated notifications to files or, well,
-actual events (such as a vCalendar); anything that you can serialize
-to JSON can go through this network.
-
-My original intention was to integrate Matrix based chat into
-Telepathy, a DBus based messaging framework used by e.g. the GNOME
-desktop (more specifically Empathy, GNOME's chat client.) After
-announcing my plans among the Matrix devs, I quickly learned some
-things:
-
-1. they are more than open to any development ideas
-1. they really wanted to see this working
-1. they would have been happy if there were a GLib or Qt based SDK
-
-With my (far from complete) knowledge in GLib I decided to move on
-with this last point, hoping that it will help me much when I finally
-implement the Telepathy plugin.
-
-## Matrix devs are open minded
-
-What I learned very quickly is that Matrix devs are very open minded
-folks from different parts of the world. They are all individuals with
-their own ideas, experiences and quirks, yet, when it comes to that,
-they steer towards their goals as a community. Thus, getting
-additional information from them while reading the spec was super
-easy.
-
-## The specification is easy to understand
-
-Except when it is not. For these cases, see the previous point.
-
-Jokes asidu, anyone who worked with communications protocols or JSON
-APIs before can get along with it fast. The endpoints are all
-documented, and if something is unclear, they are happy to help
-(especially if you patch up the spec afterwards.)
-
-
-## Copying the SDK for a different language is not (always) what you want
-
-I started my SDK in C, trying to mimic the Python SDK. This was a
-double fail: the Python SDK was a volatile WiP, and C and Python are
-fundamentally different.
-
-During the upcoming weeks this became clear and I switched to the Vala
-language. It is much easier to write GObject based stuff in Vala,
-although I had to fall back to C to get some features working. I also
-planned and implemented a more object oriented API, which is easier to
-use in the GObject world.

_includes/about.html

@@ -1,27 +0,0 @@
-<p>
-    Gergely Polonkai is a systems engineer of a telco company, and
-    also a freelancer self- and software developer.
-</p>
-
-<p>
-    He is learning about different IT subjects since the late
-    1990s. These include web development, application building,
-    systems engineering, IT security and many others. He also dug his
-    nose deeply into free software, dealing with different types of
-    Linux and its applications,
-    while also writing and contributing to some open source projects.
-</p>
-
-<p>
-    On this site he is writing posts about different stuff he faces
-    during work (oh my, yet another IT solutions blog), hoping they
-    can help others with their job, or just to get along with their
-    brand new netbook that shipped with Linux.
-</p>
-
-<p>
-    “I believe one can only achieve success if they follow their own
-    instincts and listen to, but not bend under others’ opinions. If
-    you change your course just because someone says so, you are
-    following their instincts, not yours.”
-</p>

_includes/blog-post.html

@@ -1,49 +0,0 @@
-<article class="{% if page.post_listing %}col-sm-5 col-md-6 {% endif%}post">
-{% if page.post_listing %}
-    <ul class="list-inline">
-        <li class="col-md-8">
-{% endif %}
-            <header class="post-header">
-{% if page.tag %}
-                <h5>
-{% else %}
-                <h3>
-{% endif %}
-{% if page.post_listing %}
-<a href="{{post.url | prepend: site.baseurl}}">
-{% endif %}
-{{post.title}}
-{% if page.post_listing %}
-</a>
-{% endif %}
-{% if layout.render_post %}
-                    <div class="plusone-container"><div class="g-plusone" data-annotation="inline" data-size="small" data-width="300"></div></div>
-{% endif %}
-{% if page.tag %}
-                </h5>
-{% else %}
-                </h3>
-{% endif %}
-                <div class="meta pull-left">
-                    {{post.author.name}}
-                </div>
-                <div class="meta pull-right">
-                    {{post.date | date: "%b %-d, %Y :: %H:%M"}}
-                </div>
-                <div class="clearfix"></div>
-            </header>
-
-            <main>
-{% if layout.render_post %}
-                {{content}}
-{% else %}
-                {{post.excerpt}}
-{% endif %}
-            </main>
-
-{% include tag-link.html %}
-{% if layout.post_listing %}
-        </li>
-    </ul>
-{% endif %}
-</article>

_includes/disqus.html

@@ -1,14 +0,0 @@
-<div id="disqus_thread"></div>
-<script type="text/javascript">
-    var disqus_shortname = 'gergelypolonkai';
-
-    (function() {
-        var dsq = document.createElement('script');
-        dsq.type = 'text/javascript';
-        dsq.async = true;
-        dsq.src = '//' + disqus_shortname + '.disqus.com/embed.js';
-        (document.getElementsByTagName('head')[0] || document.getElementsByTagName('body')[0]).appendChild(dsq);
-    })();
-</script>
-<noscript>Please enable JavaScript to view the <a href="http://disqus.com/?ref_noscript">comments powered by Disqus.</a></noscript>
-<a href="http://disqus.com" class="dsq-brlink">comments powered by <span class="logo-disqus">Disqus</span></a>

_includes/head.html

@@ -1,16 +0,0 @@
-        <meta charset="UTF-8">
-        <meta name="viewport" content="width=device-width, initial-scale=1.0">
-        <meta name="keywords" content="{{page.keywords}}">
-        <meta name="description" content="Personal page of Gergely Polonkai">
-        <title>Gergely Polonkai{% if page.title %}: {{page.title}}{% endif %}</title>
-
-        <link rel="icon" type="image/x-icon" href="{{'/favicon.ico' | prepend: site.baseurl}}">
-        <link href="https://fonts.googleapis.com/css?family=Open+Sans:400,300,300italic,400italic,600,600italic,700,700italic,800,800italic" rel="stylesheet" type="text/css">
-        <link rel="alternate" type="application/rss+xml" title="Gergely Polonkai's Blog - RSS Feed" href="{{site.url}}/blog/atom.xml">
-        <link rel="stylesheet" type="text/css" href="https://maxcdn.bootstrapcdn.com/bootstrap/3.3.4/css/bootstrap.min.css">
-        <link rel="stylesheet" href="{{'/css/style.css' | prepend: site.baseurl}}">
-        <link href="https://cdnjs.cloudflare.com/ajax/libs/jquery.terminal/1.6.3/css/jquery.terminal.min.css" rel="stylesheet"/>
-
-        <script type="text/javascript" src="//code.jquery.com/jquery-2.1.3.min.js"></script>
-        <script src="https://maxcdn.bootstrapcdn.com/bootstrap/3.3.4/js/bootstrap.min.js"></script>
-        <script src="https://cdnjs.cloudflare.com/ajax/libs/jquery.terminal/1.6.3/js/jquery.terminal.min.js"></script>

_includes/header.html

@@ -1,35 +0,0 @@
-            <div class="navbar navbar-inverse navbar-fixed-top">
-                <div class="container-fluid">
-                    <div class="navbar-header">
-                        <button type="button" class="navbar-toggle" data-toggle="collapse" data-target="#gp-navbar">
-                            <span class="icon-bar"></span>
-                            <span class="icon-bar"></span>
-                            <span class="icon-bar"></span>
-                        </button>
-                        <a class="navbar-brand" href="{{'/' | prepend: site.baseurl}}"><img src="{{'/images/profile.svg' | prepend: site.baseurl}}" alt="Gergely Polonkai" style="background-color: white; height: 45px; margin-top: -13px;"></a>
-{% if page.url != '/' %}
-                        <a class="navbar-brand" href="{{'/' | prepend: site.baseurl}}">Gergely Polonkai</a>
-{% endif %}
-                    </div>
-                    <div class="collapse navbar-collapse" id="gp-navbar">
-                        <ul class="nav navbar-nav">
-                            <li><a href="{{'/blog' | prepend: site.baseurl}}">Blog</a></li>
-                            <li><a href="{{'/resume' | prepend: site.baseurl}}">Resume</a></li>
-                            <li><a href="{{'/stories' | prepend: site.baseurl}}">Stories</a></li>
-                        </ul>
-                        <ul class="nav navbar-nav navbar-right">
-                            <li><a href="https://about.me/gergely.polonkai">about.me</a></li>
-                            <li><a href="{{'/disclaimer' | prepend: site.baseurl}}">Disclaimer</a></li>
-                            <li class="dropdown">
-                                <a href="#" class="dropdown-toggle" data-toggle="dropdown" role="button" aria-expanded="false"><span class="glyphicon glyphicon-pencil"></span> Contact me <span class="caret"></span></a>
-                                <ul class="dropdown-menu" role="menu">
-{% for contact in site.data.contacts %}
-                                    <li><a href="{{contact.link}}" target="_blank"><img src="{{'/images/contact/' | prepend: site.baseurl | append: contact.image}}" alt="" /> {{contact.text}}</a></li>
-{% endfor %}
-                                    <li><a href="{{'/blog/atom.xml' | prepend: site.baseurl}}"><img src="{{'/images/contact/feed.png' | prepend: site.baseurl}}" alt="" /> RSS Feed</a></li>
-                                </ul>
-                            </li>
-                        </ul>
-                    </div>
-                </div>
-            </div>

_includes/pagination.html

@@ -1,17 +0,0 @@
-<nav>
-    <ul class="pagination">
-        <li{% if paginator.previous_page == null %} class="disabled"{% endif %}>
-            <a href="{{paginator.previous_page_path | prepend: site.baseurl | replace: '//', '/'}}" aria-label="Previous page">
-                <span aria-hidden="true">&laquo;</span>
-            </a>
-        </li>
-{% for page in (1..paginator.total_pages) %}
-        <li{% if paginator.page == page %} class="active"{% endif %}><a href="{% if page == 1 %}{{'/blog' | prepend: site.baseurl}}{% else %}{{site.paginate_path | prepend: site.baseurl | replace: '//', '/' | replace: ':num', page}}{% endif %}">{{page}}</a></li>
-{% endfor %}
-        <li{% if paginator.next_page == null %} class="disabled"{% endif %}>
-            <a href="{{paginator.next_page_path | prepend: site.baseurl | replace: '//', '/'}}" aria-label="Next page">
-                <span aria-hidden="true">&raquo;</span>
-            </a>
-        </li>
-    </ul>
-</nav>

_includes/post-list.html

@@ -1,9 +0,0 @@
-<div class="container-fluid">
-{% for post in posts limit: post_limit %}
-{%     capture counter %}{% cycle 'odd', 'even' %}{% endcapture %}
-{%     include blog-post.html %}
-{%     if counter == 'even' %}
-    <div class="clearfix"></div>
-{%     endif %}
-{% endfor %}
-</div>

_includes/tag-link.html

@@ -1,11 +0,0 @@
-{% capture tagsize %}{{post.tags | size}}{% endcapture %}
-{% if tagsize != '0' %}
-            <footer>
-                <p class="article-tags">
-{%     for tag in post.tags %}
-                    <a href="{{tag | prepend: '/blog/tag/' | prepend: site.baseurl}}" class="tag-label">{{tag}}</a>
-{%     endfor %}
-                </p>
-                <br class="clearfix">
-            </footer>
-{% endif %}

_layouts/default.html

@@ -1,130 +0,0 @@
-<!DOCTYPE html>
-<html>
-    <head>
-{% include head.html %}
-    </head>
-    <body>
-{% include header.html %}
-        <div class="container" id="main-container">
-
-            {{content}}
-
-{% if page.name != 'about.html' %}
-            <div class="well well-sm small">
-                <div class="pull-left" id="about-well-image">
-                    <a href="{{ site_url }}/about/">
-                        <img src="{{'/images/profile.svg' | prepend: site.baseurl}}" alt="">
-                    </a>
-                </div>
-{%     include about.html %}
-                <div class="clearfix"></div>
-            </div>
-{% endif %}
-        </div>
-        <script type="text/javascript">
-         $(document).ready(function() {
-             $('#tagcloud-button').click(function() {
-                 $('#tag-cloud').toggle('slow');
-             });
-         });
-
-         (function() {
-             var po = document.createElement('script');
-             po.type = 'text/javascript';
-             po.async = true;
-             po.src = 'https://apis.google.com/js/client:plusone.js?onload=start';
-
-             var s = document.getElementsByTagName('script')[0];
-             s.parentNode.insertBefore(po, s);
-         })();
-
-         (function(i,s,o,g,r,a,m){i['GoogleAnalyticsObject']=r;i[r]=i[r]||function(){(i[r].q=i[r].q||[]).push(arguments)},i[r].l=1*new Date();a=s.createElement(o), m=s.getElementsByTagName(o)[0];a.async=1;a.src=g;m.parentNode.insertBefore(a,m)})(window,document,'script','//www.google-analytics.com/analytics.js','ga');
-
-         ga('create', 'UA-43569023-1', 'polonkai.eu');
-         ga('send', 'pageview');
-
-         jQuery.extend_if_has = function(desc, source, array) {
-             for (var i=array.length;i--;) {
-                 if (typeof source[array[i]] != 'undefined') {
-                     desc[array[i]] = source[array[i]];
-                 }
-             }
-             return desc;
-         };
-
-         (function($) {
-             $.fn.tilda = function(eval, options) {
-                 if ($('body').data('tilda')) {
-                     return $('body').data('tilda').terminal;
-                 }
-                 this.addClass('tilda');
-                 options = options || {};
-                 eval = eval || function(command, term) {
-                     term.echo("you don't set eval for tilda");
-                 };
-                 var settings = {
-                     prompt: 'guest@gergely.polonkai.eu> ',
-                     name: 'tilda',
-                     height: 400,
-                     enabled: false,
-                     greetings: 'Welcome to my Terminal.  Type `help\' to list the available commands.\n\nPowered by http://terminal.jcubic.pl',
-                     keypress: function(e) {
-                         if (e.which == 96) {
-                             return false;
-                         }
-                     }
-                 };
-                 if (options) {
-                     $.extend(settings, options);
-                 }
-                 this.append('<div class="td"></div>');
-                 var self = this;
-                 self.terminal = this.find('.td').terminal(eval, settings);
-                 var focus = false;
-                 $(document.documentElement).keypress(function(e) {
-                     console.log(e);
-                     if (e.which == 96) {
-                         self.slideToggle('fast');
-                         self.terminal.focus(focus = !focus);
-                         self.terminal.attr({
-                             scrollTop: self.terminal.attr("scrollHeight")
-                         });
-                     }
-                 });
-                 $('body').data('tilda', this);
-                 this.hide();
-                 return self;
-             };
-         })(jQuery);
-
-         String.prototype.strip = function(char) {
-             return this.replace(new RegExp("^\\s*"), '')
-                        .replace(new RegExp("\\s*$"), '');
-         }
-
-         jQuery(document).ready(function($) {
-             $('#tilda').tilda(function(command, terminal) {
-                 command = command.strip();
-
-                 switch (command) {
-                     case 'help':
-                         terminal.echo('about  - Go to the about page');
-                         terminal.echo(' ');
-                         terminal.echo('More commands will follow soon!');
-
-                         break;
-                     case 'about':
-                         location = '{{ site_url }}/about/';
-
-                         break;
-                     default:
-                         terminal.echo(command + ': command not found');
-
-                         break;
-                 }
-             });
-         });
-        </script>
-        <div id="tilda"></div>
-    </body>
-</html>

_layouts/page.html

@@ -1,15 +0,0 @@
----
-layout: default
----
-<div class="post">
-
-    <header class="post-header">
-        <h2>{{page.title}}</h2>
-        <div class="clearfix"></div>
-  </header>
-
-  <article class="post-content">
-  {{content}}
-  </article>
-
-</div>

_layouts/post.html

@@ -1,20 +0,0 @@
----
-layout: default
-render_post: true
----
-{% assign post = page %}
-{% include blog-post.html %}
-<div class="g-plus" data-action="share" data-height="15"></div>
-
-<nav>
-    <ul class="pager">
-{% if page.previous %}
-        <li class="previous"><a href="{{page.previous.url | prepend: site.baseurl}}">&larr; {{page.previous.title}}</a></li>
-{% endif %}
-{% if page.next %}
-        <li class="next"><a href="{{page.next.url | prepend: site.baseurl}}">{{page.next.title}} &rarr;</a></li>
-{% endif %}
-    </ul>
-</nav>
-
-{% include disqus.html %}

_layouts/posts-by-tag.html

@@ -1,15 +0,0 @@
----
-layout: default
-post_listing: true
----
-<h3 class="tag">{{ page.tag }}</h3>
-{{content}}
-
-<h4>Articles under this tag</h4>
-
-{% if site.tags[page.tag] %}
-{%     assign posts = site.tags[page.tag] %}
-{%     include post-list.html %}
-{% else %}
-No posts with this tag.
-{% endif %}

_make_tags.sh

@@ -1,43 +0,0 @@
-#! /bin/sh
-#
-# Find all tags in all posts under _posts, and generate a file for
-# each under blog/tag. Also, if a tag page does not contain the tag:
-# or layout: keywords, the script will include them in the front
-# matter.
-
-layout="posts-by-tag"
-
-for tag in `grep -h ^tags: _posts/* | sed -re 's/^tags: +\[//' -e 's/\]$//' -e 's/, /\n/g' | sort | uniq`
-do
-    tag_file="blog/tag/${tag}.md"
-    echo -n "[$tag] "
-
-    if [ ! -f $tag_file ]
-    then
-        echo "creating ($tag_file)"
-
-        cat <<EOF > $tag_file
----
-layout: $layout
-tag:    $tag
----
-EOF
-    else
-        updated=0
-        if ! egrep "^tag: +${tag}$" $tag_file 2>&1 > /dev/null; then
-            echo "adding tag"
-            sed -i "0,/---/! s/---/tag:    $tag\\n---/" $tag_file
-            updated=1
-        fi
-
-        if ! egrep "^layout: +" $tag_file 2>&1 > /dev/null; then
-            echo "adding layout"
-            sed -i "0,/---/! s/---/layout: $layout\\n---/" $tag_file
-            updated=1
-        fi
-
-        if [ $updated = 0 ]; then
-            echo ""
-        fi
-    fi
-done

_posts/2011-05-12-ethical-hacking-2011.markdown

@@ -1,29 +0,0 @@
----
-layout:    post
-title:     "Ethical Hacking 2012"
-date:      2011-05-12 20:54:42
-tags:      [conference]
-permalink: /blog/2011/5/12/ethical-hacking-2011
-published: true
-author:
-    name: Gergely Polonkai
-    email: gergely@polonkai.eu
----
-
-Today I went to the Ethical Hacking conference with my boss. It was my first
-appearance at such conferences, but I hope there will be more. Although we
-just started to redesign our IT security infrastructure with a 90% clear goal,
-it was nice to hear that everything is vulnerable. I was thinking if we should
-sell all our IT equipments, fire all our colleagues (you know, to prevent
-social engineering), and move to the South Americas to herd llamas or sheep,
-so the only danger would be some lurking pumas or jaguars. Or I simply leave
-my old background image on my desktop, from the well-known game, which says:
-Trust is a weakness.
-
-Anyways, the conference was really nice. We heard about the weaknesses of
-Android, Oracle, and even FireWire. They showed some demos about everything,
-exploited some free and commercial software with no problem at all. We have
-seen how much power the virtualisation admin has (although I think it can be
-prevented, but I’m not sure yet). However, in the end, we could see that the
-Cloud is secure (or at least it can be, in a few months or so), so I’m not
-totally pessimistic. See you next time at Hacktivity!

