If you know Google’s Chromium, you probably have heard of SRWare’s Iron. That’s the name of a Chromium fork with some minor changes. Compared to Google Chrome, Iron offers much more privacy and that is just what they advertise for. However, Chrome is itself based on Chromium, their open source project. And guess what – when you build just Chromium, you will have no user metrics (basically all the spy stuff except for the following) and no crash reporting! Auto suggestion features can also be turned off, so when you do that, it treats your privacy like any other browser like Firefox does these days, in other words: Your privacy is safe as it can be from only the browser’s side.

So if we take SRWare’s Iron vs. Chrome and think of it as Iron vs. Chromium, there’s not much left at all. In fact it is just the Other important differences section. Let’s have a closer look…

The current version of Google Chrome uses Version 530.5 of the Rendering-Engine WebKit and Version 1.1.10 of the V8 Javascript-Engine. WIron [sic!] uses the most up-to-date Version (532.0) of Webkit and V8 (1.2.14) which is significantly faster.

This statement is out of date. Chromium 4.0.222.5 and upwards uses WebKit 532.2, which is actually a newer version than the one announced on the Iron website. I could not really find out which version of V8 the most recent chromium revision uses, but the fact that Chromium uses a newer WebKit version should make obvious which browser is out of date here.

Chrome doesn’t have an [sic!] built-in adblocker. Iron has an easy to use and built-in adblocker which can be configured by a single file.

I don’t know how good that ad-blocker works. It is true that Chromium does not have an integrated ad-blocking system (and seem not to like to have one anyway), so this might be the only real upside of the project (on a side note, I think easy to use is what I would call AdBlock Plus for Firefox, because you can select elements to block from within the browser, not something that is just based on a config file).

The User-Agent in Chrome is only be changeable [sic!] with parameters over a link or command, which isn’t really ideal for permanent usage. The User-Agent in Iron can be flexiblel [sic!] and permanently changed by UA.ini.

By that, they mean that chromium supports a different user agent by sending a command line parameter. While it might be helpful for some people, it is not a big benefit to be able to change that option in a config file.

To sum it up, the only reason why one would prefer Iron over Chromium would be the ad-blocker.

However, I still have a major issue with the development of that browser. Everyone knows Sourceforge, Google Code and similar websites, right? These allow you to host open source applications for free and provide very good code handling, for both developers and users, with a version control system such as SVN or GIT. What the hell do you have to snort up your nose to think it’s a good idea to upload the huge code of a chromium fork on Rapidshare? What the fuck? By doing that, you make distribution harder (no one wants to see a timer before downloading OSS, automatic package building does not work, …) and you can’t modify parts of the code without re-uploading the whole archive! It is plain stupid!

Also at the same time they provide pre-built binaries on their own server. Why don’t you put them on Rapidshare, too? Don’t they usually make a whole lot more traffic than the source?

I could even go as far and argue that SRWare does not want you to see the sourcecode, because they have hidden something bad inside. Now that would be a logical reason to do such a horrible thing!

In my opinion, this is cancer to the OSS world and should therefore not be supported. Iron developers, register a Google Code account (would be fitting since Chromium is also a Google project) and make Iron more transparent if you really care. Best would be to provide just a patchset for the Chromium source code and pre-built binaries for Windows Users. That would be the most transparent way I can think of – and it would be certainly the best way for the community!

tl;dr: Do not use Iron until they switch to an acceptable source code location, Rapidshare or any other one-click-hoster are definitely not an option! If you need proper ad-blocking in Chromium, set up Privoxy, it should fit your needs at least as well as Iron.

In some cases you would like to package software for a Linux distribution which seems not quite to want that itself („just copy everything into your home folder“). I’ve already written an article about how to package JDownloader, which is a program that requires write access to all its files, as it has its own updater.

This time I will write about programs that require access to a folder inside its application directory. I know that there is more software of that kind (although it should not be in the unix world, properly programmed software should only modify data in the home folder) like dolphin-emu, but I will focus on gtkradiant-svn in this case, because I maintain it.

