[Adium-devl] The plan to start fixing trac

[Chris Forsythe]

Hey guys,

    Sorry for all the emails today guys, but I figured most of you would 
appreciate an update. Based on the feedback from everyone, every single 
backtrace you guys gave me had a database lock problem. I am hesitant to 
say that fixing that will fix everything else, however, moving away from 
sqlite is probably the first good move we can make, or the second.

    Andreas has also mentioned a few fixes that we'll receive once we 
move from 0.10.x to a later revision of 0.10. I think that'll help as well.

    So the game plan:

1) Sunday I will be testing from a backup of our trac setup (the adium 
one) and see how much will change when moving to postgresql, and if we 
are going to run into any problems.

2) If that goes really well, then the next thing we're going to do is 
test it with Virtue Desktops. If that goes well, then Virtue will be the 
first trac instance to get completely moved to postgre. My next free 
time is the following weekend.

3) Next goes Growl. Hopefully this can be done at the same time as Growl

4) Finally Adium. Adium is much larger than both Virtue and Growl 
combined, so testing backup strategies and what not makes more sense on 
smaller installs. We're going to let Virtue and Growl run for at least 
one week before converting Adium's trac.

5) Update trac to 0.10.latest release, barring a question to the trac 
team regarding the best version to go with is, and review of their 
mailing list for which version seems most suitable.

6) Enable logging once more and ask for any further problems to be 
reported once more.

If anyone has any experience in converting from sqlite to postgre, or 
any other tips/tricks to this whole bit then definitely let myself and 
James know. I'm cc'ing Noah since he's been haras... asked politely 
several things and always has very good feedback on what to try in some 
situations.

If updating the version before moving to postgre is a better idea, 
someone let me know.

Chris