_posts/2011-05-12-gentoo-hardened-desktop-with-gnome-3-round-one.markdown

@@ -1,88 +0,0 @@
----
-layout:    post
-title:     "Gentoo hardened desktop with GNOME 3 – Round one"
-date:      2011-05-12 20:32:41
-tags:      [gentoo, gnome3, selinux]
-permalink: /blog/2011/5/12/gentoo-hardened-desktop-with-gnome-3-round-one
-published: true
-author:
-    name: Gergely Polonkai
-    email: gergely@polonkai.eu
----
-
-After having some hard times with Ubuntu (upgrading from 10.10 to 11.04), I
-decided to switch back to my old friend, Gentoo. As I’m currently learning
-about Linux hardening, I decided to use the new SELinux profile, which
-supports the v2 reference policy.
-
-Installation was pretty easy, using the [Gentoo x86
-Handbook](http://www.gentoo.org/doc/hu/handbook/handbook-x86.xml). This profile
-automatically turns on the `USE=selinux` flag (so does the old SELinux
-profile), but deprecated `FEATURE=loadpolicy` (which is turned on by the
-profile, so portage will complain about it until you disable it in
-`/etc/make.conf`).
-
-For the kernel, I chose `hardened-sources-2.6.37-r7`. This seems to be recent
-enough for my security testing needs. I turned on both SELinux, PaX and
-grsecurity. So far, I have no problem with it, but I don’t have X installed
-yet, which will screw up things for sure.
-
-After having those hard times with Ubuntu mentioned before, I decided not to
-install Grub2 yet, as it renders things unusable (eg. my Windows 7
-installation, which I sometimes need at the office). So I installed Grub 0.97
-(this is the only version marked as stable, as I remember), touched
-`/.autorelabel`, and reboot.
-
-My first mistake was using an UUID as the root device on the kernel parameter
-list (I don’t want to list all the small mistakes like forgetting to include to
-correct SATA driver from my kernel and such). Maybe I was lame, but after
-including `/dev/sda5` instead of the UUID thing, it worked like…
-
-Well, charm would not be the good word. For example, I forgot to install the
-lvm2 package, so nothing was mounted except my root partition. After I
-installed it with the install CD, I assumed everything will be all right, but
-I was wrong.
-
-udev and LVM is a critical point in a hardened environment. udev itself
-doesn’t want to work without the `CONFIG_DEVFS_TEMPFS=y` kernel option, so I
-also had to change that. It seemed that it can be done without the install CD,
-as it compiled the kernel with no problems. However, when it reached the point
-when it compresses the kernel with gzip, it stopped with a `Permission denied`
-message (although it was running with root privileges).
-
-The most beautiful thing in the hardened environment with Mandatory Access
-Control enabled) is that root is not a real power user any more by default.
-You can get this kind of messages many times. There are many tools to debug
-these, I will talk about these later.
-
-So, my gzip needed a fix. After digging a bit on the Internet, I found that
-the guilty thing is text relocation, which can be corrected if gzip is
-compiled with PIC enabled. Thus, I turned on `USE=pic` flag globally, and
-tried to remerge gzip. Of course it failed, as it had to use gzip to unpack
-the gzip sources. So it did when I tried to install the PaX tools and gradm to
-turn these checks off. The install CD came to the rescue again, with which I
-successfully recompiled gzip, and with this new gzip, I compressed my new
-kernel, with which udev started successfully. So far, so good, let’s try to
-reboot!
-
-Damn, LVM is still not working. So I decided to finally consult the Gentoo
-hardened guide. It says that the LVM startup scripts under `/lib/rcscripts/…`
-must be modified, so LVM will put its lock files under `/etc/lvm/lock` instead
-of `/dev/.lvm`. After this step and a reboot, LVM worked fine (finally).
-
-The next thing was the file system labelling. SELinux should automatically
-relabel the entire file system at boot time whenever it finds the
-`/.autorelabel` file. Well, in my case it didn’t happen. After checking the
-[Gentoo Hardening](http://wiki.gentoo.org/wiki/Hardened_Gentoo) docs, I realised that the `rlpkg` program does exactly the same
-(as far as I know, it is designed specifically for Gentoo). So I ran `rlpkg`,
-and was kind of shocked. It says it will relabel ext2, ext3, xfs and JFS
-partitions. Oh great, no ext4 support? Well, consulting the forums and adding
-some extra lines to `/etc/portage/package.keywords` solved the problem (`rlpkg`
-and some dependencies had to have the `~x86` keyword set). Thus, `rlpkg`
-relabelled my file systems (I checked some directories with `ls -lZ`, it seemed
-good for me).
-
-Now it seems that everything is working fine, except the tons of audit
-messages. Tomorrow I will check them with `audit2why` or `audit2allow` to see if
-it is related with my SELinux lameness, or with a bug in the policy included
-with Gentoo.

_posts/2011-05-13-zabbix-performance-tip.markdown

@@ -1,35 +0,0 @@
----
-layout:    post
-title:     "Zabbix performance tip"
-date:      2011-05-13 19:03:31
-tags:      [zabbix, monitoring]
-permalink: /blog/2011/5/13/zabbix-performance-tip
-published: true
-author:
-    name: Gergely Polonkai
-    email: gergely@polonkai.eu
----
-
-Recently I have switched from [MRTG](http://oss.oetiker.ch/mrtg/) + [Cacti](http://www.cacti.net/) + [Nagios](http://www.nagios.org/) + [Gnokii](http://www.gnokii.org/) to [Zabbix](http://www.zabbix.com/), and I
-must say I’m more than satisfied with it. It can do anything the former tools
-did, and much more. First of all, it can do the same monitoring as Nagios did,
-but it does much more fine. It can check several parameters within one
-request, so network traffic is kept down. Also, its web front-end can generate
-any kinds of graphs from the collected data, which took Cacti away. Also, it
-can do SNMP queries (v1-v3), so querying my switches’ port states and traffic
-made easy, taking MRTG out of the picture (I know Cacti can do it either, it
-had historical reasons we had both tools installed). And the best part: it can
-send SMS messages via a GSM modem natively, while Nagios had to use Gnokii.
-The trade-off is, I had to install Zabbix agent on all my monitored machines,
-but I think it worths the price. I even have had to install NRPE to monitor
-some parameters, which can be a pain on Windows hosts, while Zabbix natively
-supports Windows, Linux and Mac OS/X.
-
-So I only had to create a MySQL database (which I already had for NOD32
-central management), and install Zabbix server. Everything went fine, until I
-reached about 1300 monitored parameters. MySQL seemed to be a bit slow on disk
-writes, so my Zabbix “queue” filled up in no time. After reading some forums,
-I decided to switch to PostgreSQL instead. Now it works like charm, even with
-the default Debian settings. However, I will have to add several more
-parameters, and my boss wants as many graphs as you can imagine, so I’m more
-than sure that I will have to fine tune my database later.

_posts/2011-05-18-gentoo-hardened-desktop-with-gnome-3-round-two.markdown

@@ -1,29 +0,0 @@
----
-layout:    post
-title:     "Gentoo hardened desktop with GNOME 3 – Round two"
-date:      2011-05-18 10:28:14
-tags:      [gentoo, gnome3, selinux]
-permalink: /blog/2011/5/18/gentoo-hardened-desktop-with-gnome-3-round-two
-published: true
-author:
-    name: Gergely Polonkai
-    email: gergely@polonkai.eu
----
-
-After several hours of `package.keywords`/`package.use` editing and package
-compiling, I managed to install GNOME 3 on my notebook. Well, I mean, the
-GNOME 3 packages. Unfortunately the fglrx driver didn’t seem to recognise my
-ATI Mobility M56P card, and the open source driver didn’t want to give me GLX
-support. When I finally found some clues on what should I do, I had to use my
-notebook for work, so I installed Fedora 14 on it. Then I realised that GNOME
-3 is already included in Rawhide (Fedora 15), so I quickly downloaded and
-installed that instead. Now I have to keep this machine in a working state for
-a few days, so I will learn SELinux stuff in its native environment.
-
-When I installed Fedora 14, the first AVC message popped up after about ten
-minutes. That was a good thing, as I wanted to see `setroubleshoot` in action.
-However, in Fedora 15, the AVC bubbles didn’t show up even after a day. I
-raised my left eyebrow and said that’s impossible, SELinux must be disabled.
-And it’s not! It’s even in enforcing mode! And it works just fine. I like it,
-and I hope I will be able to get the same results with Gentoo if I can get
-back to testing…

_posts/2011-05-27-citrix-xenserver-vs-debian-5-0-upgrade-to-6-0.markdown

@@ -1,41 +0,0 @@
----
-layout:    post
-title:     "Citrix XenServer 5.5 vs. Debian 5.0 upgrade to 6.0"
-date:      2011-05-27 17:33:41
-tags:      [citrix-xenserver, debian]
-permalink: /blog/2011/5/27/citrix-xenserver-vs-debian-5-0-upgrade-to-6-0
-published: true
-author:
-    name: Gergely Polonkai
-    email: gergely@polonkai.eu
----
-
-Few weeks ago I’ve upgraded two of our Debian based application servers from
-5.0 to 6.0. Everything went fine, as the upgraded packages worked well with
-the 4.2 JBoss instances. For the new kernel we needed a reboot, but as the
-network had to be rebuilt, I postponed this reboot until the network changes.
-With the network, everything went fine again, we successfully migrated our
-mail servers behind a firewall. Also the Xen server (5.5.0, upgrade to 5.6
-still has to wait for a week or so) revolted well with some storage disks
-added. But the application servers remained silent…
-
-After checking the console, I realised that they don’t have an active console.
-And when I tried to manually start them, XenServer refused with a message
-regarding pygrub.
-
-To understand the problem, I had to understand how XenServer boots Debian. It
-reads the grub.conf on the first partition’s root or `/boot` directory, and
-starts the first option, without asking (correct me, if I’m mistaken
-somewhere). However, this pygrub thing can not parse the new, grub2 config.
-This is kinda frustrating.
-
-For the first step, I quickly installed a new Debian 5.0 system from my
-template. Then I attached the disks of the faulty virtual machine, and mounted
-all its partitions. This way I could reach my faulty 6.0 system with a chroot
-shell, from which I could install the `grub-legacy` package instead of grub,
-install the necessary kernel and XenServer tools (which were missing from both
-machines somehow), then halt the rescue system, and start up the original
-instance.
-
-Next week I will do an upgrade on the XenServer to 5.6.1. I hope no such
-problems will occur.

_posts/2011-05-27-oracle-database-incompatible-with-oracle-linux.markdown

@@ -1,25 +0,0 @@
----
-layout:    post
-title:     "Oracle Database “incompatible” with Oracle Linux?"
-date:      2011-05-27 17:53:31
-tags:      [linux, oracle]
-permalink: /blog/2011/5/27/oracle-database-incompatible-with-oracle-linux
-published: true
-author:
-    name: Gergely Polonkai
-    email: gergely@polonkai.eu
----
-
-Today I gave a shot to install [Oracle
-Linux](http://www.oracle.com/us/technologies/linux/overview/index.html). I thought I could easily install
-an Oracle DBA on it. Well, I was naive.
-
-As only the 5.2 version is supported by XenServer 5.5, I downloaded that
-version of Oracle Linux. Installing it was surprisingly fast and easy, it
-asked almost nothing, and booted without any problems.
-
-After this came the DBA, 10.2, which bloated an error message in my face
-saying that this is an unsupported version of Linux. Bah.
-
-Is it only me, or is it really strange that Oracle doesn’t support their own
-distro?

_posts/2011-06-10-proxy-only-non-existing-files-with-mod-proxy-and-mod-rewrite.markdown

@@ -1,22 +0,0 @@
----
-layout:    post
-title:     "Proxy only non-existing files with mod_proxy and mod_rewrite"
-date:      2011-06-10 14:20:43
-tags:      [apache]
-permalink: /blog/2011/6/10/proxy-only-non-existing-files-with-mod-proxy-and-mod-rewrite
-published: true
-author:
-    name: Gergely Polonkai
-    email: gergely@polonkai.eu
----
-
-Today I got an interesting task. I had to upload some pdf documents to a site.
-The domain is ours, but we don’t have access to the application server that is
-hosting the page yet. Until we get it in our hands, I did a trick.
-
-I enabled `mod_rewrite`, `mod_proxy` and `mod_proxy_http`, then added the following
-lines to my apache config:
-
-{% gist 47680bfa44eb29708f20 redirect-non-existing.conf %}
-
-I’m not totally sure it’s actually secure, but it works for now.

_posts/2011-09-18-inverse-of-sort.markdown

@@ -1,30 +0,0 @@
----
-layout:    post
-title:     "Inverse of `sort`"
-date:      2011-09-18 14:57:31
-tags:      [linux, command-line]
-permalink: /blog/2011/9/18/inverse-of-sort
-published: true
-author:
-    name: Gergely Polonkai
-    email: gergely@polonkai.eu
----
-
-I’m using \*NIX systems for about 14 years now, but it can still show me new
-things. Today I had to generate a bunch of random names. I’ve create a small
-perl script which generates permutations of some usual Hungarian first and
-last names, occasionally prefixing it with a ‘Dr.’ title or using double first
-names. For some reasons I forgot to include uniqueness check in the script.
-When I ran it in the command line, I realized the mistake, so I appended
-`| sort | uniq` to the command line. So I had around 200 unique names, but in
-alphabetical order, which was awful for my final goal. Thus, I tried shell
-commands like rand to create a random order, and when many of my tries failed,
-the idea popped in my mind (not being a native English speaker): “I don’t have
-to create «random order», but «shuffle the list». So I started typing `shu`,
-pressed Tab in the Bash shell, and voilà! `shuf` is the winner, it does just
-exactly what I need:
-
-    **NAME**
-      shuf - generate random permutations
-
-Thank you, Linux Core Utils! :)

_posts/2011-12-11-why-you-should-always-test-your-software-with-production-data.markdown

@@ -1,16 +0,0 @@
----
-layout:    post
-title:     "Why you should always test your software with production data"
-date:      2011-12-11 12:14:51
-tags:      [development, testing, ranting]
-permalink: /blog/2011/12/11/why-you-should-always-test-your-software-with-production-data
-published: true
-author:
-    name: Gergely Polonkai
-    email: gergely@polonkai.eu
----
-
-I’m writing a software for my company in PHP, using the Symfony 2 framework.
-I’ve finished all the work, created some sample data, it loaded perfectly. Now
-I put the whole thing into production and tried to upload the production data
-into it. Guess what… it didn’t load.

_posts/2012-03-20-php-5-4-released.markdown

@@ -1,29 +0,0 @@
----
-layout:    post
-title:     "PHP 5.4 released"
-date:      2012-03-20 13:31:12
-tags:      [php]
-permalink: /blog/2012/3/20/php-5-4-released
-published: true
-author:
-    name: Gergely Polonkai
-    email: gergely@polonkai.eu
----
-
-After a long time of waiting, PHP announced 5.4 release on 1 March (also,
-today they announced that they finally migrate to Git, which is sweet from my
-point of view, but it doesn’t really matter).
-
-About a year ago we became very agressive towards a developer who created our
-internal e-learning system. Their database was very insecure, and they didn’t
-really follow industry standards in many ways. Thus, we forced them to move
-from Windows + Apache 2.0 + PHP 5.2 + MySQL 4.0 to Debian Linux 6.0 + Apache
-2.2 + PHP 5.3 + MySQL 5.1. It was fun (well, from our point of view), as their
-coders… well… they are not so good. The code that ran “smoothly” on the
-old system failed at many points on the new one. So they code and code, and
-write more code. And they still didn’t finish. And now 5.4 is here. Okay, I
-know it will take some time to get into the Debian repositories, but it’s
-here. And they removed `register_globals`, which will kill that funny code again
-at so many points that they will soon get to rewrite the whole code to make it
-work. And I just sit here in my so-much-comfortable chair, and laugh. Am I
-evil?

_posts/2012-03-27-fast-world-fast-updates.markdown

@@ -1,34 +0,0 @@
----
-layout:    post
-title:     "Fast world, fast updates"
-date:      2012-03-27 06:18:43
-tags:      [linux]
-permalink: /blog/2012/3/27/fast-world-fast-updates
-published: true
-author:
-    name: Gergely Polonkai
-    email: gergely@polonkai.eu
----
-
-We live in a fast world, that’s for sure. When I first heard about Ubuntu
-Linux and their goals, I was happy: they gave a Debian to everyone, but in
-different clothes. It had fresh software in it, and even they gave support of
-a kind. It was easy to install and use, even if one had no Linux experience
-before. So people liked it. I’ve even installed it on some of my servers
-because of the new package versions that came more often. Thus I got an up to
-date system. However, it had a price. After a while, security updates came
-more and more often, and when I had a new critical update every two or three
-days, I’ve decided to move back to Debian. Fortunately I did this at the time
-of a new release, so I didn’t really loose any features.
-
-After a few years passed, even Debian is heading this very same way. But as I
-see, the cause is not the same. It seems that upstream software is hitting
-these bugs, and even the Debian guys don’t have the time to check for them. At
-the time of a GNOME version bump (yes, GNOME 3 is a really big one for the
-UN\*X-like OSes), when hundreds of packages need to be checked, security bugs
-show off more often. On the other hand however, Debian is releasing a new
-security update every day (I had one on each of the last three days). This, of
-course, is good from one point of view as we get a system that is more secure,
-but most administrators don’t have maintenance windows this often. I can think
-of some alternatives like Fedora, but do I really have to change? Dear fellow
-developers, please code more carefully instead!