GTKradiant consists of two application files (one for compiling maps, one for the actual editor), which both live in the same folder, that also contains some sub-directories with other data, such as bitmap files for the editor. I’ve put all that stuff in /usr/share/gtkradiant, as it should be (NOT in /opt or anywhere else!) and that’s just where the problem begins. GTKradiant will crash if it does not have access to the games folder (relative to the application’s main binary file).

What to do now? Some tutorials suggest to just install GTKradiant in your home folder. But that is not an option, since I am talking about a package here, which is expected to install the software system-wide.

One option would be to create a group called games (if it does not exist yet) at install-time and give members of that group write access to the games folder. This would require the user to add his normal account to the games group, which is not so dramatic. One downside is however, that when one user modifies files inside the games directory, it is also modified for all the other user’s installations and vice versa. I think that this solution is good for huge games/other applications that need access to too many files (not talking about folders now), that it would waste too much space and/or traffic to put a copy of them in every user’s home folder and update each of them separately.

However, with GTKradiant we only have that tiny folder, which would be better off in every user’s home directory. My first solution was to link /usr/share/gtkradiant/games to ~/.radiant/1.6.0/games (~/.radiant/1.6.0/ is used by the application anyway) on package building time. But this did not work as expected – the way I wrote it, bash replaced ~ with the real user name, so this workaround only worked for the user who built the package. This is not the way I intended it at all. I’ve experimented around a bit and it turned out that it is not possible to link with ~ as variable, so that it replaces that letter with the username, depending on who’s asking.

Somehow sad that this solution did not work the way I wanted I thought further and came up with this:

How about linking all files from the application’s root folder (in this case /usr/share/gtkradiant) to some temporary folder each time we run the software, plus linking a games folder inside that temporary folder to somewhere in home?

And I am glad that this idea, with symbolically linking all the folders and hardlinking the application itself (because otherwise radiant „thinks“ that it’s root path is /usr/share/gtkradiant and therefore wants to access /usr/share/gtkradiant/games instead of /tmp/gtkradiant-username/games – the symbolic link to somewhere in the current user’s home) and creating a temp folder that contains the current user’s name, works out exactly as I want it to.

Here’s my launch script, it can be customized easily:

#!/bin/bash
#
# A wrapper for applications that need write access
# to a directory that is write protected. We just link
# it to somewhere in home and it will work.
# - thoughtcrime
#
#
# Configuration
#
# Path to the program
path_program="/usr/share/gtkradiant"
#
# Where will we put the files?
# (Somewhere in home would make sense)
path_target=~/.radiant/1.6.0
#
# Make sure that the temp path is unique
# for each user!
path_temp="/tmp/gtkradiant-`id -un`"
#
# $path_link are the folders where write access
# is required. Both variables are relative to the
# program folder.
path_link=("games")
path_app="radiant.bin"
#
# This is required if the application would find out
# its correctly installed path otherwise
hardlink_app="true"
#
# Configuration end
#

# Install the temporary folder
install -d "$path_temp"

# Link everything from the program in the temporary folder
for file_target in ${path_program}/*; do
 if [[ "$hardlink_app" = "true" && "$path_temp/${file_target##*/}" = "$path_temp/$path_app" ]]; then
 if [ ! -e "$path_temp/$path_app" ]; then
 ln "$file_target"  "$path_temp/$path_app"
 fi
 elif [ ! -L "$path_temp/${file_target##*/}" ]; then
 ln -s "$file_target" "$path_temp/${file_target##*/}"
 fi
done

# Link all folders that need write access
for file_link in ${path_link[@]}; do
 if [ ! -d  "$path_target/$file_link" ]; then
 install -d "$path_target/$file_link"
 fi
 if [ ! -L "$path_temp/$file_link" ]; then
 ln -s "$path_target/$file_link" "$path_temp/$file_link"
 fi
done

# Start the application
cd "$path_temp"
./"$path_app" $*

Please use this for software that requires write access to one or more of the folders inside its root directory – or (depending on the application!) – either the group- (see above) or the full-app-in-home-solution (see the JDownloader article for the last one). Because reading messages like…

==> It will likely be necessary for you to change the ownership of /opt/dolphin-emu/
==>  before anything useful (or fun) can be done with this program.
==>  It may be sufficient to just own the User/ subdirectory.
==> Ex: chown -R youruser:users /opt/dolphin-emu/User

…is just not necessary.

