Sunday, 27 May 2007

Two weeks later...

It has been two weeks from my last post there... due to this end of semester being a bit thougher than the last ones in my university I couldn't do everything I wanted, particularly on the Summer of Code front. Summer of Code I believe this blog should be on SoC's planet by now. Hello, planet :) I hope the SoC can be a great time for all of the people involved. Today, I finally got some time to make code to talk with debbugs (Debian's bug tracking system software). Thanks to the help of Don Armstrong on debian-debbugs, I have been able to get a python function to get informations from a given bug report through SOAP, but unfortunately it doesn't handle non-ascii characters. I've also made a similar function parsing the debbugs-generated HTML pages; but parsing html doesn't sound reliably at all. BTW, the code to parse HTML was mostly taken from Bastian Venthur's reportbug-ng; thanks Bastian. Free Software rocks because you don't have to rewrite the wheel. Debian Packages On the Debian Packaging front, libgda 3.0.1-1 was uploaded recently (thanks, Loïc Minier). Also, my patch to use libgcrypt instead of openssl was commited on libgnomedb upstream's repository (this time, the thanks goes to Vivien Malerba); I hope a new upstream release is made soon.

Saturday, 12 May 2007

Bug Triage and Forward: The Beginning

I've got some free time over the last two days, so I've started to write some code for my Summer of Code project, the Bug Triage and Forward Tool (bugtaf). As I had only a laptop without internet connection in hand, it wasn't possible to implement anything on the tracker integration part, so I've started the user interface. This is the first python application I'm writing, and so far it seems to be a great language. The biggest difficulty I've had so far with it was to find how to call the parent class constructor. I got stuck there yesterday when making QueryBuilder (derived from gtk.Dialog), but today (with internet), I found the answear on Dive Into Python. It's pretty straightforward BTW, I guess I need to sleep more. The pyGTK stuff is pretty simple; there, the background with C GTK programming helps a lot. The only quirk I had there is that calling set_border_width on a already packaged GtkBox (eg, on an Dialog's vbox) doesn't work :( After all that, I've git pushed it to alioth, and git pulled on my desktop. Just then I remembered I had created an ui branch, as I wasn't sure on how my first try with python/pyGTK would be and wished to test the git branching. It took me some fight with git, but everything seems to be properly merged in the master branch now.

Sunday, 6 May 2007

Playing with alioth and git

I've been playing a bit with alioth's git support. I didn't find the full procedure to create a new git repository on alioth anywhere (of course, I didn't search a lot though), so I'll post how I did it here. First of all, I asked for a git repository on the Site Admin Support Requests Tracker. They created a directory with the right permissions in (bug-triage is the name of my alioth project). After that, I loged on through ssh (the ssh keys registered on alioth work fine) and initialized a new empty git repository there: $ mkdir /git/bug-triage/bug-triage.git; cd /git/bug-triage/bug-triage.git $ git --bare init-db --shared Then, I logged out of and created a git repository on my machine, with an empty README file, and tested pushing it to, which worked fine: $ mkdir ~/bug-triage; cd ~/bug-triage; touch README $ git init $ git add README $ git commit -m "Add an empty README" $ git push --all ssh:// Finally, I've set an origin remote entry on my machine's repository, so that I can use push, fetch and pull without typing the path: $ git remote add origin ssh:// To test that, I've edited the README file and "commited" the change to