_posts/2012-06-14-wordpress-madness.markdown

@@ -1,28 +0,0 @@
----
-layout:    post
-title:     "Wordpress madness"
-date:      2012-06-14 06:40:12
-tags:      [wordpress, ranting]
-permalink: /blog/2012/6/14/wordpress-madness
-published: true
-author:
-    name: Gergely Polonkai
-    email: gergely@polonkai.eu
----
-
-I’m a bit fed up that I had to install [MySQL](http://www.mysql.com/) on my
-server to have [Wordpress](http://wordpress.org/) working, so I’ve Googled a
-bit to find a solution for my pain. I found
-[this](http://codex.wordpress.org/Using_Alternative_Databases). I don’t know when
-this post was written, but I think it’s a bit out of date. I mean come on, PDO
-is the part of PHP for ages now, and they say adding a DBAL to the dependencies
-would be a project as large as (or larger than) WP itself. Well,
-yes, but PHP is already a dependency, isn’t it? Remove it guys, it’s too
-large!
-
-Okay, to be serious… Having a heavily MySQL dependent codebase is a bad
-thing in my opinion, and changing it is no easy task. But once it is done, it
-would be a child’s play to keep it up to date, and to port WP to other
-database backends. And it would be more than enough to call it 4.0, and
-raising version numbers fast is a must nowadays (right, Firefox and Linux
-Kernel guys?)

_posts/2012-06-18-ssh-login-failed-on-red-hat-enterprise-linux-6-2.markdown

@@ -1,28 +0,0 @@
----
-layout:    post
-title:     "SSH login FAILed on Red Had Enterprise Linux 6.2"
-date:      2012-06-18 18:28:45
-tags:      [linux, selinux, ssh, red-hat]
-permalink: /blog/2012/6/18/ssh-login-failed-on-red-hat-enterprise-linux-6-2
-published: true
-author:
-    name: Gergely Polonkai
-    email: gergely@polonkai.eu
----
-
-Now this was a mistake I should not have done…
-
-About a month ago I have moved my AWS EC2 machine from Amazon Linux to RHEL
-6.2. This was good. I have moved all my files and stuff, recreated my own
-user, everything was just fine. Then I copied my
-[gitosis](https://github.com/tv42/gitosis) account (user `git` and its home
-directory). Then I tried to log in. It failed. I was blaming OpenSSH for a week
-or so, changed the config file in several ways, tried to change the permissions
-on `~git/.ssh/*`, but still nothing. Permission were denied, I was unable to
-push any of my development changes. Now after a long time of trying, I
-coincidently `tail -f`-ed `/var/log/audit/audit.log` (wanted to open `auth.log`
-instead) and that was my first good point. It told me that `sshd` was unable to
-read `~git/.ssh/authorized_keys`, which gave me the idea to run `restorecon` on
-`/home/git`. It solved the problem.
-
-All hail SELinux and RBAC!

_posts/2012-06-22-upgrades-requiring-a-reboot-on-linux-at-last.markdown

@@ -1,35 +0,0 @@
----
-layout:    post
-title:     "Upgrades requiring a reboot on Linux? At last!"
-date:      2012-06-22 20:04:51
-tags:      [linux]
-permalink: /blog/2012/6/22/upgrades-requiring-a-reboot-on-linux-at-last
-published: true
-author:
-    name: Gergely Polonkai
-    email: gergely@polonkai.eu
----
-
-I’ve recently received an article on Google+ about Fedora’s new idea: package
-upgrades that require a reboot. The article said that Linux guys have lost
-their primary adoo: “Haha! I don’t have to reboot my system to install system
-upgrades!” My answer was always this: “Well, actually you should…”
-
-I think this can be a great idea if distros implement it well. PackageKit was
-a good first step on this road. That software could easily solve such an
-issue. However, it is sooo easy to do it wrong. The kernel, of course, can not
-be upgraded online (or could it be? I have some theories on this subject,
-wonder if it can be implemented…), but other packages are much different.
-From the users’ point of view the best would be if the packages would be
-upgraded in the background seemlessly. E.g. PackageKit should check if the
-given executable is running. If not, it should upgrade it, while notifying the
-user like “Hey dude, don’t start Anjuta now, I’m upgrading it!”, or simply
-denying to start it. Libraries are a bit different, as PackageKit should check
-if any running executables are using the library. Meanwhile, PK should also
-keep a notification somewhere telling the users that some packages could be
-upgraded, but without stopping this-and-that, it can not be done.
-
-I know these things are easier said than done. But I think (a) users should
-tell such ideas to the developers and (b) developers (mostly large companies,
-like Microsoft or Apple) should listen to them, and at least think of these
-ideas. Some users are not as stupid as they think…

_posts/2012-09-05-some-thoughts-about-that-dead-linux-desktop.markdown

@@ -1,80 +0,0 @@
----
-layout:    post
-title:     "Some thoughts about that dead Linux Desktop"
-date:      2012-09-05 09:01:31
-tags:      [linux]
-permalink: /blog/2012/9/5/some-thoughts-about-that-dead-linux-desktop
-published: true
-author:
-    name: Gergely Polonkai
-    email: gergely@polonkai.eu
----
-
-There were some arguments in the near past on [What Killed the Linux
-Desktop](http://tirania.org/blog/archive/2012/Aug-29.html). After reading many
-replies, like [Linus
-Torvalds’](http://www.zdnet.com/linus-torvalds-on-the-linux-desktops-popularity-problems-7000003641/),
-I have my own thoughts, too.
-
-I know my place in the world, especially in the online community. I’m a Linux
-user for about 15 years and a Linux administrator for 10 years now, beginning
-with WindowMaker and something that I remember as GNOME without a version
-number. I have committed some minor code chunks and translations in some minor
-projects, so I’m not really into it from the “write” side (well, until now,
-since I have began to write this blog, and much more, but don’t give a penny
-for my words until you see it).
-
-I’m using Linux since 2.2 and GNOME since 1.whatever. It’s nice that a program
-compiled years ago still runs on today’s Linux kernel, especially if you see
-old DOS/Windows software failing to start on a new Windows 7 machine. I
-understand Linus’ point that breaking external APIs is bad, and I think it can
-work well on the kernel’s level. But the desktop level is much different. As
-the Linux Desktop has such competitors (like OS/X and Windows’ Aero and Metro),
-they have to give something new to the users almost every year to keep up with
-them. Eye candies are a must (yes, of course my techy fellows, they are
-worthless, but users *need* it), and they can not be created without extending
-APIs. And the old API… well, it fades away fast. I don’t really understand
-however, why they have to totally disappear, like
-[GTK_DIALOG_NO_SEPARATOR](http://developer.gnome.org/gtk/stable/GtkDialog.html#GtkDialogFlags)
-in Gtk3. It could be replaced with a 0 value (e.g: it won’t do anything). This
-way my old Gtk2 program could compile with Gtk3 nicely. Also, there could be a
-small software that goes through your source code and warn you about such
-deprecated (and no-doer but still working) things. Porting applications between
-Gtk (and thus, GNOME) versions became a real pain, which makes less enthusiast
-programmers stop developing for Linux. Since I’m a GNOME guy for years, I can
-tell nothing about Qt and KDE, but for the GNOME guys, this is a bad thing. As
-of alternatives, there is Java. No, wait… it turned out recently that [it has
-several security
-bugs](http://www.theregister.co.uk/2012/08/31/critical_flaw_found_in_patched_java).
-Also it’s not that multiplatform as they say (I can’t find the article on
-that at the moment, but I have proof). Also, the JVMs out there eat up so much
-resources, which makes it a bit hard and expensive to use.
-
-Also, I see another problem: those blasted package managers. RPM, DPKG,
-Portage, whatever. What the hell? Why are there so many? Why do developers
-reinvent the wheel? The nave is too small or there are to few spokes? Come on…
-we live in an open source world! Contribute to the one and only package manager
-(which one is that I don’t actually care)! I’m sure the two (three, many)
-bunches of develoeprs could make a deal. Thus, it could become better and
-“outsider” companies would be happier to distribute their software for Linux
-platforms.
-
-And now that we get to the big companies. I don’t really understand them.
-nVidia and ATI made their own closed source drivers for Linux. Some other
-hardware vendors also write Linux drivers, and as the kernel API doesn’t really
-change, they will work for a long time. But what about desktop
-application vendors? Well, they try to stick to a desktop environment or two,
-and if they change too frequently, they stop developing for Linux, like Skype
-did (OK, maybe Skype has other reasons, but you see my point). But why? The
-main part for Linux programs is the Linux kernel and the basic userland like
-libc/stdlib++. If you write graphical software, it will have to use X-Windows.
-Yes, it’s much different in many ways, mostly because they have a… well… pretty
-ugly design by default. But still, it’s the same on every Linux distributions,
-as it became somewhat an industry standard, as it was already on the market
-back in the old UN\*X days. The protocol itself changed just like the Linux
-kernel: almost no change at all, just some new features.
-
-So what kills the Linux desktop in my opinion is these constant wars inside,
-and the lack of support from the outside. Open Source is good, but until these
-(mostly the first) problems are not resolved, Linux Desktop can do nothing on
-the market. It’s a downward spiral hard to escape.

_posts/2012-09-07-how-to-start-becoming-a-web-developer.markdown

@@ -1,76 +0,0 @@
----
-layout:    post
-title:     "How to start becoming a web developer"
-date:      2012-09-07 18:12:12
-tags:      [development, technology]
-permalink: /blog/2012/9/7/how-to-start-becoming-a-web-developer
-published: true
-author:
-    name: Gergely Polonkai
-    email: gergely@polonkai.eu
----
-
-A friend of mine asked me today how to become a web developer. It took me a
-while, but I made up a checklist. It’s short, but it’s enough for the first
-steps.
-
-####  First of all, learn English
-
-Well, if you read this, maybe this was a bad first point…
-
-####  Choose a language and stick to it!
-
-For the UN\*X/Linux line, there is PHP. It’s free, easy to learn, and has many
-free tools and documentations available. It can be used in a functional or an
-object-oriented way.
-
-C# is another good way to start, but for the Windows line. It’s fully object-
-oriented, and the web is full of tutorials, how-tos and other resources.
-
-####  Learn the basics of the system you are working on
-
-To become a good developer, learn at least the basics of the system you are
-working on. Basic commands can always come in handy. Debugging (yes, you will
-do tons of bugs for sure) can become much easier if you know the huge set of
-tools provided by your OS. You should also try to develop in the chosen
-environment. Chose PHP? Get a Linux desktop! ASP.NET? Get a Windows.
-Everything will be much easier!
-
-####  Learn the basics of the web server you are using
-
-PHP can run on [Apache](http://httpd.apache.org/) (as a module), or any
-CGI-capable webserver, like [lighttpd](http://www.lighttpd.net/) or
-[nginx](http://nginx.org/) (well, it can also run on IIS, but trust me: you
-don’t want that). ASP.NET is designed for IIS, and although some scripts can
-be run under a mono-capable server, it should still stay there.
-
-Whichever you choose, learn the basics! How to start and stop the service,
-basic configuration methods, modules/extensions, and so on. It’s more than sure
-that you will face some issues while developing, so it can never hurt.
-
-####  Keep your versions under control
-
-Version control is critical nowadays. It gives you a basic backup solution,
-can come in handy with debugging, and if you ever want to work in a team, you
-will badly need it.
-
-Subversion is a bit out of date now, and it’s kind of hard to set up.
-
-Git is no easy. You will have to learn a lot of stuff, but basicly it’s just
-another version control system. Just choose if you want to stick to
-merge-then-commit or rebase-then-commit, get a client, and get on the run.
-
-Microsoft’s Team Foundation is another good way if you are working in a team.
-It provides several other features besides version controlling, and is well
-integrated into Visual Studio, which is highly recommended for Windows based
-development.
-
-####  Choose an environment to work in
-
-There are so many good tools out there. You should choose according to the
-language and OS on what you are working on. [Zend
-Studio](http://www.zend.com/en/products/studio) or
-[Netbeans](https://netbeans.org/) are both good tools for PHP development,
-while [Visual Studio](http://www.visualstudio.com/) is a best buy for Windows
-development. Both of these have many ups and downs, but once you get in touch
-with their deeper parts, you will like them.

_posts/2012-09-10-do-not-track-in-ie10-vs-apache.markdown

@@ -1,19 +0,0 @@
----
-layout:    post
-title:     "Do-Not-Track in IE10 vs. Apache"
-date:      2012-09-10 20:22:32
-tags:      [apache, technology]
-permalink: /blog/2012/9/10/do-not-track-in-ie10-vs-apache
-published: true
-author:
-    name: Gergely Polonkai
-    email: gergely@polonkai.eu
----
-
-[Apache developer decided not to accept Do-Not-Track headers from IE10
-users](http://arstechnica.com/security/2012/09/apache-webserver-updated-to-ignore-do-not-track-settings-in-ie-10/),
-because it’s enabled by default. So… if I install a plugin that hides the
-fact from the web server that I’m using IE10, I become eligible of using
-it. But if I do this, I simply became eligible because I consciously installed
-that addon, so I could actually use it without hiding the fact. Sorry if
-I’m a bit Philosoraptorish…

_posts/2012-09-16-symfony-2-create-role-and-class-based-acls-with-your-roles-coming-from-the-orm.markdown

@@ -1,64 +0,0 @@
----
-layout:    post
-title:     "Symfony 2 – Create role- and class-based ACLs with your roles coming from the ORM"
-date:      2012-09-16 18:39:25
-tags:      [php, symfony]
-permalink: /blog/2012/9/16/symfony-2-create-role-and-class-based-acls-with-your-roles-coming-from-the-orm
-published: true
-author:
-    name: Gergely Polonkai
-    email: gergely@polonkai.eu
----
-
-During the last weeks I had some serious issues with one of my private Symfony
-2 projects. One of my goals was to create a dynamic security system, e.g my
-administrators wanted to create roles, and grant these roles access to
-different object types (classes) and/or objects.
-
-So I have created a `User` entity, which implements `UserInterface` and
-`AdvancedUserInterface`, the latter for the possibility to enable/disable
-accounts and such. It had a `$roles` property, which was a `ManyToMany` relation
-to the `Role` entity, which implemented `RoleInterface`. Also I have created my
-own role hierarchy service that implements `RoleHierarchyInterface`.
-
-So far so good, first tests. It soon turned out that if `User::getRoles()`
-returns a `DoctrineCollection` as it does by default, then the standard
-
-{% gist 883ace4f35e440f6fe0f WhatEver.php %}
-
-doesn’t work. I know, it should not be hard coded, as my roles and permission
-tables are dynamic, I have just tested. So I fixed my `User` entity so
-`getRoles()` returns an array of `Role` objects instead of the
-`DoctrineCollection`. Also I implemented a `getRolesCollection()` method to
-return the original collection, but I think it will never be used.
-
-After that, I had to implement some more features so I put this task away.
-Then, I tried to create my first ACL.
-
-{% gist 883ace4f35e440f6fe0f WhatEver2.php %}
-
-I was about to check if the user who is logged in has an `OWNER` permission on
-the `User` class.
-
-{% gist 883ace4f35e440f6fe0f WhatEver3.php %}
-
-The ACL was defined based on a role, so everyone who had the `ROLE_ADMIN` role
-should gain access to the user listing page. But they didn’t. It took several
-weeks to find the cause, I have put it on
-[stackoverflow](http://stackoverflow.com/questions/12057795/symfony-2-1-this-getsecurity-context-isgrantedrole-admin-returns-fa)
-and the Symfony Google Group, but no usable answers.
-
-Then I went off for debugging. Setting up NetBeans for xdebug-based PHP
-debugging was real fun under Fedora, but that’s another story. After a while I
-have found that Symfony’s basic access decision manager checks for
-`$role->getRole()` only if `$role` is an instance of
-`Symfony\Component\Security\Core\Role\Role`, instead of checking if the object
-implements `Symfony\Component\Security\Core\Role\RoleInterface`. So I’ve
-checked if the bug is already reported. It turned out that it was, and my
-solution was available in a specific commit about a year ago, but as [Johannes
-Schmitt commented, it would introduce a security
-issue](https://github.com/symfony/symfony/commit/af70ac8d777873c49347ac828a817a400006cbea),
-so it was reverted. Unfortunately neither Johannes Schmitt, nor Fabien
-Potencier (nor anyone else) could (or wanted) to tell about this issue. So the
-final (and somewhat hack-like) solution was to extend
-`Symfony\Component\Security\Core\Role\Role`. And boom! It worked.

_posts/2012-10-07-smsgateway-and-smssender.markdown

@@ -1,25 +0,0 @@
----
-layout:    post
-title:     "SmsGateway and SmsSender"
-date:      2012-10-07 00:10:26
-tags:      [development, php, symfony]
-permalink: /blog/2012/10/7/smsgateway-and-smssender
-published: true
-author:
-    name: Gergely Polonkai
-    email: gergely@polonkai.eu
----
-
-I have just uploaded my SmsGateway, SmsSender and SmsSenderBundle packages to
-[GitHub](http://github.com/gergelypolonkai) and
-[Packagist](http://packagist.org). I hope some of you will find it useful.
-
-* SmsGateway
-  * [GitHub](https://github.com/gergelypolonkai/smsgateway)
-  * [Packagist](https://packagist.org/packages/gergelypolonkai/smsgateway)
-* SmsSender
-  * [GitHub](https://github.com/gergelypolonkai/smssender)
-  * [Packagist](https://packagist.org/packages/gergelypolonkai/smssender)
-* SmsSenderBundle
-  * [GitHub](https://github.com/gergelypolonkai/smssender-bundle)
-  * [Packagist](https://packagist.org/packages/gergelypolonkai/smssender-bundle)

_posts/2012-10-13-changing-the-session-cookie-s-name-in-symfony-2.markdown

@@ -1,20 +0,0 @@
----
-layout:    post
-title:     "Changing the session cookie’s name in Symfony 2"
-date:      2012-10-13 12:49:28
-tags:      [symfony, development]
-permalink: /blog/2012/10/13/changing-the-session-cookie-s-name-in-symfony-2
-published: true
-author:
-    name: Gergely Polonkai
-    email: gergely@polonkai.eu
----
-
-I have a development server, on which I have several Symfony 2.x projects under
-the same hostname in different directories. Now I’m facing a funny problem
-which is caused by that the cookies Symfony places for each of my projects have
-the same name.
-
-To change this, you will have to modify the `config.yml` file like this:
-
-{% gist c695670ecca2809f7c93 %}

_posts/2012-12-20-symfony-2-configuration-array-of-associative-arrays.markdown

@@ -1,22 +0,0 @@
----
-layout:    post
-title:     "Symfony 2 Configuration – Array of associative arrays"
-date:      2012-12-20 12:03:23
-tags:      [php, symfony]
-permalink: /blog/2012/12/20/symfony-2-configuration-array-of-associative-arrays
-published: true
-author:
-    name: Gergely Polonkai
-    email: gergely@polonkai.eu
----
-
-Few days ago I have struggled with a problem using Symfony2 configuration. I
-wanted to add the following kind of configuration to `config.yml`:
-
-{% gist 30440e25f7a447730064 config.yml %}
-
-The problem was that the stuff under `transitions` is dynamic, so those
-`hc_cba` and `cba_hc` tags can be pretty much anything. After hitting many
-errors, I came to the solution:
-
-{% gist 30440e25f7a447730064 DynarrayConfiguration.php %}

_posts/2013-01-05-development-man-pages-on-fedora.markdown

@@ -1,14 +0,0 @@
----
-layout:    post
-title:     "Development man pages on Fedora"
-date:      2013-01-05 18:20:41
-tags:      [development, fedora]
-permalink: /blog/2013/1/5/development-man-pages-on-fedora
-published: true
-author:
-    name: Gergely Polonkai
-    email: gergely@polonkai.eu
----
-
-If you use Fedora (like me), and can’t find the development manual pages for
-e.g. `printf(3)` (like me), just `yum install man-pages` (like me).

