#archlinux32 | Logs for 2023-07-14
Back
[01:54:43] -!- bill-auger has quit [Remote host closed the connection]
[01:56:50] -!- bill-auger has joined #archlinux32
[02:06:14] -!- bill-auger has quit [Ping timeout: 246 seconds]
[02:34:38] -!- bill-auger has joined #archlinux32
[05:15:22] <bill-auger> does anyone know the schema of the statefiles? pacman 6.0.2-7 6.0.2-7 a9e5c309e2c4347a402651c4d14ee883f1af6932
[05:16:10] <bill-auger> why does it have the same version repeated? - which shuold i be interested in? is one or the other possibly different than the released package version?
[09:34:22] <Foxboron> it's not repeated
[09:36:24] <Foxboron> bill-auger: the format is: $pkgbase $pkgver $gitlabTag $checksumOfPkgbase
[09:45:50] <bill-auger> awesome thanks
[15:14:25] -!- bill-auger has quit [Ping timeout: 250 seconds]
[17:29:06] -!- lodger has joined #archlinux32
[17:29:11] -!- lodger has quit [Remote host closed the connection]
[17:45:58] -!- sunshavi has joined #archlinux32
[20:30:48] <girls> KitsuWhooa: yeah, that's the price, one pays for allowing to run too many db queries in parallel: lots of timeouts
[20:31:21] <girls> we could increase the number of retrials (it's currently at 10)
[20:31:58] <KitsuWhooa> ah, I see
[20:32:11] <KitsuWhooa> worth a try I think
[20:33:47] <girls> ok, there are now 15 retries. Additionally, the delay between the retries increases linearly, now
[20:33:49] <KitsuWhooa> it's interesting how fragile this entire thing is
[20:34:06] <girls> the db queries are quite complex
[20:34:35] <girls> I tried to make them perfect from a logical standpoint - which I had to buy with many joins, and thus a huge execution time :-D
[20:34:47] <KitsuWhooa> how does upstream arch do this?
[20:34:56] <girls> they don't
[20:35:07] <KitsuWhooa> hmmm
[20:35:18] <girls> they do everything manually
[20:35:31] <girls> i.e. someone says "move package xyz from extra-testing to extra"
[20:35:39] <KitsuWhooa> I see
[20:36:31] <girls> we have checks, that this will not break any package in extra (because it depends on something that is only provided by the old version of xyz, or because xyz itself depends on something, that is still in extra-testing)
[20:37:03] <girls> same goes for building: upstream builds manually - i.e. someone says bump pkgrel of xyz, rebuild it, upload
[20:37:32] <girls> we try to track all this and if a library gets updated, the buildmaster should know, that the dependent packages will break, thus need to be recompiled
[20:37:58] <girls> but as you have seen, most probably, this does not always work (mainly, because the reschedule job is so slow, that we don't run it that often ...(
[20:37:59] <girls> ))
[20:40:32] <KitsuWhooa> yeah, I see
[20:40:36] <KitsuWhooa> thanks for the explanation
[21:36:32] -!- bill-auger has joined #archlinux32