Thanks to those 2 comments. Although placing Xft.dpi: 96 in my ~/.Xdefaults did not help, but I managed to do it through the appearance control center applet. I had to purge the gtk-qt theme engine because it was freezing the dialog and setting it via gconf-editor didn't work as expected.

I had to fight with my fonts this morning but here's my configuration. Maybe it'll be useful to anyone:
DejaVu sans book 12
96 DPI
greyscale smoothing, subpixel slight RGB.

I'm not yet satisfied but it's more usable now.

Thanks Phaeronix for the help too.

I have 2 laptops. My personal one at home and my work one at the office. Both are running testing. A month or two ago, I did a dist-upgrade on the home one. Later on I rebooted (DUH!) it for a reason I can't remember just to end up with very tiny fonts under GNOME. It took me some time to tweak them back again and play with the GNOME anti-aliasing dialog. Now if you know me well, you know that the last thing someone should do to me is to touch my font settings. I have a very sensitive eye (Because I have a very low vision) and I hardly tweak the fonts. All went somehow fine after that. Problem is I wasn't sure whether it was fglrx or GNOME who's responsible for that.

So, today Multi GNOME terminal has been removed from debian as part of the efforts to get rid of gtk 1.x.

I'm neither ranting nor blaming. I'm just expressing my feelings.

I knew before it happened and I acknoledged the removel. It's dead upstream and I don't think it can be easily ported. I've been using this terminal since 4 or 5 years. Let's see if I can live without it. I know there are a lot but this one was lightweight and it was using zvt which was noticeably faster than vte.
This was one of the packages I worked on with "He" as part of the NM T&S and I quite learned a lot from it.

This is a recommended update.

tarball, md5sum and gpg signed md5sum file
ChangeLog:

I was trying to package hunspell-ar for Chinook so I ended up packaging:
* hunspell (libhunspell and its development headers)
* enchant (hunspell and ispell providers but I'll disable ispell soon)
* hunspell-ar
* dictionaries-common and dictionaries-common-dev (hunspell-ar depends on them)

As for dictionaries-common and dictionaries-common-dev, they are different than the Debian ones. Yes I'm using the same source but I'm only providing the bits related to myspell. I didn't include the policy too.

him-arabic is also available as well as a few Arabic fonts:

I noticed this because of my work on him-arabic.

If your plugin provides a language other than the 16 "known" languages, It will be displayed as " ()".

Ideally, The fix should be somewhere else but not all of the components are open so even if I can fix it, I won't be able to publish patches or packages.

It seems that the source provided with the SDK is a bit outdated so I had to grab the latest trunk: https://stage.maemo.org/svn/maemo/projects/haf/trunk/hildon-input-method/

And here's a patch. It'll return the language code if it fails to get the language description. (I know it's a bit hackish :|):

Apply, build the packages and install hildon-input-method_1.99.33-2_armel.deb and libhildon-im-ui3_1.99.33-2_armel.deb

I hope it will be fixed otherwise I'll be providing packages soon. I know I promised to provide a lot of things (-:

Have a nice weekend.

This is my Arabic related TODO list for maemo. It's my personal one not the Nokia employee one :-)

* katoob should be fully functional (I'm learning here). Actually I did more work today and I guess it's fine.
* Package and publish some Arabic fonts (I know there's one provided by the browser team).
* See what can be done regarding input (Low priority. Sorry to say that I don't like the HIM plugin interface).
* Check arabeyes.org apps. Is there anything useful and needs "porting" ?
* glibc/x11 Arabic locale ?
* Any suggestions ?

I'll not translate the UI for 3 reasons:
* I'm not of a translator.
* The license you agree to while downloading the OS image prevents me from doing that.
* Because we are using logical IDs in the code, we will f*** up the UI unless all of the strings are translated.

I've just uploaded the result of the ayaspell-dic project packaged for debian as hunspell-ar (The Arabic hunspell dictionary).
The packages are now waiting in new.

The same packages can be downloaded from here. Maybe someone other than me'd like to use them ;-)

* Yes, they can upload crappy packages but we can ask to revoke an account.
* Yes, they can upload crappy packages but someone must sponsor the 1st upload.
* Yes, they can upload crap but a DD is advocating them. If they are smart enough to fool a DD to get an DM account then they are smart enough to fool him for the DD process.
* Yes, they can seek sponsors but finding sonsors is not that easy (At least for me).

Perhaps there's a way to identify the DM uploads so we can do more checks "not part of the process" ?

Just because I can't run sa-learn on my 15,000+ messages spam folder. It'll crash due to some hardware problems.
I thought that splitting the mailbox into smaller files will allow me to feed it to sa-learn.
I'm not sure something similar doesn't exist but I wrote mine anyway ;-)

The only problem is it consumes a lot of CPU and RAM, it was killed/crashed multiple times but it worked and allowed me to feed my spam mailbox to spamassassin!

Here it is in case someone needs it: split_mailbox.py. Needs python 2.5