_posts/2013-01-06-registering-an-enum-type-in-glib-s-type-system.markdown

@@ -1,32 +0,0 @@
----
-layout:    post
-title:     "Registering an enum type in GLib’s type system"
-date:      2013-01-06 02:34:03
-tags:      [c, development, glib]
-permalink: /blog/2013/1/6/registering-an-enum-type-in-glib-s-type-system
-published: true
-author:
-    name: Gergely Polonkai
-    email: gergely@polonkai.eu
----
-
-I faced a problem in my [GLib](https://developer.gnome.org/glib/) self-teaching
-project, [wMUD](https://github.com/gergelypolonkai/wmud) today. I wanted to
-register a signal for a `GObject`, whose handler should accept two `enum`
-parameters for which I had to register a new `GEnum` type in the `GObject` type
-system. However, the [documentation on this
-feature](https://developer.gnome.org/gobject/unstable/gtype-non-instantiable.html)
-(thanks for pointing out goes to hashem on `#gnome-hackers`) is not… uhm…
-obvious. Making the long story short, I have checked with the `GIO` sources for
-an example, and using that, I have created this small, working chunk:
-
-{% gist 47794b6fb94484f8160b client-state.h %}
-
-{% gist 47794b6fb94484f8160b client-state.c %}
-
-Still, it can be made more perfect by using the
-[glib-mkenums](http://developer.gnome.org/gobject/stable/glib-mkenums.html)
-tool. I will read through the GLib Makefiles tomorrow for some hints on
-this.
-
-Edit: you can find the glib-mkenums solution [here]({% post_url 2014-08-16-registering-an-enum-type-in-glib-glib-mkenums-magic %}).

_posts/2013-01-14-git-rm-cached-madness.markdown

@@ -1,17 +0,0 @@
----
-layout:    post
-title:     "git rm --cached madness"
-date:      2013-01-14 21:38:00
-tags:      [development, git]
-permalink: /blog/2013/1/14/git-rm-cached-madness
-published: true
-author:
-    name: Gergely Polonkai
-    email: gergely@polonkai.eu
----
-
-I have recently learned about `git rm --cached`. It’s a very good tool, as it
-removes a file from tracking, without removing your local copy of it. However,
-be warned that if you use `git pull` in another working copy, the file will be
-removed from there! If you accidentally put the configuration of a production
-project, and remove it on your dev machine, it can cause a lot of trouble ;)

_posts/2013-01-17-jms-diextrabundle-s-greppatternfinder-grep-exits-with-status-code-2-on-fedora-18.markdown

@@ -1,52 +0,0 @@
----
-layout:    post
-title:     "JMS\\DiExtraBundle’s GrepPatternFinder – grep exits with status code 2 on Fedora 18"
-date:      2013-01-17 00:32:12
-tags:      [fedora, selinux, symfony]
-permalink: /blog/2013/1/17/jms-diextrabundle-s-greppatternfinder-grep-exits-with-status-code-2-on-fedora-18
-published: true
-author:
-    name: Gergely Polonkai
-    email: gergely@polonkai.eu
----
-
-Yesterday I’ve upgraded my development machines from Fedora 17 to Fedora
-18. Although it went well, my [Symfony](http://symfony.com/) projects stopped
-working with a message like this:
-
-    RuntimeException: Command "/usr/bin/grep --fixed-strings --directories=recurse --devices=skip --files-with-matches --with-filename --color=never --include=*.php 'JMS\DiExtraBundle\Annotation'
-    '/var/www/html/gergelypolonkaiweb/app/../src'
-    '/var/www/html/gergelypolonkaiweb/vendor/symfony/symfony/src/Symfony/Bundle/FrameworkBundle'
-    '/var/www/html/gergelypolonkaiweb/vendor/symfony/symfony/src/Symfony/Bundle/SecurityBundle'
-    '/var/www/html/gergelypolonkaiweb/vendor/symfony/symfony/src/Symfony/Bundle/TwigBundle'
-    '/var/www/html/gergelypolonkaiweb/vendor/symfony/monolog-bundle/Symfony/Bundle/MonologBundle'
-    '/var/www/html/gergelypolonkaiweb/vendor/symfony/swiftmailer-bundle/Symfony/Bundle/SwiftmailerBundle'
-    '/var/www/html/gergelypolonkaiweb/vendor/symfony/assetic-bundle/Symfony/Bundle/AsseticBundle'
-    '/var/www/html/gergelypolonkaiweb/vendor/doctrine/doctrine-bundle/Doctrine/Bundle/DoctrineBundle'
-    '/var/www/html/gergelypolonkaiweb/vendor/sensio/framework-extra-bundle/Sensio/Bundle/FrameworkExtraBundle'
-    '/var/www/html/gergelypolonkaiweb/vendor/jms/aop-bundle/JMS/AopBundle'
-    '/var/www/html/gergelypolonkaiweb/vendor/jms/security-extra-bundle/JMS/SecurityExtraBundle'
-    '/var/www/html/gergelypolonkaiweb/vendor/doctrine/doctrine-migrations-bundle/Doctrine/Bundle/MigrationsBundle'
-    '/var/www/html/gergelypolonkaiweb/vendor/friendsofsymfony/jsrouting-bundle/FOS/JsRoutingBundle'
-    '/var/www/html/gergelypolonkaiweb/vendor/avalanche123/imagine-bundle/Avalanche/Bundle/ImagineBundle'
-    '/var/www/html/gergelypolonkaiweb/vendor/genemu/form-bundle/Genemu/Bundle/FormBundle'
-    '/var/www/html/gergelypolonkaiweb/src/GergelyPolonkai/FrontBundle'
-    '/var/www/html/gergelypolonkaiweb/src/GergelyPolonkai/GeshiBundle'
-    '/var/www/html/gergelypolonkaiweb/vendor/symfony/symfony/src/Symfony/Bundle/WebProfilerBundle'
-    '/var/www/html/gergelypolonkaiweb/vendor/sensio/distribution-bundle/Sensio/Bundle/DistributionBundle'
-    '/var/www/html/gergelypolonkaiweb/vendor/sensio/generator-bundle/Sensio/Bundle/GeneratorBundle'" exited with non-successful status code "2".
-
-After getting through my logs and such, I’ve finally found out that the new
-SELinux policy is causing the trouble together with git. Eventually, my
-`.git/logs` directory is tagged as `unconfined_u:object_r:httpd_log_t:s0`.
-`httpd_log_t` type is not readable by the `system_u:system_r:httpd_t:s0` user,
-which makes `/usr/bin/grep` throw an access denied error. To fix this, I needed
-to do
-
-    semanage fcontext -a -t httpd_sys_content_t '/var/www(/.*)?/\.git/logs(/.*)?'
-
-as root. This makes `.git` directories readable for the httpd process, thus,
-for `grep`. The optimal solution would be to tell `GrepPatternFinder` to ignore
-version control stuff, so the `httpd` process would have no access to them at
-all. Also, in production, removing the `.git` or `.svn` directories could be a
-good idea.

_posts/2013-02-19-mount-device-or-resource-busy-after-enabling-multipath.markdown

@@ -1,32 +0,0 @@
----
-layout:    post
-title:     "mount: device or resource busy after enabling multipath"
-date:      2013-02-19 23:09:05
-tags:      [linux, heartbeat-cluster]
-permalink: /blog/2013/2/19/mount-device-or-resource-busy-after-enabling-multipath
-published: true
-author:
-    name: Gergely Polonkai
-    email: gergely@polonkai.eu
----
-
-We have a heartbeat cluster with two nodes. It has been running for several
-months without problems. The shared storage is on an IBM DS3400, on which we
-have a large volume formatted with ext4.
-
-Today I decided to reboot the active node for security reasons. So I’ve
-switched to the passive node, which failed at the first step: it was unable to
-mount the storage (`/dev/sda1`). After whining for a few moments, I tried to
-mount it by hand, which told me
-
-    /dev/sda1 already mounted or /data is busy
-
-I’ve quickly made sure that none of that was true. After checking
-this-and-that, it turned out that the passive node had `multipathd` running, so
-I looked under `/dev/mapper`, and found two symlinks there, `<long-long WWN>`
-and `<long-long WWN>-part1`. As the partition table and the disk size was the
-same as on `/dev/sda`, I tried to
-
-    mount /dev/<long-long WWN>-part1 /data
-
-and voilà! It worked like charm!

_posts/2013-02-27-why-i-stopped-using-annotation-based-routing-in-symfony-today.markdown

@@ -1,27 +0,0 @@
----
-layout:    post
-title:     "Why I stopped using annotation based routing in Symfony today"
-date:      2013-02-27 23:10:24
-tags:      [development, symfony]
-permalink: /blog/2013/2/27/why-i-stopped-using-annotation-based-routing-in-symfony-today
-published: true
-author:
-    name: Gergely Polonkai
-    email: gergely@polonkai.eu
----
-
-I have read several opinions about routing configuration in Symfony. I stayed
-with annotation based routing as it was convinient for me to see the URL right
-above the controller action. This was because by just checking the URL, I
-remembered the controlling code, as they always were fresh ones. Well, until
-today.
-
-I had to take a look into an old (Sf 2.0, last commit was about 3 months ago)
-project of mine. In the same run I’ve upgraded the whole project to 2.2 (it was
-a fast one, thanks for [JMikola@GitHub](https://github.com/jmikola) for the
-quick reply on my issue with
-[JmikolaJsAssetsHelperBundle](https://github.com/jmikola/JmikolaJsAssetsHelperBundle)
-again!). After that I went on to the requested change. Now, finding a route in
-about 40 controller files spread between 3 bundles can really be a pain! So
-I’ve finished with annotation based routing. It’s still a nice feature, it’s
-simply not for me.

_posts/2013-03-01-programming-as-i-see-it.markdown

@@ -1,62 +0,0 @@
----
-layout:    post
-title:     "Programming, as I see it"
-date:      2013-03-01 23:32:35
-permalink: /blog/2013/3/1/programming-as-i-see-it
-published: false
-author:
-    name: Gergely Polonkai
-    email: gergely@polonkai.eu
----
-
-Since my age of around 11, I write code. I began with BASIC, which is,
-well, the most basic language I have ever seen. Simply writing
-<code>10 PRINT "Hello World!"</code> does the job (with Assembly it
-would be tens of lines as I recall). Then I moved to Pascal, then
-Delphi (which is basically the same thing). The next step was a bit
-longer, as I started learning more languages after this, like Perl
-(for dynamic web pages), C (for desktop applications), TCL (for
-eggdrop programming. Yes, I might have been a weird kid), PHP (again,
-for dynamic web pages. It was becoming mainstream back then).
-
-Many of my classmates looked down on me, as they thought I was a geek (hell I
-was, but I wouldn’t have confessed it then), and called me a nerd. For a few
-months maybe I was depressed, but after that I realised that this is the thing
-I want to do in my life, this is the thing I’m good at.
-
-Most people I ask why don’t they code say “it’s too hard”. I’ve attended some
-courses (both online and offline, and I was like “Whoa! Coding is extremely
-hard! What the hell! I will never learn it!”, but right after the course I
-realised that everything is just fine, I can still write programs, and it’s
-eeeeasy. So then, what’s the problem?
-
-After looking through many course papers, I found that most teachers do it
-totally wrong. A programming language is just that: a language. You don’t start
-learning Spanish by going into a classic literature conference in Madrid and
-doing a speech, but learn the basic vocabulary and grammar. The same goes for
-coding. You learn the vocabulary (the basic commands or keywords) and grammar
-(syntax). I had several ideas how this could be taught, just didn’t have the
-background to do it.
-
-The idea of teaching programming lingers in my head for years now, and a few
-days ago, I’ve bumped into [this
-video](https://www.youtube.com/watch?v=dU1xS07N-FA). So it seems that
-technology superstars like Bill Gates and Mark Zuckerberg wants to do the same.
-Maybe they don’t have enough high quality coders at hand. Well of course,
-if teachers make it awfully hard to learn it! So a bunch of guys sat together
-and created [code.org](http://www.code.org/) to achieve my old dream. I like
-the idea. And although I have almost no visitor on this blog of mine, allow me
-to give you a few points on how I see programming.
-
-####  Great learning process
-
-When you write programs, especially during the first years, you adapt a new way
-of thinking and learning. If you learn it as an adult, it can be a bit of a
-pain, but as a child, it’s easy as learning how the wheels of those little cars
-spin).
-
-####  A job
-
-####  Art
-
-####  Magic

_posts/2013-03-05-fedora-can-t-change-active-directory-password-via-kpasswd.markdown

@@ -1,38 +0,0 @@
----
-layout:    post
-title:     "Fedora can’t change Active Directory password via kpasswd"
-date:      2013-03-05 08:55:04
-tags:      [fedora, kerberos, active-directory]
-permalink: /blog/2013/3/5/fedora-can-t-change-active-directory-password-via-kpasswd
-published: true
-author:
-    name: Gergely Polonkai
-    email: gergely@polonkai.eu
----
-
-I wanted to change my AD password today. As the AD is actually a Kerberos
-server, I was pretty sure that `kpasswd` will do the trick. However, `kpasswd`
-output looked like this:
-
-    $ kpasswd
-    Password for polonkai.gergely@EXAMPLE.LOCAL:
-    Enter new password:
-    Enter it again:
-    kpasswd: Cannot find KDC for requested realm changing password
-
-I’ve checked `kinit` and `klist`, everything looked fine. After a while it came
-to my mind that password changing is done through the kadmin server, not
-through the KDC. It seems that when I set up the Active Directory membership,
-the `admin_server` directive is not get written to `krb5.conf`. So all I had to
-do was to put
-
-    admin_server = ad.example.local
-
-in that file, and voilà!
-
-    $ kpasswd
-    Password for polonkai.gergely@EXAMPLE.LOCAL:
-    Enter new password:
-    Enter it again:
-    Password changed.
-

_posts/2013-03-05-haversine-in-mysql.markdown

@@ -1,17 +0,0 @@
----
-layout:    post
-title:     "Haversine in MySQL"
-date:      2013-03-05 12:49:28
-permalink: /blog/2013/3/5/haversine-in-mysql
-tags:      [mysql, development]
-published: true
-author:
-    name: Gergely Polonkai
-    email: gergely@polonkai.eu
----
-
-Just insert it in your database, feed them two Google coordinates, and you get
-the distance in kilometres. If you happen to need it in miles, change the
-constant `12756.200` in the `RETURN` row to `7922.6` instead.
-
-{% gist bdad1cf2d410853bef35 %}

_posts/2013-03-13-dvorak-and-me.markdown

@@ -1,28 +0,0 @@
----
-layout:    post
-title:     "Dvorak and me"
-date:      2013-03-13 21:20:13
-tags:      [linux]
-permalink: /blog/2013/3/13/dvorak-and-me
-published: true
-author:
-    name: Gergely Polonkai
-    email: gergely@polonkai.eu
----
-
-A few months ago I have decided to switch to the Dvorak layout. After using
-QWERTY (well, QWERTZ, to be precise) for almost 17 years, it was a hard
-decision, but now I think it worthed the try. I started with the UK (Dvorak
-with UK punctuation) layout, and in about four weeks, I’ve almost reached my
-original typing speed. Today I have modified the Hungarian xkb definitions file
-to add the Hungarian accended letters like ű to the layout, so I don’t have to
-use dead keys anymore (which apparently turned out to be a problem, as the
-Linux version of Java doesn’t support dead keys at all).
-
-Best thing is, as I never learned proper 10-finger typing, but learned Dvorak
-that way, I can switch between QWERTY and Dvorak more or less painlessly (about
-10 minutes of confusion, so to say).
-
-Conclusion: I don’t know yet if this was actually a good decision, but it
-wasn’t bad, after all. But seeing people’s faces when they try to type on my
-machine totally worths it.

