<quote who="Erik de Castro Lopo"> > People, > > The following offer is open till 8am tomorrow: > > I hereby offer to try and find/fix any bug in any piece of software that > mets the following conditions: > > 0) Software is released under any of the standard Open Source or Free > Software licenses. > 1) Is available as an official Debian package. > 2) The version of the package in Debian unstable has the bug. > 3) The software runs in user space (ie not kernel code). > 4) Is written in C, C++ or Python. > 5) Has a easy recipe for recreating the problem. > 6) You notify me in advance of the Debian package name (by private > email please). > 7) You can be present to confirm that the bug is fixed. > 8) Time permits :-). > > Hecklers, rubbernecks, supporters, people who think they could do better, > people wanting to help etc are all welcome. I will attempt to give a running > commentary the debugging progresses. > > All bugs found/fixed will be submitted as bug reports with a patch to > Debian's bug tracking system. > > So far I have two to work on: > > http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=193966 > nominated by Ken Caldwell > > http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=149088 > nominated by Peter Chubb > > Unfortunately, Ken can't be there until sometime in the afternoon and the > second one looks real difficult. I would therefore like to an easier one > to warm up on, especially since I am hoping to pass on some of my debugging > techniques. Congratulations Erik!
Two bugs submitted, two bugs resolved. I didn't arrive till early afternoon so Erik started on the "hard" bug first. By the time I looked in the master had disposed of the first bug and was ready to start on the second. This was a bug in the window manager aewm which caused it to (frequently) ignore mouse clicks on the root window. Some left clicks, most middle clicks and no right clicks were ignored. This bug turned out to be much harder to crack than the first. In fact it took over six hours. (from about 2pm to 9pm with a short refueling stop for a late lunch) The program was recompiled many times with debugging messages strategically placed to try and isolate the section of code which was not behaving as expected. Progress was quite satisfactory at first and it was readily shown that the program was detecting all the mouse button events. The difficult part was determining why the subprogram was often not run following the detection of a button event. A chance discovery that the bug could be consistently triggered by being very slow on the button led to the solution of the problem. I would like to publicly thank Erik for his marathon effort. There must have been a dozen other people hacking away all day on various projects but I have not seen any reports either here or on slug chat. What did anyone else achieve? (I know it may be hard to write about programming but not impossible ;-) cheers, Ken -- SLUG - Sydney Linux User's Group - http://slug.org.au/ More Info: http://lists.slug.org.au/listinfo/slug