You have probably heard of Google Chrome, right? That newish browser Google wants you to use. Actually Google has installed some spy crap in that browser, so I recommend not to use it. However, big G are nice enough to provide the browser on an open source basis – without some of the privacy relevant features (and the other ones can be switched off in its preferences). This is what they call chromium.

As I am writing this, Archlinux users may install chromium from the testing repository. If you do not want to enable testing, just pick your favorite mirror and install the package directly from there. Example for x86_64:

pacman -U http://ftp-stud.hs-esslingen.de/pub/Mirrors/archlinux/testing/os/x86_64/chromium-4.0.249.30-1-x86_64.pkg.tar.gz

Chromium with an alternative skin

Chromium comes with a very minimal interface, it does not even have a status bar, as status information is displayed above the normal website content. The scrollbar does not have buttons to scroll anymore (but hey, no one uses them anyway), it is just the scrollbar itself. I guess that all this comes from netbook optimization.

Compared to Firefox, Chromium starts up lightning fast. You basically click the icon on the desktop (or type it into a terminal, whatever) and it is there. Firefox always takes some seconds to start (I guess it spends those looking for add-on updates). The browsing does also feel fast (but not faster, it is basically the same).

The developer tools shipped with Chromium are pretty equal to Firebug for Firefox. It is also possible to change the browser’s background, just like it is built in with Firefox 3.6 (beta) and above (I use 3.6b4 right now), although the Chromium website for changing skins does not feature as much designs by far as the Firefox page does.

Speed dial in Chromium

Chromium also copied the speed dial feature from Opera (you see a bunch of screenshots for your favorite websites when you open a new tab) which I think looks pretty nice.

Under the hood there’s also some code that allows every tab from chromium to be its own process. This is a great step in the right direction, because if one tab crashes, the other ones are not affected. Also this provides better support for multicore processors.

A huge reason not to use Chromium however, is that it can not have something like NoScript by design. Adblocking can be done with Privoxy as a workaround, but it is not quite as usable as a real add-on, because you can’t say right from the browser: block this!

So as you see I won’t use Chromium as my main browser – at least until it supports NoScript. It has some great ideas though and I hope other browsers – and especially Firefox – take advantage from that.

Realurl.org, that is a website that tries to get attention by doing trackbacks on blogs when someone tweets about them. In my blog’s case, there are two Twitter accounts tweeting about every new post, because I want it that way. This means that when I check the spamfilter, as I did today, I get a lot of these annoying trackbacks.

Anyway, I got a bit curious about the URLs (the same mistake can be found in the title and the website itself by the way). They look like:

http://realurl.org/twitted.php?id=randomnumber

Why the hell twitted? The owners of realurl spam probably every blog ever mentioned on Twitter and they do not know that the verb for writing a status message on Twitter is to tweet? I know I do not speak or write perfect English by far, but even I notice that and it looks really stupid to me. What were they thinking?

Squid is a proxy that has great cache features. I use it to speed up some web applications by caching everything except for files with the text/html type. Why? Because they get changed most of the time. With the Web 2.0 there are even websites which change every time you access them (for example the powered by line on omploader), so caching that does not make too much sense. Pictures on the other hand do not get changed that often – and even if they do, its hardly relevant. Besides, good admins know about cache and change the picture’s URLs, too.

In order to make squid not cache ‘text/html’ stuff, add this to /etc/squid/squid.conf (or wherever your configuration file is):

# Do not cache HTML pages
acl nocache_html rep_mime_type text/html
no_cache allow nocache_html

PS: That syntax highlighting feature is great, here’s how to use it on a wp.com blog.

qooxdoo is a rich internet application framework, which basically means that it allows you to create javascript based web applications that look pretty similar to native applications by providing common GUI elements which you can not get by using just HTML and recreates some others in order to give the GUI a consistent design. The framework part means that in order to create your own qooxdoo application, you need to install some software on the operating system you are developing with.

This software are actually a bunch of python scripts that merge many javascripts together, strip them down and do some other useful tasks. That means however, that you need to build your project, before you can release it.

I haven’t done much with that framework yet – except for creating an AUR package – but I guess I will code something with it in the future

If this stuff sounds interesting to you, check out the demo browser.