_posts/2013-04-09-renaming-a-symfony-2-bundle.markdown

@@ -1,28 +0,0 @@
----
-layout:    post
-title:     "Renaming a Symfony 2 bundle"
-date:      2013-04-09 22:29:48
-tags:      [development, symfony]
-permalink: /blog/2013/4/9/renaming-a-symfony-2-bundle
-published: true
-author:
-    name: Gergely Polonkai
-    email: gergely@polonkai.eu
----
-
-Today I’ve realised that the name I gave to one of my Symfony 2 bundles should
-be something else. To rename a bundle, one must do four things (at least).
-
-1. Change the namespace from `Vendor\OldBundle` to `Vendor\NewBundle` in every
-   PHP class (sounds like pain? It is…)
-1. Change the name of files and classes. Some files under
-   `src/Vendor/OldBundle` (and the classes in them) contain the name of the
-   bundle, like `OldBundle/DependencyInjection/VendorOldBundleExtension.php`
-   and `OldBundle/VendorOldBundle.php`. You should rename them, or Symfony
-   won’t find the classes defined in them! When done, rename the whole bundle
-   directory either.
-1. Change the configuration files accordingly, including `AppKernel.php`. These
-   config files are usually `routing.yml`, `services.yml`, and in some cases,
-   `config.yml`
-1. Change the references in other parts of your code. A `grep OldBundle .` will
-   usually help…

_posts/2013-05-06-installing-otrs-in-fedora-18-with-selinux-enabled.markdown

@@ -1,111 +0,0 @@
----
-layout:    post
-title:     "Installing OTRS in Fedora 18 with SELinux enabled"
-date:      2013-05-06 06:01:52
-tags:      [fedora, selinux, otrs]
-permalink: /blog/2013/5/6/installing-otrs-in-fedora-18-with-selinux-enabled
-published: true
-author:
-    name: Gergely Polonkai
-    email: gergely@polonkai.eu
----
-
-I’ve read somewhere in an OTRS installation howto that if you want to install
-OTRS, you will have to disable SELinux. Well, I won’t.
-
-During the last few months, I have been using Fedora 18 with SELinux on all of
-my desktop machines and on my notebook, and I had no problems at all.
-Meanwhile I got familiar with SELinux itself, and got used to solving problems
-caused by it. So I started `tail -f /var/log/httpd/error_log` in one terminal
-(to see if something Apache related thing appears),
-`tail -f /var/log/audit/audit.log` in another (to see errors caused by
-SELinux), opened the admin manual at the installation chapter, took a deep
-breath, and went on.
-
-Throughout this article, I will refer to OTRS 3.2.6 as OTRS and Fedora 18
-(with only “stock” repositories) as Fedora. I assume that you have already
-installed OTRS in a non-SELinux environment before, and that you have at least
-some basic knowledge about SELinux, MAC, RBAC, and all the like. I’m
-installing OTRS in `/opt/otrs`, so if you install it somewhere else, you will
-have to modify the paths below. Also, if you happen to install under
-`/var/www` (I wouldn’t recommend it), that directory already has the
-`httpd_sys_content_t` type, so you won’t have to set it explicitly.
-
-As the first step I have unpacked the archive to `/opt/otrs`. This directory
-is the OTRS default, many config files have it hardcoded, and changing it is
-no easy task.
-
-Running `otrs.CheckModules.pl` gave me a list of missing perl modules. Red Hat
-and Fedora makes it easy to install these, as you don’t have to know the RPM
-package name, just the perl module name:
-
-    yum install 'perl(Crypt::SSLeay)' \
-                'perl(DBD::Pg)' \
-                'perl(GD)' \
-                'perl(JSON::XS)' \
-                'perl(GD::Text)' \
-                'perl(GD::Graph)' \
-                'perl(Mail::IMAPClient)' \
-                'perl(Net::DNS)' \
-                'perl(PDF::API2)' \
-                'perl(Text::CSV_XS)' \
-                'perl(YAML::XS)'
-
-I also needed to install `mod_perl`. Although `otrs.CheckModules.pl` didn’t
-mention it, the default settings use syslog as the logging module, so unless
-you change it in `Config.pm`, you will also need to install
-`'perl(Unix::Syslog)'`, either.
-
-The default SELinux policy doesn’t permit any network connection to be
-initiated by Apache httpd. As OTRS needs to connect to its database, you
-need to enable it explicitly. In older distributions, the
-`httpd_can_network_connect` was the SELinux boolean for this, but recent
-installations also have a `httpd_can_network_connect_db` flag. As far as I
-know, this enables all network connections to the well-known database
-servers’ default port, but I will have to check for it. For me, with a
-MySQL listening on its standard port, the
-`setsebool httpd_can_network_connect_db=1` command just did it.
-
-With SELinux enabled, Apache won’t be able to read anything that’s not
-marked with the `httpd_sys_content_t` type, nor write anywhere without the
-`httpd_sys_rw_content_t` type. The trivial, quick and dirty solution is to
-label all the files as `httpd_sys_rw_content_t`, and let everything go.
-However, the goal of SELinux is just the opposite of this: grant access
-only to what is really needed. After many trial-and-error steps, it finally
-turned out that for OTRS to work correctly, you must set
-
-* `httpd_sys_content_t`
-  * on `/opt/otrs/var/httpd/htdocs`
-* `httpd_script_exec_t`
-  * on `/opt/otrs/bin/cgi-bin`
-* `httpd_sys_rw_content_t`
-  * on `/opt/otrs/Kernel`
-  * on `/opt/otrs/var/sessions`
-  * on `/opt/otrs/var/log` (unless you use syslog for logging)
-  * on `/opt/otrs/var/packages` (this is used only when you download an .opm
-    package)
-  * on `/opt/otrs/var/stats`
-  * on `/opt/otrs/var/tmp`
-  * on `/opt/otrs/bin` (I wonder why this is required, though)
-
-To do this, use the following command:
-
-    # semanage fcontext -a -t <context> <directory regex>
-
-Where `<directory regex>` is something like `/opt/otrs/Kernel(/.*)?`. When
-this is done, all you have to do is running `restorecon -vR /opt/otrs` so
-it will relabel everything with the correct types (you can omit -v, I just
-like to see what my software does).
-
-The last thing I faced is that Fedora is more restrictive on reading
-directories other than `/var/www`. It has a `Require all denied` on
-`<Directory />`, and a `Require all granted` on `<Directory /var/www>`, so
-`/opt/otrs/var/httpd/htdocs` will throw a
-`403 Forbidden (client denied by server configuration)` error. To get rid
-of this, I had to modify `scripts/apache2-httpd.include.conf` and add
-`Require all granted` to both the `cgi-bin` and `htdocs` directories.
-
-As I will have to use OTRS in a production environment soon with SELinux
-enabled, it is more than sure that this list will change in the near future.
-As there are no official documentation on this (I haven’t find one yet), I
-have to do it with the trial-and-error way, so be patient!

_posts/2013-09-16-swe-glib-final-release.markdown

@@ -1,30 +0,0 @@
----
-layout:    post
-title:     "SWE-GLib final release"
-date:      2013-09-16 21:37:17
-tags:      [development, astrology]
-permalink: /blog/2013/9/16/swe-glib-final-release
-published: true
-author:
-    name: Gergely Polonkai
-    email: gergely@polonkai.eu
----
-
-Few of you may know that I’m interested in astrology. About two months ago
-I have decided to create an astrologers’ software for the GNOME desktop.
-Since then, I have contacted Jean-André Santoni, who created a software
-called [Astrognome](https://code.google.com/p/astrognome/) some years ago.
-We exchanged some e-mails, and after several weeks of coding, I’m proud to
-present [SWE-GLib](https://github.com/gergelypolonkai/swe-glib) 1.0.1. This
-is “just” a library which wraps around [Swiss
-Ephemeris](http://www.astro.com/swisseph/), creating a nice GLib-ish
-interface around it. See the project page and the built-in GTK-Doc document
-for more information.
-
-The astrologer’s software I’m writing will be
-[Astrognome](https://github.com/gergelypolonkai/astrognome) (thanks for
-Jean-André for letting me use the name). It is currently in pre-alpha
-status, but already utilizes SWE-GLib (it just can’t display the results
-yet). If you happen to be interested in astrology and/or Astrognome, fork
-the repository and contribute! You can also contact me (or open an
-enhancement issue on GitHub) if you have any ideas.

_posts/2013-09-24-from-symfony-to-django-in-two-days.markdown

@@ -1,25 +0,0 @@
----
-layout:    post
-title:     "From Symfony to Django in two days"
-date:      2013-09-24 14:05:22
-tags:      [development, symfony, django]
-permalink: /blog/2013/9/24/from-symfony-to-django-in-two-days
-published: true
-author:
-    name: Gergely Polonkai
-    email: gergely@polonkai.eu
----
-
-I was a Python hater for a long time, although I can’t really tell why. It
-didn’t fit in my mind, maybe. I was programming in BASIC, Pascal (none of
-these would come to my mind, though), C, PHP, Perl, JavaScript, and
-different shell “languages” like awk, sed or bash.
-
-After I could not fit my next Symfony app on my cloud server (it is pretty
-low on storage), I have decided to move slowly to Django. My first task was
-simple: transition my web page (this one) from PHP + Symfony 2 to Python +
-Django. The results: the “static” pages are already working, the blog
-listing is almost ready (some styling issues are still around), only
-tagging remains. And this is after about 6 hours of work. Oh, and the admin
-site is included with Django, so I don’t have to port that. I have also
-decided to finally integrate a comment feature in the Django version.

_posts/2013-11-05-first-impressions-of-windows-8.markdown

@@ -1,29 +0,0 @@
----
-layout:    post
-title:     "First impressions of Windows 8"
-date:      2013-11-05 08:14:50
-tags:      [windows]
-permalink: /blog/2013/11/5/first-impressions-of-windows-8
-published: true
-author:
-    name: Gergely Polonkai
-    email: gergely@polonkai.eu
----
-
-Many of you may know my commitment to Linux and Open Source Software. But this
-doesn’t mean I hate proprietary software like many others do. I think
-everything has its own place in the world, and this goes for software as well.
-
-A few days ago I got my hands on a new notebook, thanks to my company. It was
-shipped with Windows 8 by default, and although I installed Fedora 19 in an
-instant (which went smoothlessly, even with Secure Boot enabled), I’ve decided
-to give a try to this new Windows Version.
-
-Being a heavy Windows 7 user, my first thought was “What the hell is this?”
-But in a day, I got totally used to it. I don’t miss the Start button at all.
-The applications already installed were almost enough for me (I still need
-Office. Maybe I’ll also enroll for Office 365 later…), and the games are great
-and beautiful too. So overall, this new version may be totally different (by
-the looks), but it seems almost the same Windows as we know it. So if you
-don’t freak out by touching something new, go give it a try: don’t
-instant-remove 8 in favour of 7!

_posts/2014-07-18-list-git-branches-and-their-remote-tracking-branches-side-by-side.markdown

@@ -1,33 +0,0 @@
----
-layout:    post
-title:     "List Git branches and their remote tracking branches side by side"
-date:      2014-07-18 21:46:45
-tags:      [git]
-permalink: /blog/2014/7/18/list-git-branches-and-their-remote-tracking-branches-side-by-side
-published: true
-author:
-    name: Gergely Polonkai
-    email: gergely@polonkai.eu
----
-
-I had a hard time following my own branches in a project. They got pretty
-numerous, and I wasn’t sure if I pushed them to origin at all.
-`git branch -a` can list all the branches, including remote ones, but, as
-my list grew too big, it was impossible to follow it any more.
-
-Thus, I have created a small script called git-branches-with-remotes, which
-does the work for me. Its only requirements are git (of course), and the
-`column` command, which is pretty obviously present on every POSIX
-compliant systems (even OSX).
-
-{% gist 8af6a3e86b57dd4c250e %}
-
-I just put it in my path, and `git branches-with-remotes` does the work!
-
-Edit (16 August): I have added some code to mark the current branch (if any)
-with an asterisk. Also, I have put this script [in a
-gist](https://gist.github.com/gergelypolonkai/8af6a3e86b57dd4c250e).
-
-Edit (26 February, 2015): It turns out that `git branch -vv` shows the same
-information and some more: it also shows if the branches are diverged, and the
-first line of the last commit’s message.

_posts/2014-08-16-registering-an-enum-type-in-glib-glib-mkenums-magic.markdown

@@ -1,36 +0,0 @@
----
-layout:    post
-title:     "Registering an enum type in GLib, glib-mkenums magic"
-date:      2014-08-16 15:10:54
-tags:      [development, c, glib]
-permalink: /blog/2014/8/16/registering-an-enum-type-in-glib-glib-mkenums-magic
-published: true
-author:
-    name: Gergely Polonkai
-    email: gergely@polonkai.eu
----
-
-In [this
-post](/blog/2013/1/6/registering-an-enum-type-in-glib-s-type-system) I said
-I will get through the GLib Makefiles to add an enum type to GLib in a more
-sophisticated way.
-
-In my other project,
-[SWE-GLib](https://github.com/gergelypolonkai/swe-glib) I already used this
-method. The following two rules in `Makefile.am` create `gswe-enumtypes.h`
-and `gswe-enumtypes.c`.
-
-{% gist 1e2fdedb136de3ca67f0 Makefile %}
-
-`$(GLIB_MKENUMS)` is set in `configure` with
-`AC_PATH_PROG([GLIB_MKENUMS], [glib-mkenums])`.
-
-This approach requires the GNU Autotools (you can get rid of it by changing
-`$(GLIB_MKENUMS)` to the path to `glib-mkenums` binary), and two template
-files, one for the header and one for the code. `$(gswe_enum_headers)`
-contains a list of all the header files that have enum types defined
-throughout the project.
-
-{% gist 1e2fdedb136de3ca67f0 gswe-enumtypes.h %}
-
-{% gist 1e2fdedb136de3ca67f0 gswe-enumtypes.c %}

_posts/2014-09-17-nyanmacs.markdown

@@ -1,16 +0,0 @@
----
-layout:    post
-title:     "NyanMacs"
-date:      2014-09-17 12:45:42
-tags:      [emacs]
-permalink: /blog/2014/9/17/nyanmacs
-published: true
-author:
-    name: Gergely Polonkai
-    email: gergely@polonkai.eu
----
-
-I was a Vi/ViM user for years. For several reasons I had to change to Emacs
-now and then. And then, I found
-[this](https://www.emacswiki.org/emacs/NyanMode). I surrender. Emacs is
-just better. (And it’s working even in plain text mode without graphics)

_posts/2014-10-07-rounding-numbers-to-n-decimals-in-emacs.markdown

@@ -1,25 +0,0 @@
----
-layout:    post
-title:     "Rounding numbers to N decimals in Emacs"
-date:      2014-10-07 10:28:50
-tags:      [emacs, development]
-permalink: /blog/2014/10/7/rounding-numbers-to-n-decimals-in-emacs
-published: true
-author:
-    name: Gergely Polonkai
-    email: gergely@polonkai.eu
----
-
-I have recently faced a problem, where I had a bunch of SVG files with a
-large amount of fraction numbers in the path definitions. These images were
-displayed in small size, so this amount of precision was irrelevant, and
-these numbers took almost half of my SVG images’ size. So I created an
-Elisp defun to round these numbers to 2 decimals:
-
-{% gist 9c721ceda6d3079b4f05 %}
-
-This finds the first digit of the number under point (the cursor), and
-reduces its digits to the given amount (or the number given with `C-u`). It
-has some drawbacks, though, as it cannot handle exponential forms (e.g.
-`1e-1234`), but these were rare in my case, and its hard to iterate through
-all numbers. I will come over this latter problem soon(ish).

_posts/2015-02-26-using-git-bisect-to-find-the-first-good-commit.markdown

@@ -1,46 +0,0 @@
----
-layout:    post
-title:     "Using Git bisect to find the first good commit"
-date:      2015-02-26 10:42:56
-tags:      [git]
-permalink: /blog/2015/2/26/using-git-bisect-to-find-the-first-good-commit
-published: true
-author:
-    name: Gergely Polonkai
-    email: gergely@polonkai.eu
----
-
-Few months ago we “implemented” a bug in our software, which was released
-to the customers. We continued development for two weeks when the first
-customer ticket arrived about the bug. We successfully reproduced it with
-the customer’s version, but not with the development sources; it turned out
-that one of the developers unconsciously fixed the bug. The devs spent some
-hours finding where the fix lied before coming to me like “There is
-`git-bisect` which we can use to find the commit where we messed up things.
-Is there a way to find where we fixed it?”
-
-For those who don’t know this feature, you have to mark a known “good” and
-“bad” commit, then git-bisect will go through the commits between this two,
-present you the corresponding snapshots, and you have to mark each of them
-as “good” or “bad”. At the end, you will get a commit hash where the bug
-first occured.
-
-As it turned out, our developers’ problem rooted in the naming convention
-of git-bisect: they assumed that the “good” commit must be a working one,
-while a “bad” one must be the buggy. In this case, we did the following:
-
-The commit with the customer’s release tag was marked as good (even though
-this had the bug), and the latest commit on our development branch was
-marked as “bad” (even though the bug was fixed by then). Now with every
-snapshot presented by git-bisect we had to do the opposite what you usually
-do: mark commits still having the bug as “good”, and commits that don’t as
-“bad”. At the end, we had the hash of the commit that fixed the bug (among
-some other things; luckily, the developer who pushed that commit had a
-workflow that introduced a lot of cherry-picking and squashing before the
-push, so he could easily find the bit that actually fixed the problem in
-his local repository with the same technique).
-
-[This StackOverflow answer](http://stackoverflow.com/a/17153598/1305139)
-suggests the very same, but with some aliases:
-
-{% gist a98f4aab84659d60364e %}

_posts/2015-04-25-good-bye-digital-ocean-hello-again-github.markdown

@@ -1,23 +0,0 @@
----
-layout:    post
-title:     "Good bye, Digital Ocean! Hello again, GitHub!"
-date:      2015-04-25 21:18:56
-permalink: /blog/2015/4/25/good-bye-digital-ocean-hello-again-github
-published: true
-author:
-    name: Gergely Polonkai
-    email: gergely@polonkai.eu
----
-
-Few years ago I have signed up for a
-[Digital Ocean](https://www.digitalocean.com/) account. I used one
-single droplet for my private needs, like hosting my private Git
-repositories and my blog. However, as I didn’t host anything else there
-except my blog, I decided to shut it down. From now on, my blog is
-on [GitHub Pages](https://pages.github.com/), as it provides just
-everything I need (except automatically converting my resume to
-PDF. But I can live without that.)
-
-I’m really sorry, Digital Ocean Guys, your hosting is awesome and I’ll
-keep recommending you to others, but paying for a droplet for one
-single blog is overkill.

_posts/2015-04-28-cross-browser-border-radius-sass-mixin-with-varargs.markdown

@@ -1,32 +0,0 @@
----
-layout:    post
-title:     "Cross browser border-radius SASS mixin with varargs"
-date:      2015-04-27 22:59:56
-tags:      [css, sass]
-permalink: /blog/2015/4/28/cross-browser-border-radius-sass-mixin-with-varargs
-published: true
-author:
-    name: Gergely Polonkai
-    email: gergely@polonkai.eu
----
-
-Few days ago I needed to create style sheets with many rounded boxes,
-where different corners had to be rounded differently (think about
-Bootstrap’s [button
-groups](http://getbootstrap.com/components/#btn-groups)).
-
-CSS has this nifty shorthand to specify border width in one line, like
-with `border-width: 1px 2px 3px 4px`, but it lacks the same for
-`border-radius`. So I decided to create something similar using [Sass
-mixins](http://sass-lang.com/guide#topic-6) with dynamic
-parameters. Another nice feature you get using the `border-width`
-shorthand is that you can specify less than four parameters, and the
-values will be applied on different sides of your box, so in the end
-all side will have the whole `border-width` set.
-
-I wanted to achieve the same for my `border-radius` mixin, although I
-could not start specifically with the `top` side. I decided to go with
-the top right corner for the first parameter, while trying to keep a
-sane repeating pattern. Here is the result:
-
-{% gist 313b227434ecc5d85d7b border-radius.sass %}

_posts/2015-06-07-paramconverter-a-la-django.markdown

@@ -1,39 +0,0 @@
----
-layout:    post
-title:     "@ParamConverter à la Django"
-date:      2015-06-07 18:14:32
-tags:      [python, django]
-published: true
-author:
-    name:  Gergely Polonkai
-    email: gergely@polonkai.eu
----
-One thing I really miss from [Django](https://www.djangoproject.com/)
-is [Symfony](http://symfony.com)’s
-[@ParamConverter](http://symfony.com/doc/current/bundles/SensioFrameworkExtraBundle/annotations/converters.html). It
-made my life so much easier while developing with Symfony. In Django,
-of course, there is
-[get_object_or_404](https://docs.djangoproject.com/en/dev/topics/http/shortcuts/#get-object-or-404),
-but, for example, in one of my projects I had a view that had to resolve 6(!)
-objects from the URL, and writing `get_object_or_404` six times is not what a
-programmer likes to do (yes, this view had a refactor later on). A quick Google
-search gave me one [usable
-result](http://openclassrooms.com/forum/sujet/middleware-django-genre-paramconverter-doctrine)
-(in French), but it was very generalized that I cannot always use. Also, it was
-using a middleware, which may introduce performance issues
-sometimes<sup>[citation needed]</sup>. So I decided to go with decorators, and
-at the end, I came up with this:
-
-{% gist 498a32297f39b4960ad7 helper.py %}
-
-Now I can decorate my views, either class or function based, with
-`@convert_params(User, (Article, 'aid'), (Paragraph, None, 'pid'),
-(AnotherObject, None, None, 'obj'))` and all the magic happens in the
-background. The `user_id` parameter passed to my function will be
-popped off, and be resolved against the `User` model by using the `id`
-field; the result is put in the new `user` parameter. For Article, the
-`aid` parameter will be matched against the `id` field of the
-`Article` model putting the result into `article`, and finally, the
-`another_object_id` will be matched against the `id` field of the
-`AnotherObject` model, but in this case, the result is passed to the
-original function as `obj`.

_posts/2015-08-19-foss-failmeter.md

@@ -1,20 +0,0 @@
----
-layout:    post
-title:     "F/OSS Fail meter"
-date:      2015-08-19 10:12:19
-tags:      [development]
-published: true
-author:
-    name: Gergely Polonkai
-    email: gergely@polonkai.eu
----
-
-I have recently bumped into [this
-article](http://spot.livejournal.com/308370.html). Naturally, I quickly
-calculated the FAIL metrics for all my projects (most of them are pretty high).
-To ease calculation, I made up a
-[small page]({{ '/failmeter/' | prepend: site.baseurl }}) based on this list
-(although I have divided the points by 5; I really don’t understand why spot is
-using such big points if all of them can be divided by 5). Feel free to use it,
-and if you have any recommendations (point additions/removal, new categories,
-etc.), leave me a comment!

_posts/2015-08-27-how-my-email-gets-to-that-other-guy.markdown

@@ -1,228 +0,0 @@
----
-layout:    post
-title:    "How my e-mail gets to that other guy?"
-date:      2015-08-27 21:47:19
-tags:      [technology]
-published: true
-author:
-    name: Gergely Polonkai
-    email: gergely@polonkai.eu
----
-
-A friend of mine asked me how it is possible that she pushes buttons on her
-keyboard and mouse, and in an instant her peer reads the text she had in her
-mind. This is a step-by-step introduction of what happens in-between.
-
-#### From your mind to your computer
-
-When you decide to write an e-mail to an acquaintance of yours, you open up
-your mailing software (this document doesn’t cover using mail applications
-you access through your browsers, just plain old Thunderbird, Outlook or
-similar programs. However, it gets the same after the mail left your
-computer), and press the “New Mail” button. What happens during this process
-is not covered in this article, but feel free to ask me in a comment! Now
-that you have your Mail User Agent (MUA) up and running, you begin typing.
-
-When you press a button on your keyboard or mouse, a bunch of bits gets
-through the wire (or through air, if you went wireless) and get into your
-computer. I guess you learned about Morse during school; imagine two
-[Morse operators](http://www.uscupstate.edu/academics/education/aam/lessons/susan_sawyer/morse%20code.jpg),
-one in your keyboard/mouse, and one in your computer. Whenever you press a
-key, that tiny creature sends a series of short and long beeps (called 0 or
-1 bits, respectively) to the operator in your computer (fun fact: have you
-ever seen someone typing at an amazing speed of 5 key presses per second?
-Now imagine that whenever that guy presses a key on their keyboard, that
-tiny little Morse operator pressing his button 16 times for each key press,
-with perfect timing so that the receiving operator can decide if that was a
-short or long beep.)
-
-Now that the code got to the operator inside the machine, it’s up to him to
-decode it. The funny thing about keyboards and computers is that the
-computer doesn’t receive the message “Letter Q was pressed”, but instead
-“The second button on the second row was pressed” (a number called scan
-code). At this time the operator decodes this information (in this example
-it is most likely this Morse code: `···-···· -··-····`) and checks one of
-his tables titled “Current Keyboard Layout.” It says this specific key
-corresponds to letter ‘Q’, so it forwards this information (I mean the
-letter; after this step your computer doesn’t care which plastic slab you
-hit, just the letter ‘Q’) to your MUA, inserts it into the mail in its
-memory, then displaying it happily (more about this step later).
-
-When you finish your letter you press the send button of your MUA. First it
-converts all the pretty letters and pictures to something a computer can
-understand (yes, those Morse codes, or more precisely, zeros and ones,
-again). Then it adds loads of meta data, like your name and e-mail address,
-the current date and time including the time zone and pass it to the sending
-parts of the MUA so the next step can begin.
-
-#### IP addresses, DNS and protocols
-
-The Internet is a huge amount of computers connected with each other, all of
-them having at least one address called IP address that looks something like
-this: `123.234.112.221`. These are four numbers between 0 and 255 inclusive,
-separated by dots. This makes it possible to have 4,294,967,296 computers.
-With the rules of address assignment added, this is actually reduced to
-3,702,258,432; a huge number, still, but it is not enough, as in the era of
-the Internet of Things everything is interconnected, up to and possibly
-including your toaster. Thus, we are slowly transitioning to a new
-addressing scheme that looks like this:
-`1234:5678:90ab:dead:beef:9876:5432:1234`. This gives an enormous amount of
-340,282,366,920,938,463,463,374,607,431,768,211,456 addresses, with only
-4,325,185,976,917,036,918,000,125,705,034,137,602 of them being reserved,
-which gives us only a petty
-335,957,180,944,021,426,545,374,481,726,734,073,854 available.
-
-Imagine a large city with
-[that many buildings](http://www.digitallifeplus.com/wp-content/uploads/2012/07/new-york-city-aerial-5.jpg),
-all of them having only a number: their IP address. No street names, no
-company names, no nothing. But people tend to be bad at memorizing numbers,
-so they started to give these buildings names. For example there is a house
-with the number `216.58.209.165`, but between each other, people call it
-`gmail.com`. Much better, isn’t it? Unfortunately, when computers talk, they
-only understand numbers so we have to provide them just that.
-
-As remembering this huge number of addresses is a bit inconvenient, we
-created Domain Name Service, or DNS for short. A “domain name” usually (but
-not always) consist of two strings of letters, separated by dots (e.g.
-polonkai.eu, gmail.com, my-very-long-domain.co.uk, etc.), and a hostname is
-a domain name occasionally prefixed with something (e.g. **www**.gmail.com,
-**my-server**.my-very-long-domain.co.uk, etc.) One of the main jobs of DNS
-is to keep record of hostname/address pairs. When you enter `gmail.com`
-(which happens to be both a domain name and a hostname) in your browser’s
-address bar, your computer asks the DNS service if it knows the actual
-address of the building that people call `gmail.com`. If it does, it will
-happily tell your computer the number of that building.
-
-Another DNS job is to store some meta data about these domain names. For
-such meta data there are record types, one of these types being the Mail
-eXchanger, or MX. This record of a domain tells the world who is handling
-incoming mails for the specified domain. For `gmail.com` this is
-`gmail-smtp-in.l.google.com` (among others; there can be multiple records of
-the same type, in which case they usually have priorities, too.)
-
-One more rule: when two computers talk to each other they use so called
-protocols. These protocols define a set of rules on how they should
-communicate; this includes message formatting, special code words and such.
-
-#### From your computer to the mail server
-
-Your MUA has two settings called SMTP server address SMTP port number (see
-about that later). SMTP stands for Simple Mail Transfer Protocol, and
-defines the rules on how your MUA, or another mail handling computer should
-communicate with a mail handling computer when *sending* mail. Most probably
-your Internet Service Provider gave you an SMTP server name, like
-`smtp.aol.com` and a port number like `587`.
-
-When you hit that send button of yours, your computer will check with the
-DNS service for the address of the SMTP server, which, for `smtp.aol.com`,
-is `64.12.88.133`. The computer puts this name/address pair into its memory,
-so it doesn’t have to ask the DNS again (this technique is called caching
-and is widely used wherever time consuming operations happen).
-
-Then it will send your message to the given port number of this newly
-fetched address. If you imagined computers as office buildings, you can
-imagine port numbers as departments and there can be 65535 of them in one
-building. The port number of SMTP is usually 25, 465 or 587 depending on
-many things we don’t cover here. Your MUA prepares your letter, adding your
-e-mail address and the recipients’, together with other information that may
-be useful for transferring your mail. It then puts this well formatted
-message in an envelope and writes “to building `64.12.88.133`, dept. `587`”,
-and puts it on the wire so it gets there (if the wire is broken, the
-building does not exist or there is no such department, you will get an
-error message from your MUA). Your address and the recipient’s address are
-inside the envelope; other than the MUA, your own computer is not concerned
-about it.
-
-The mailing department (or instead lets call it the Mail Transfer Agent,
-A.K.A. MTA) now opens this envelope and reads the letter. All of it, letter
-by letter, checking if your MUA formatted it well. More than probably it
-also runs your message through several filters to decide if you are a bad
-guy sending some unwanted letter (also known as spam), but most importantly
-it fetches the recipients address. It is possible, e.g. when you send an
-e-mail within the same organization, that the recipient’s address is handled
-by this very same computer. In this case the MTA puts the mail to the
-recipient’s mailbox and the next step is skipped.
-
-#### From one server to another
-
-Naturally, it is possible to send an e-mail from one company to another, so
-these MTAs don’t just wait for e-mails from you, but also communicate with
-each other. When you send a letter from your `example@aol.com` address to me
-at `gergely@polonkai.eu`, this is what happens.
-
-In this case, the MTA that initially received the e-mail from you (which
-happened to be your Internet Service Provider’s SMTP server) turns to the
-DNS again. It will ask for the MX record of the domain name specified by the
-e-mail address, (the part after the `@` character, in my case,
-`polonkai.eu`), because the server mentioned there must be contacted, so
-they can deliver your mail for me. My domain is configured so its primary MX
-record is `aspmx.l.google.com` and the secondary is
-`alt1.aspmx.l.google.com` (and 5 more. Google likes to play it safe.) The
-MTA then gets the first server name, asks the DNS for its address, and tries
-to send a message to the `173.194.67.27` (the address of
-`aspmx.l.google.com`), same department. But unlike your MUA, MTAs don’t have
-a pre-defined port number for other MTAs (although there can be exceptions).
-Instead, they use well-known port numbers, `465` and `25`. If the MTA on
-that server cannot be contacted for any reason, it tries the next one on the
-list of MX records. If none of the servers can be contacted, it will retry
-based on a set of rules defined by the administrators, which usually means
-it will retry after 1, 4, 24 and 48 hours. If there is still no answer after
-that many attempts, you will get an error message back, in the form of an
-e-mail sent directly by the SMTP server.
-
-Once the other MTA could be contacted, your message is sent there. The
-original envelope you used is discarded, and a new one is used with the
-address and dept. number (port) of the receiving MTA. Also, your message
-gets altered a little bit, as most MTAs are kind enough (ie. not sneaky) to
-add a clause to your message stating “the MTA at <organization> has checked
-and forwarded this message.”
-
-It is possible, though not likely, that your message gets through more than
-two MTAs (one at your ISP and one at the receiver’s) before arriving to its
-destination. At the end, an MTA will say that “OK, this recipient address is
-handled by me”, your message stops and stays there, put in your peer’s
-mailbox.
-
-##### The mailbox
-
-Now that the MTA has passed your mail to the mailbox team (I call it a team
-instead of department because the tasks described here are usually handled
-by the MTA, too), it reads it. (Pesky little guys are these mail handling
-departments, aren’t they?) If the mailbox has some filtering rules, like “if
-XY sends me a letter, mark it as important” or “if the letter has a specific
-word in its subject, put it in the XY folder”, it executes them, but the
-main point is to land the message in the actual post box of the recipient.
-
-#### From the post box to the recipients computer
-
-When the recipient opens their MUA, it will look to a setting usually called
-“Incoming mail server”. Just like the SMTP server, it has a name and port
-number, along with a server type. This type can vary from provider to
-provider, and is usually one of POP3 (pretty old protocol, doesn’t even
-support folders on its own), IMAP (a newer one, with folders and message
-flags like “important”), MAPI (a dialect of IMAP, created by Microsoft as
-far as I know), or plain old mbox files on the receiving computer (this last
-option is pretty rare nowadays, so I don’t cover this option. Also, if you
-use these, you most probably don’t really need this article to understand
-how these things work.) This latter setting defines the protocol, telling
-your MUA how to “speak” to the post box.
-
-So your MUA turns to the DNS once more to get the address of your incoming
-mail server and contacts it, using the protocol set by the server type. At
-the end, the recipients computer will receive a bunch of envelopes including
-the one that contains your message. The MUA opens them one by one and reads
-them, making a list ordered by their sender or subject, or the date of
-sending.
-
-#### From the recipient’s comupter to their eyes
-
-When the recipient then clicks on one of these mails, the MUA will fetch all
-the relevant bits like the sender, the subject line, the date of sending and
-the contents itself and sends it to the “printing” department (I use quotes
-as they don’t really print your mail on paper, they just convert it to a
-nice image so the recipient can see it. This is sometimes referred to as a
-rendering engine). Based on a bunch of rules they pretty-print it and send
-it to your display as a new series of Morse codes. Your display then decides
-how it will present it to the user: draw the pretty pictures if it is a
-computer screen, or just raise and lower some hard dots that represents
-letters on a Braille terminal.

_posts/2016-01-13-emacs-implement-a-gobject-s-virtual-function.md

@@ -1,53 +0,0 @@
----
-layout:    post
-title:     "Emacs: Implement a GObject’s virtual function"
-date:      2016-01-13 13:31:12
-tags:      [c, development, emacs]
-published: true
-author:
-  name: "Gergely Polonkai"
-  email: "gergely@polonkai.eu"
----
-
-I have recently started creating a GLib implementation of the
-Matrix.org API. For that, I have created a GObject interface,
-MatrixAPI, which has as many virtual functions as API calls (which is
-a lot, and expanding). This way I ended up with the following scenario.
-
-In `matrix-api.h` I had a struct like this, with a lot more elements:
-
-    typedef struct {
-        void (*initial_sync)(MatrixAPI *api,
-                             MatrixAPICallback callback,
-                             gpointer user_data,
-                             GError **error);
-        void (*sync)(MatrixAPI *api,
-                     MatrixAPICallback callback,
-                     gpointer user_data,
-                     GError **error);
-        …
-
-And in `matrix-http-api.c`, which implements `MatrixAPI`, I have a
-function like this (again, with a lot more elements):
-
-    static void
-    matrix_http_api_matrix_api_init(GObjectInterface *iface)
-    {
-        iface->initial_sync = i_initial_sync;
-        iface->sync = i_sync;
-        …
-    }
-
-And every time I wanted to implement a new function from the vtable, I
-had to copy the prototype, and add an `iface->foo_bar = i_foo_bar`
-line and an actual function header for `i_foo_bar` with the same
-parameters. That’s a cumbersome job for more than 40 function
-headers. But emacs comes to the rescue!
-
-{% gist bfd36be8b515edced3d2 implement-gobject-vfunc.el %}
-
-Now all I have to do is to copy the whole vtable entry into
-`matrix_http_api_matrix_api_init()`, execute `M-x
-implement-gobject-vfunc`, then put the same vtable entry somewhere
-before the interface init function, and execute `M-x
-implement-gobject-vfunc-prototype`.

_posts/2016-02-26-vala-interface-madness.md

@@ -1,33 +0,0 @@
----
-layout:    post
-title:     "Vala interface madness"
-date:      2016-02-26 13:07:52
-tags:      [vala, development]
-published: true
-author:
-  name: "Gergely Polonkai"
-  email: "gergely@polonkai.eu"
----
-
-Although I have just started making it in C, I decided to move my
-Matrix GLib SDK to Vala. First to learn a new language, and second
-because it is much easier to write GObject based stuff with it.
-
-For the first step I created a `.vapi` file from my existing sources,
-so the whole SDK prototype was available for me in Vala.
-
-I had a `MatrixEvent` class that implemented the `GInitable`
-interface, and many others were subclassed `MatrixEvent`. For some
-reason I don’t remember, I created the following header for one of the
-event classes:
-
-    public class MatrixPresenceEvent : GLib.Object, GLib.Initable {
-
-This is nice and everything, but as I didn’t create an `init()` method
-for `MatrixPresenceEvent`, it tried to use the one from the parent
-class and somehow got into an infinite loop. The Vala transformer
-(`valac`), however, doesn’t mention this.
-
-Lessons learned: if you implement an interface on a subclass that is
-implemented by the parent don’t forget to add the necessary functions
-to the subclass.