PS: Yes, this article will probably appear twice on twitter and/or your feed reader. Why? Because using wordpress.com with effective cache settings seems to be a bad idea. I guess the other article I wrote this day just replaced this one. Luckily I had a tab open and so I just copy pasted the whole thing.

Speaking of AUR packages, I have spent a lot of time on trying to make a working lightning-hg package. In case you do not know what hg is, it is the name of the binary that one can use to access a mercurial repository (something like git, svn, cvs etc). Basically it should provide the most bleeding edge version on the interwebs, which works with thunderbird 3.

Anyway, the mozilla build process is very complicated as there is one repository for many mozilla apps and some of their add-ons. I made a PKGBUILD which downloads the source properly, but I couldn’t get the compile part right yet (and I’m too bored to try harder right now).

So if you think you are a skilled AUR contributor and/or would also like to see a lightning-hg package, write something in my archbbs request thread. Thanks

…I finally found a lyrics display program, that fits my needs. As always, I am looking for some specific features I want definitely to be implemented. In this case, I required realtime scrolling with the text (so it displays just the line that gets spoken on the song right now), support for Amarok 1.4 and the ability to act as OSD (on screen display). Last but not least I wanted to get the lyrics downloaded automatically from the intertubes. Well, after spending too much time trying out all the lyrics programs I could find – twice – I decided to stick with osd-lyrics. It has a few bugs though, but it basically does what I want it to

osd-lyrics in action

Posted in English Tagged: amarok, lyrics, music, osd, osd-lyrics ]]> http://thoughtyblog.wordpress.com/2009/11/22/after-a-long-search/feed/ 0 thoughtyblog osd-lyrics http://thoughtyblog.wordpress.com/2009/11/15/my-first-wall-texture/ http://thoughtyblog.wordpress.com/2009/11/15/my-first-wall-texture/#comments Sun, 15 Nov 2009 11:31:21 +0000 thoughtyblog http://thoughtyblog.wordpress.com/?p=1210 ]]>

I decided to make a wall texture and see how it turns out, but the only think I really like about it are the colors. Compared to a wall texture I used from another Urban Terror map just for testing, this one has much bigger bricks (but that isn’t too bad as it is theoretically possible to scale it within the level editor).

But the thing that really bugs me is the blurry middle part… in case you don’t know too much about such textures (as I did yesterday until reading some tutorials), you take a square of a (free-for-use) photo and then set an offset for the image in GIMP (or what you like to use) to x/2 and y/2 (if your imagination isn’t that good, here are some example pictures). The goal is to make a transition between all the edges (which are now in the middle) and in my bad example, it looks like the that part is painted or something :-\

I’m glad I did a wall anyway, because now I know exactly how it works. But for good looking textures, I’ll probably use some from the default UrT maps and change the color in order to get what I want.

…that I can use to write an article in my blog. That software I develop, ghostship, runs very stable in my experience. I think *someone* should package it for other Linux distributions. I’ve got already someone working on a Debian port, but I am not quite sure if and when he will actually do it – maybe I will do it myself in a few weeks.

Twitter became a handy tool for me, now I am able to post things like „the new southpark episode is [my opinion] [link]“ in just a few seconds rather than expanding this information to a whole blog article and finally posting it on wordpress (like I used to do previously). Another cool thing I noticed is that this information does get indexed on my blog by Google however, because the most recent (and who reads old ones anyway) tweets are shown in the navigation. That way you don’t even need to „follow“ me or add my twitter account to your feed reader

I’ve experimented around a bit with GTKRadiant (a map editor) in combination with Urban Terror as I wrote earlier. By now I am trying to remake cs_1337_assault from Counter-Strike 1.6, as this is one of my favorite maps with some outdated graphics though. On the other hand, I don’t like the Source version, because it is too different from the original cs_assault – I’ll try to mix it up with my own ideas. But by now I don’t have much yet except for the one building which is directly in front of the CT spawn, crappy lighting and some lame textures which I just use from the default Urban Terror maps. Well, if I have enough time and the map turns out to look great and be very playable without changing the original cs_assault feeling too much, I’ll definitely post it on this blog.