_posts/2016-09-28-emacs-package-to-generate-gobject-boilerplate.md

@@ -1,46 +0,0 @@
----
-layout:    post
-title:     "Emacs package to generate GObject boilerplate"
-date:      2016-09-28 15:40:15
-tags:      [gnome, development]
-published: true
-author:
-    name: Gergely Polonkai
-    email: gergely@polonkai.eu
----
-
-Before I started using Vala (and sometimes even after that) I often
-needed to generate new classes based
-on [GObject](https://developer.gnome.org/gobject/stable/).
-
-If you have ever worked with GObject in C, you know how tedious it can
-be. You need a pretty long boilerplate just to register your class,
-and, if you want to be introspectable (and readable, actually), your
-function names can grow really long.
-
-To overcome this problem back in my ViM days, I used template files,
-where I could replace class prefixes and names with a few keyboard
-macros. As I never really dug into ViM scripting other than using some
-plugins, I never got farther than
-that. [Then came Emacs]({% post_url 2014-09-17-nyanmacs %}).
-
-I use Emacs for about two years now very extensively, up to and
-including GLib-based development. I tried the template approach, but
-it felt to be a really poor experience, especially given that I made
-my feet wet with Emacs Lisp. So I dug deeper, and created a package
-for that.
-
-![A screenshot of GobGen in action]({{ site_url }}/images/screenshot-gobgen.png)
-
-GobGen has its own buffer with some widgets, a bit similar to
-`customize`. You can enter the name of your new object and its parent,
-specify some settings. Then you press Generate, and you are presented
-with two new buffers, one for the `.c` and another for the `.h`
-boilerplate.
-
-There are a lot of things to do, actually. There is already an open
-issue for creating a major mode for this buffer, and there are some
-minor switches I’d like to add, but it is already usable. You can grab
-it from [MELPA](https://melpa.org/#/gobgen) (my first package there;
-woo!) or from
-my [GitHub account](https://github.com/gergelypolonkai/gobgen.el).

_posts/2016-10-04-git-merge-stages.md

@@ -1,65 +0,0 @@
----
-layout:    post
-title:     "git-merge stages"
-date:      2016-10-04 12:46:00
-tags:      [git]
-published: true
-author:
-    name: Gergely Polonkai
-    email: gergely@polonkai.eu
----
-
-This was a mail to my company’s internal Git mailing list, after I
-realised many colleagues can’t wrap their heads around merge
-conflicts.
-
->Hello all,
->
->I just saw this on
->the [git-users](https://groups.google.com/forum/#!forum/git-users)
->list and thought it could help you when you bump into a merge
->conflict. It is an excerpt from a mail by Konstantin Khomoutov (one
->of the main contributors on the list), with a few modifications of
->mine. Happy debugging :)
->
->>When a merge conflict is detected for a file, Git:
->>
->>1. Updates the entry for that file in the index to make it contain
->>   several so-called “stages”:
->>  * `0`: “Ours” version – that one which was there in this index entry
->>    before we begun to merge. At the beginning of the conflict, like
->>    right after the `git merge` or `git rebase` command this won’t
->>    exist (unless you had the file in the index, which you didn’t, did
->>    you?). When you resolve the conflict and use `git add
->>    my/conflicting/file.cc`, this will be the version added to the
->>    staging area (index), thus, the resolution of the conflict.
->>  * `1`: The version from the common ancestor commit, ie. the version
->>    of the file both of you modified.
->>  * `2`: The version from `HEAD`. During a merge, this is the current
->>    branch. During a rebase, this is the branch or commit you are
->>    rebasing onto, which usually will be `origin/develop`).
->>  * `3`: The version being merged, or the commit you are rebasing.
->>2. Updates the file in the work tree to contain conflict markers and
->>   the conflicting chunks of text between them (and the text from the
->>   common ancestor if the `diff3` style of conflict markers was set).
->>
->>Now you can use the numbers in point 1 to access the different stages
->>of the conflicting file. For example, to see the common ancestor (the
->>version both of you modified), use
->>
->>```
->>git show :1:my/conflicting/file.cc
->>```
->>
->>Or, to see the difference between the two conflicting versions, try
->>
->>```
->>git diff :2:my/conflicting/file.cc :3:my/conflicting/file.cc
->>```
->>
->>**Note** that you can’t use the `:0:` stage *before* you stage your
->>resolution with `git add`, and you can’t use the `:2:` and `:3:`
->>stages *after* you staged the resolution.
->>
->>Fun fact: behind the scenes, these are the files (*revisions*) `git mergetool`
->>accesses when it presents you the conflict visually.

_posts/2016-11-03-how-i-started-with-emacs.md

@@ -1,98 +0,0 @@
----
-layout:    post
-title:     "How I started with Emacs"
-date:      2016-11-03 09:58:41
-tags:      [emacs]
-published: true
-authon:
-    name: Gergely Polonkai
-    email: gergely@polonkai.eu
----
-
-Sacha Chua has a nice
-[Emacs chat intro](http://sachachua.com/blog/2013/04/emacs-chat-intro/) article
-back from 2013. I write this post half because she asks there about my
-(OK, anyone’s) first Emacs moments, and half because I plan to do it
-for months now.
-
-I wanted to start using Emacs 6(ish) years ago, and I was like
-“<kbd>C-x</kbd> what”? (Note that back around 1998, I was among the
-people who exited `vi` by killing it from another terminal after a
-bunch of tries & fails like
-[these](http://osxdaily.com/2014/06/12/how-to-quit-vim/).)
-
-I tried to come back to Emacs a lot of times. And I mean a *lot*,
-about every two months. I suddenly learned what these cryptic key
-chord descriptions mean (`C` is for <kbd>Control</kbd> and `M` is for
-<kbd>Meta</kbd>, which is actually <kbd>Alt</kbd>), but somehow it
-didn’t *click*. I remained a ViM power user with a huge pile of
-3<sup>rd</sup> party plugins.
-Then [I found Nyan-macs]({% post_url 2014-09-17-nyanmacs %}),
-which converted me to Emacs, and it is final now. Many of my friends
-thought I’m just kidding this being the cause, but I’m not. I’m a huge
-fan of Nyan cat (did you know there is even a site
-called [nyan.cat](http://nyan.cat/)?) and since then I have it in my
-mode line:
-
-![Nyan modeline]({{ site_url }}/images/nyan-modeline.png)
-
-…in my `eshell` prompt:
-
-![IMAGE HERE]({{ site_url }}/images/nyan-eshell.png)
-
-…and I also [zone out](https://www.emacswiki.org/emacs/ZoneMode) with
-Nyan cat:
-
-![IMAGE HERE]({{ site_url }}/images/nyan-zone.png)
-
-Now on to more serious stuff. After browsing through all the packages
-provided by [ELPA](http://elpa.gnu.org/), I found tons of useful (and
-sometimes, less useful) packages,
-like
-[Helm](https://github.com/emacs-helm/helm/wiki),
-[company](http://company-mode.github.io/),
-[gtags](https://www.emacswiki.org/emacs/GnuGlobal) (which introduced
-me to GNU Global, removing Exuberant ctags from my
-life),
-[magit](https://magit.vc/),
-[Projectile](http://batsov.com/projectile/),
-and [Org](http://orgmode.org/) (OK, it’s actually part of Emacs for a
-while, but still). I still use these few, but in a month or two, I
-started
-to [version control](https://github.com/gergelypolonkai/my-emacs-d) my
-`.emacs.d` directory, so I can easily transfer it between my home and
-work machine (and for a few weeks now, even to my phone: I’m using
-Termux on Android). Then, over these two years I wrote some packages
-like [GobGen](https://github.com/gergelypolonkai/gobgen.el), and a
-small addon for Calendar
-providing
-[Hungarian holidays](https://github.com/gergelypolonkai/hungarian-holidays),
-and I found a lot more (in no particular
-order):
-[git-gutter](https://github.com/syohex/emacs-git-gutter),
-[multiple-cursors](https://github.com/magnars/multiple-cursors.el),
-[origami](https://github.com/gregsexton/origami.el),
-[ace-window](https://github.com/abo-abo/ace-window),
-[avy](https://github.com/abo-abo/avy),
-[beacon](https://github.com/Malabarba/beacon), and a lot more.
-
-What is more important (to me) is that I started using
-the [use-package](https://github.com/jwiegley/use-package) package,
-which can automatically download packages that are not installed on my
-current local system. Together
-with
-[auto-package-update](https://github.com/rranelli/auto-package-update.el),
-it is *very* practical.
-
-In addition, I started to follow the blogs of a bunch of Emacs
-users/gurus. I’ve already
-mentioned [Sacha Chua](http://sachachua.com/). She’s a charming,
-cheerful person, writing a lot about Emacs and project management
-(among other things). Another one
-is [Bozhidar Batsov](http://batsov.com/), who, among other things, had
-an initiate to lay down the foundation of
-a
-[common Elisp coding style](https://github.com/bbatsov/emacs-lisp-style-guide). Another
-favourite of mine
-is [Endless Parentheses](http://endlessparentheses.com/), whence I got
-a lot of ideas.

_posts/2016-11-10-edit-file-as-other-user-in-emacs.md

@@ -1,27 +0,0 @@
----
-layout:    post
-title:     "Edit file as another user in Emacs"
-date:      2016-11-10 08:57:12
-tags:      [development, emacs]
-published: true
-author:
-    name: Gergely Polonkai
-    email: gergely@polonkai.eu
----
-
-I have recently found
-[this article](http://emacsredux.com/blog/2013/04/21/edit-files-as-root/) by
-Bozhidar Batsov on opening the current file as root. I barely use
-[tramp](https://www.gnu.org/software/tramp/) for sudo access, but when I do,
-I almost never use root as the target user. So I decided to fix it for my
-needs.
-
-{% gist 192c83aa0556d5cdaf4018f57b75a84b %}
-
-If the user is not specified, the default is still root. Also, if the
-current buffer is not visiting a file, I prompt for a filename. As I’m not
-an `ido` user, I didn’t bother calling
-`ido-read-file-name`; [`helm`](https://github.com/emacs-helm/helm/wiki)
-overrides `read-file-name` for me anyway.
-
-Unlike Bozhidar, I barely use this feature, so I didn’t bind this to a key.

_posts/2016-11-18-get-passwords-from-id-manager.md

@@ -1,28 +0,0 @@
----
-layout:    post
-title:     "Get account data programatically from id-manager"
-date:      2016-11-18 12:43:13
-tags:      [emacs]
-published: true
-author:
-    name: Gergely Polonkai
-    email: gergely@polonkai.eu
----
-
-I recently started
-using [`id-manager`](https://github.com/kiwanami/emacs-id-manager). It is a
-nice little package that can store your passwords, encrypting them with
-GPG. My original reason was to store my GitHub access token
-for [`github-notifier`](https://github.com/xuchunyang/github-notifier.el),
-but it soon turned out, it’s not *that* easy.
-
-`id-manager` is a nice package when it comes to storing your password, and
-retrieving them for your own eyes. But it cannot retrieve account data
-programatically. Taking a look into its source code, I came up with this
-solution:
-
-{% gist 8bad70502ac563864080f754fce726c3 idm.el %}
-
-I currently need only the account ID (ie. the username) and the password,
-but it’s pretty easy to add a macro to get the `memo` or `update-time`
-fields, too.

_posts/2016-11-30-add-python-docstring-to-the-beginning-of-anything.md

@@ -1,47 +0,0 @@
----
-layout:    "post"
-title:     "Add Python docstring to the beginning of anything in Emacs"
-date:      2016-11-30 07:52:37
-tags:      [development, python, emacs]
-published: true
-author:
-    name: Gergely Polonkai
-    email: gergely@polonkai.eu
----
-
-Now that I write Python code for a living, I write a lot of functions,
-classes, and modules.  What I still tend to forget, and also find tedious,
-is adding docstrings.  Unlike many developers, writing documentation is not
-an enemy of mine, but it usually comes to my mind when I finish
-implementation.  The procedure, roughly, is this:
-
-* Decide where I am (in a function, in a class but not in one of its
-  methods, or not inside such a block at all)
-* Go to the beginning of the thing
-* Insert `"""`
-* Leave a blank line
-* Insert `"""`
-
-One of my mottos is if something takes more than one step and you have to do
-it more than twice, you should automate it after the first time.  This puts
-a small(ish) overhead on the second invocation (when you implement the
-automation), but it usually worth the time.
-
-Since I use Emacs for pretty much everything coding-related (and many more,
-but that’s the topic of a different post), I wrote a small function to do it
-for me.
-
-{% gist 7b062a00d3b8a2555024521273cecfee python-docstring.el %}
-
-There are still a lot of things to improve:
-
-* it always inserts double quotes (althoug I couldn’t show a use-case when
-single quotes are preferred)
-* it doesn’t check for an existing docstring, just happily inserts a new one
-(leaving the old one alone, but generating a syntax error this way)
-* it would also be nice if I could jump to the beginning of a file even from
-a class method.  I guess I will use prefix keys for that, but I’m not sure
-yet.
-
-You can bet I will implement these features, so check back soon for an
-updated version!

_posts/2016-12-08-slugify-in-python3.md

@@ -1,20 +0,0 @@
----
-layout:    post
-title:     "Slugify in Python 3"
-date:      2016-12-08 12:54:19
-tags:      [development, python]
-published: true
-author:
-    name: Gergely Polonkai
-    email: gergely@polonkai.eu
----
-
-Today I needed a function to create a slug (an ASCII-only representation of
-a string).  I went Googling a bit, and found an
-excellend [Flask snippet](http://flask.pocoo.org/snippets/5/).  Problem is,
-it is designed for Python 2, so I came up with a Python 3 version.
-
-{% gist 1866fd363f75f4da5f86103952e387f6 slugify.py %}
-
-As I don’t really like the transliteration done in the first example
-(e.g. converting ü to ue), I went with the second example.

_posts/2016-12-22-finding-non-translated-strings-in-python-code.md

@@ -1,115 +0,0 @@
----
-layout:    post
-title:     "Finding non-translated strings in Python code"
-date:      2016-12-22 09:35:11
-tags:      [development, python]
-published: true
-author:
-    name: Gergely Polonkai
-    email: gergely@polonkai.eu
----
-
-When creating multilingual software, be it on the web, mobile, or desktop,
-you will eventually fail to mark strings as translatable.  I know, I know,
-we developers are superhuman and never do that, but somehow I stopped
-trusting myself recently, so I came up with an idea.
-
-Right now I assist in the creation of a multilingual site/web application,
-where a small part of the strings come from the Python code instead of HTML
-templates.  Call it bad practice if you like, but I could not find a better
-way yet.
-
-As a start, I tried to parse the source files with simple regular
-expressions, so I could find anything between quotation marks or
-apostrophes.  This attempt quickly failed with strings that had such
-characters inside, escaped or not; my regexps became so complex I lost all
-hope.  Then the magic word “lexer” came to mind.
-
-While searching for ready made Python lexers, I bumped into the awesome
-`ast` module.  AST stands for Abstract Syntax Tree, and this module does
-that: parses a Python file and returns a tree of nodes.  For walking through
-these nodes there is a `NodeVisitor` class (among other means), which is
-meant to be subclassed.  You add a bunch of `visitN` methods (where `N` is
-an `ast` class name like `Str` or `Call`), instantiate it, and call its
-`visit()` method with the root node.  For example, the `visitStr()` method
-will be invoked for every string it finds.
-
-#### How does it work?
-
-Before getting into the details, let’s me present you the code I made:
-
-{% gist 1a16a47e5a1971ca33e58bdfd88c5059 string-checker.py %}
-
-The class initialization does two things: creates an empty `in_call` list
-(this will hold our primitive backtrace), and saves the filename, if
-provided.
-
-`visitCall`, again, has two tasks.  First, it checks if we are inside a
-translation function.  If so, it reports the fact that we are translating
-something that is not a raw string.  Although it is not necessarily a bad
-thing, I consider it bad practice as it may result in undefined behaviour.
-
-Its second task is to walk through the positional and keyword arguments of
-the function call.  For each argument it calls the `visit_with_trace()`
-method.
-
-This method updates the `in_call` property with the current function name
-and the position of the call.  This latter is needed because `ast` doesn’t
-store position information for every node (operators are a notable example).
-Then it simply visits the argument node, which is needed because
-`NodeVisitor.visit()` is not recursive.  When the visit is done (which, with
-really deeply nested calls like `visit(this(call(iff(you(dare)))))` will be
-recursive), the current function name is removed from `in_call`, so
-subsequent calls on the same level see the same “backtrace”.
-
-The `generic_visit()` method is called for every node that doesn’t have a
-named visitor (like `visitCall` or `visitStr`.  For the same reason we
-generate a warning in `visitCall`, we do the same here.  If there is
-anything but a raw string inside a translation function call, developers
-should know about it.
-
-The last and I think the most important method is `visitStr`.  All it does
-is checking the last element of the `in_call` list, and generates a warning
-if a raw string is found somewhere that is not inside a translation function
-call.
-
-For accurate reports, there is a `get_func_name()` function that takes an
-`ast` node as an argument.  As function call can be anything from actual
-functions to object methods, this goes all down the node’s properties, and
-recursively reconstructs the name of the actual function.
-
-Finally, there are some test functions in this code.  `tst` and
-`actual_tests` are there so if I run a self-check on this script, it will
-find these strings and report all the untranslated strings and all the
-potential problems like the string concatenation.
-
-#### Drawbacks
-
-There are several drawbacks here.  First, translation function names are
-built in, to the `TRANSLATION_FUNCTIONS` property of the `ShowString` class.
-You must change this if you use other translation functions like
-`dngettext`, or if you use a translation library other than `gettext`.
-
-Second, it cannot ignore untranslated strings right now.  It would be great
-if a pragma like `flake8`’s `# noqa` or `coverage.py`’s `# pragma: no cover`
-could be added.  However, `ast` doesn’t parse comment blocks, so this proves
-to be challenging.
-
-Third, it reports docstrings as untranslated.  Clearly, this is wrong, as
-docstrings generally don’t have to be translated.  Ignoring them, again, is
-a nice challenge I couldn’t yet overcome.
-
-The `get_func_name()` helper is everything but done.  As long as I cannot
-remove that final `else` clause, there may be error reports.  If that
-happens, the reported class should be treated in a new `elif` branch.
-
-Finally (and the most easily fixed), the warnings are simply printed on the
-console.  It is nice, but it should be optional; the problems identified
-should be stored so the caller can obtain it as an array.
-
-#### Bottom line
-
-Finding strings in Python sources is not as hard as I imagined.  It was fun
-to learn using the `ast` module, and it does a great job.  Once I can
-overcome the drawbacks above, this script will be a fantastic piece of code
-that can assist me in my future tasks.

_posts/2017-01-02-rename-automatically-named-foreign-keys-with-alembic.md

@@ -1,34 +0,0 @@
----
-layout:    post
-title:     "Rename automatically named foreign keys with Alembic"
-date:      2017-01-02 09:41:23
-tags:      [mysql, development, flask, python]
-published: true
-author:
-    name: Gergely Polonkai
-    email: gergely@polonkai.eu
----
-
-I have recently messed up my Alembic migrations while modifying my
-SQLAlchemy models.  To start with, I didn’t update the auto-generated
-migration files to name the indexes/foreign keys a name, so Alembic used its
-own naming scheme.  This is not an actual problem until you have to modify
-columns that have such constraints.  I have since fixed this problem, but
-first I had to find which column references what (I had no indexes other
-than primary key back then, so I could go with foreign keys only).  Here is
-a query I put together, mostly using
-[this article](http://www.binarytides.com/list-foreign-keys-in-mysql/).
-
-``` sql
-SELECT constraint_name,
-       CONCAT(table_name, '.', column_name) AS 'foreign key',
-       CONCAT(referenced_table_name, '.', referenced_column_name) AS 'references'
-FROM information_schema.key_column_usage
-WHERE referenced_table_name IS NOT NULL AND
-      table_schema = 'my_app';
-```
-
-Now I could easily drop such constraints using
-`alembic.op.drop_constraint('users_ibfk1', 'users', type_='foreignkey')` and
-recreate them with `alembic.op.create_foreign_key('fk_user_client', 'users',
-'clients', ['client_id'], ['id'])`

_posts/2017-03-26-category-based-logging-with-flask.md

@@ -1,91 +0,0 @@
----
-layout:    post
-title:     "Category-based logging with Flask"
-date:      2017-03-26 22:00:52
-tags:      [development, python, flask]
-published: true
-author:
-    name: Gergely Polonkai
-    email: gergely@polonkai.eu
----
-
-I’m in a team who are developing a Flask-based web application, which uses
-logging extensively.  For a while now it spews out a lot of lines so the
-need arose to index them in ElasticSearch, and more importantly, to search
-through them for auditing purposes.  This latter user story brought up one
-more question: why don’t we categorize our log messages?  I quickly came up
-with an extended log format (`[auth]` is the new category name):
-
-    [2017-01-14 00:55:42,554] [8286] [INFO] [auth] invalid password for john@example.com [at __init__.py:12, in function utils.validate_login]
-
-Here, `[auth]` is the category name.  In the ideal solution, all I’d have to
-do is adding `%(category)s` to my formatter, and I could call
-`app.logger.info('auth', 'invalid password')` to achieve this output.
-Unfortunately, `Flask.logger` (and, in the background, the `logging` module)
-is not that easy to tame.
-
-As it turns out, a Flask application’s `logger` property is a subclass of
-`logging.Logger`, so my first idea was to monkey patch that class.  When the
-app’s logger is initialised, it subclasses `logging.Logger` and tweaks the
-log level so it goes down to `logging.DEBUG` if the app is running in debug
-mode.  This is done by using a different logger class depending on the app
-config.  Fortunately it doesn’t directly subclass `logging.Logger`; it calls
-`logging.getLoggerClass()` to find which class it should extend.  To achieve
-my goals, all I had to do is to subclass the original logger class, and pass
-it to `logging.setLoggerClass()` *before* I initialise my app, and I have a
-fail-safe(ish) solution.  So far so good, on to the extra category
-parameter.
-
-Now if you add a new variable to the formatter like my new `%(category)s`,
-you get a nifty `KeyError` saying there is no `category` in the format
-expansion dictionary.  If you add `category='auth` to the
-`app.logger.info()` calls and its cousins, it’s fine, because these methods
-use the magic `**kwarg` argument to swallow it. Everything goes well until
-control arrives to the `_log()` method: it complains about that extra
-`category` keyword argument.  Taking a peek at Python’s internals, I found
-two things: `info()`, `error()`, and co. pass `*args` and `**kwargs` to
-`_log()` unmodified, and the `_log()` method doesn’t have `**kwargs`
-present in its definition to swallow it.  A little doc reading later I found
-that if I want to pass extra arguments for such a formatter, I should do it
-via the `extra` keyword argument to `_log()`.  A call like
-`app.logger.info('invalid password', extra={'category': 'auth'})` solved the
-problem.  Now *that* is tedious.
-
-My first idea was to override all the standard logging methods like `info()`
-and `error()`, and handle `category` there.  But this resulted in lots of
-repeating code.  I changed the specification a bit, so my calls would look
-like `info('message', category='auth)` instead of the original plan of
-`info('auth', 'message')`: as the logging methods pass all keyword arguments
-to `_log()`, I can handle it there.  So at the end, my new logger class
-only patches `_log()`, by picking out `category` from the kwarg list, and
-inserting it to `extra` before calling `super`.
-
-As you can see, this is a bit ugly solution.  It requires me, the app
-author, to know about Flask’s internals (that I can set my own logging class
-before the app is created, and so the app will use it.)  This means if Flask
-developers change the way how logging is done, I have to adapt and find a
-workaround for the new version (well, unless they let me directly set the
-logging class.  That would make it easy.)
-
-What is worse, I must know about Python internals.  I know the `extra` kwarg
-is documented well (I just failed to notice), but this made adding a new
-formatter variable so hard.  Python version doesn’t change as often as Flask
-version in this project, and I think the standard library won’t really
-change until 4.0, so I don’t think my tampering with a “protected” method
-will cause any trouble in the future.  Still, this makes me feel a bit
-uneasy.
-
-All the above can be circumvented if this class, and the whole solution have
-some tests.  As my class uses the same method as Flask (use
-`logging.getLoggerClass()` as a base class instead of using
-`logging.Logger()` directly), if the base logger class changes in Python or
-in the running environment, my app won’t care.  By checking if the app
-logger can use my special `category` variable (ie. it doesn’t raise an
-exception *and* the category actually gets into the log output), I made sure
-my class is used as a base in Flask, so if they change the way they
-construct `app.logger`, I will know about it when I first run my tests after
-upgrading Flask.
-
-If you are interested in such functionality (and more), you can grab it
-from [GitHub](https://github.com/gergelypolonkai/flask-logging-extras), or
-via [PyPI](https://pypi.python.org/pypi/Flask-Logging-Extras/).

_posts/2017-10-02-add-sysadmin-day-to-emacs-calendar.md

@@ -1,27 +0,0 @@
----
-layout:    post
-title:     "Add SysAdmin day to Emacs Calendar"
-date:      2017-10-02 09:37:52
-tags:      [emacs]
-published: true
-author:
-    name: Gergely Polonkai
-    email: gergely@polonkai.eu
----
-I’m a SysAdmin since 1998.  Maybe a bit earlier, if you count managing our home computer.  This
-means [SysAdmin Day](http://sysadminday.com/) is also celebrating me.  However, my Emacs Calendar
-doesn’t show it for some reason.
-
-The solution is pretty easy:
-
-``` lisp
-(add-to-list 'holiday-other-holidays '(holiday-float 7 5 -1 "SysAdmin Day") t)
-```
-
-Now invoke `holidays-list` for any year, choosing “Other” as the category, and there you go:
-
-```
-…
-Friday, July 28, 2017: SysAdmin Day
-…
-```

_posts/2018-07-26-check-if-last-git-commit-has-test-coverage.md

@@ -1,214 +0,0 @@
----
-layout:    post
-title:     "Check if the last Git commit has test coverage"
-date:      2018-07-26 12:49:52
-tags:      [python,development,testing]
-published: true
-author:
-    name: Gergely Polonkai
-    email: gergely@polonkai.eu
----
-
-I use Python at work and for private projects.  I also aim to write tests for my code, especially
-recently.  And as I usually don’t start from 100% code coverage (TDD is not my game), I at least
-want to know if the code I just wrote have full coverage.
-
-The trick is to collect all the lines that changed, and all the lines that has no coverage.  Then
-compare the two, and you have the uncovered lines that changed!
-
-### Getting the list of changed lines
-
-Recently, I bumped into
-[this article](https://adam.younglogic.com/2018/07/testing-patch-has-test/).  It is a great awk
-script that lists the lines that changed in the latest commit.  I have really no problem with awk,
-but I’m pretty sure it can be done in Python, as that is my main language nowadays.
-
-```python
-def get_changed_lines():
-    """Get the line numbers that changed in the last commit
-    """
-
-    git_output = subprocess.check_output('git show', shell=True).decode('utf-8')
-
-    current_file = None
-    lines = {}
-    left = 0
-    right = 0
-
-    for line in git_output.split('\n'):
-        match = re.match(r'^@@ -([0-9]+),[0-9]+ [+]([0-9]+),[0-9]+ @@', line)
-
-        if match:
-            left = int(match.groups()[0])
-            right = int(match.groups()[1])
-
-            continue
-
-        if re.match(r'^\+\+\+', line):
-            current_file = line[6:]
-
-            continue
-
-        if re.match(r'^-', line):
-            left += 1
-
-            continue
-
-        if re.match(r'^[+]', line):
-            # Save this line number as changed
-            lines.setdefault(current_file, [])
-            lines[current_file].append(right)
-            right += 1
-
-            continue
-
-        left += 1
-        right += 1
-
-    return lines
-```
-
-OK, not as short as the awk script, but works just fine.
-
-### Getting the uncovered lines
-
-Coverage.py can list the uncovered lines with `coverage report --show-missing`.  For Calendar.social, this looks something like this:
-
-```
-Name                                     Stmts   Miss  Cover   Missing
-----------------------------------------------------------------------
-calsocial/__init__.py                      173     62    64%   44, 138-148, 200, 239-253, 261-280, 288-295, 308-309, 324-346, 354-363
-calsocial/__main__.py                        3      3     0%   4-9
-calsocial/account.py                       108     51    53%   85-97, 105-112, 125, 130-137, 148-160, 169-175, 184-200, 209-212, 221-234
-calsocial/app_state.py                      10      0   100%
-calsocial/cache.py                          73     11    85%   65-70, 98, 113, 124, 137, 156-159
-calsocial/calendar_system/__init__.py       10      3    70%   32, 41, 48
-calsocial/calendar_system/gregorian.py      77      0   100%
-calsocial/config_development.py             11     11     0%   4-17
-calsocial/config_testing.py                 12      0   100%
-calsocial/forms.py                         198     83    58%   49, 59, 90, 136-146, 153, 161-169, 188-195, 198-206, 209-212, 228-232, 238-244, 252-253, 263-267, 273-277, 317-336, 339-342, 352-354, 362-374, 401-413
-calsocial/models.py                        372     92    75%   49-51, 103-106, 177, 180-188, 191-200, 203, 242-248, 257-268, 289, 307, 349, 352-359, 378, 392, 404-409, 416, 444, 447, 492-496, 503, 510, 516, 522, 525, 528, 535-537, 545-551, 572, 606-617, 620, 652, 655, 660, 700, 746-748, 762-767, 774-783, 899, 929, 932
-calsocial/security.py                       15      3    80%   36, 56-58
-calsocial/utils.py                          42      5    88%   45-48, 52-53
-----------------------------------------------------------------------
-TOTAL                                     1104    324    71%
-```
-
-All we have to do is converting these ranges into a list of numbers, and compare it with the
-result of the previous function:
-
-```python
-def get_uncovered_lines(changed_lines):
-    """Get the full list of lines that has not been covered by tests
-    """
-
-    column_widths = []
-    uncovered_lines = {}
-
-    for line in sys.stdin:
-        line = line.strip()
-
-        if line.startswith('---'):
-            continue
-
-        if line.startswith('Name '):
-            match = re.match(r'^(Name +)(Stmts +)(Miss +)(Cover +)Missing$', line)
-            assert match
-
-            column_widths = [len(col) for col in match.groups()]
-
-            continue
-
-        name = [
-            line[sum(column_widths[0:idx]):sum(column_widths[0:idx]) + width].strip()
-            for idx, width in enumerate(column_widths)][0]
-        missing = line[sum(column_widths):].strip()
-
-        for value in missing.split(', '):
-            if not value:
-                continue
-
-            try:
-                number = int(value)
-            except ValueError:
-                first, last = value.split('-')
-                lines = range(int(first), int(last) + 1)
-            else:
-                lines = range(number, number + 1)
-
-            for lineno in lines:
-                if name in changed_lines and lineno not in changed_lines[name]:
-                    uncovered_lines.setdefault(name, [])
-                    uncovered_lines[name].append(lineno)
-
-    return uncovered_lines
-```
-
-At the end we have a dictionary that has filenames as keys, and a list of changed but uncovered
-lines.
-
-### Converting back to ranges
-
-To make the final result more readable, let’s convert them back to a nice `from_line-to_line`
-range list first:
-
-```python
-def line_numbers_to_ranges():
-    """List the lines that has not been covered
-    """
-
-    changed_lines = get_changed_lines()
-    uncovered_lines = get_uncovered_lines(changed_lines)
-
-    line_list = []
-
-    for filename, lines in uncovered_lines.items():
-        lines = sorted(lines)
-        last_value = None
-
-        ranges = []
-
-        for lineno in lines:
-            if last_value and last_value + 1 == lineno:
-                ranges[-1].append(lineno)
-            else:
-                ranges.append([lineno])
-
-            last_value = lineno
-
-        range_list = []
-
-        for range_ in ranges:
-            first = range_.pop(0)
-
-            if range_:
-                range_list.append(f'{first}-{range_[-1]}')
-            else:
-                range_list.append(str(first))
-
-        line_list.append((filename, ', '.join(range_list)))
-
-    return line_list
-```
-
-### Printing the result
-
-Now all that is left is to print the result on the screen in a format digestable by a human being:
-
-```python
-def tabular_print(uncovered_lines):
-    """Print the list of uncovered lines on the screen in a tabbed format
-    """
-
-    max_filename_len = max(len(data[0]) for data in uncovered_lines)
-
-    for filename, lines in uncovered_lines:
-        print(filename.ljust(max_filename_len + 2) + lines)
-```
-
-And we are done.
-
-### Conclusion
-
-This task never seemed hard to accomplish, but somehow I never put enough energy into it to make
-it happen.  Kudos to Adam Young doing some legwork for me!

about.html

@@ -1,24 +0,0 @@
----
-layout: page
-title: About the author
-permalink: /about/
----
-{% include about.html %}
-
-<h3>Some fun stats and links</h3>
-
-<h4>I’m a registered Linux User :)</h4>
-<p>
-    <a href="https://www.linuxcounter.net/user/416972">
-        <img src="https://www.linuxcounter.net/cert/416972.png"
-             alt="I am Linux user #416972">
-    </a>
-</p>
-
-<h4>I have my own GeekCode</h4>
-<pre>
------BEGIN GEEK CODE BLOCK-----
-Version: 3.1
-GCM/CS/IT/O d--(+) s++:- a C++$ UB++L++++$ P++ L+++$ E++>$ W+++$ N o? K w+ O M- V PS+ PE Y+ PGP+(++) t+ 5 X R !tv b+ DI++ D+ G e h----() r+++ y++++
------END GEEK CODE BLOCK-----
-</pre>

blog/atom.xml

@@ -1,30 +0,0 @@
----
----
-<?xml version="1.0" encoding="utf-8"?>
-<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom">
-    <channel>
-        <title>{{site.title | xml_escape}}</title>
-        <description>{{site.description | xml_escape}}</description>
-        <link>{{site.baseurl | prepend: site.url}}</link>
-        <atom:link href="{{'/blog/atom.xml' | prepend: site.baseurl | prepend: site.url}}" rel="self" type="application/rss+xml" />
-        <lastBuildDate>{{site.posts.first.date | date: "%a, %d %b %Y %H:%M:%S %z"}}</lastBuildDate>
-        <pubDate>{{site.posts.first.date | date: "%a, %d %b %Y %H:%M:%S %z"}}</pubDate>
-        <ttl>7200</ttl>
-        <language>en</language>
-        <image>
-            <title>Gergely Polonkai</title>
-            <url>{{'/images/profile.png' | prepend: site.baseurl | prepend: site.url}}</url>
-            <link>{{'/' | prepend: site.baseurl | prepend: site.url}}</link>
-        </image>
-{% for post in site.posts limit:10 %}
-        <item>
-            <title>{{post.title | xml_escape}}</title>
-            <link>{{post.url | prepend: site.baseurl | prepend: site.url}}</link>
-            <comments>{{post.url | prepend: site.baseurl | prepend: site.url}}#comments</comments>
-            <pubDate>{{post.date | date: "%a, %d %b %Y %H:%M:%S %z"}}</pubDate>
-            <description>{{post.excerpt | xml_escape}}</description>
-            <guid isPermaLink="true">{{post.url | prepend: site.baseurl | prepend: site.url}}</guid>
-        </item>
-{% endfor %}
-    </channel>
-</rss>

blog/index.html

@@ -1,12 +0,0 @@
----
-layout: page
-title: Blog posts
-post_listing: true
----
-{% include pagination.html %}
-
-{% assign posts = paginator.posts %}
-
-{% include post-list.html %}
-
-{% include pagination.html %}

blog/tag/active-directory.md

@@ -1,4 +0,0 @@
----
-layout: posts-by-tag
-tag:    active-directory
----

blog/tag/apache.md

@@ -1,4 +0,0 @@
----
-layout: posts-by-tag
-tag:    apache
----

blog/tag/astrology.md

@@ -1,4 +0,0 @@
----
-layout: posts-by-tag
-tag:    astrology
----

blog/tag/c.md

@@ -1,4 +0,0 @@
----
-layout: posts-by-tag
-tag:    c
----

blog/tag/citrix-xenserver.md

@@ -1,4 +0,0 @@
----
-layout: posts-by-tag
-tag:    citrix-xenserver
----

blog/tag/command-line.md

@@ -1,4 +0,0 @@
----
-layout: posts-by-tag
-tag:    command-line
----

blog/tag/conference.md

@@ -1,4 +0,0 @@
----
-layout: posts-by-tag
-tag:    conference
----

blog/tag/css.md

@@ -1,4 +0,0 @@
----
-layout: posts-by-tag
-tag:    css
----

blog/tag/debian.md

@@ -1,4 +0,0 @@
----
-layout: posts-by-tag
-tag:    debian
----

blog/tag/development.md

@@ -1,4 +0,0 @@
----
-layout: posts-by-tag
-tag:    development
----