#dev | Logs for 2023-06-23
« return
[02:03:13] -!- systemd has quit [Ping timeout: 252 seconds]
[07:43:46] -!- systemd [systemd!~systemd@487-532-424-912.res.spectrum.com] has joined #dev
[07:43:46] -!- systemd has quit [Changing host]
[07:43:46] -!- systemd [systemd!~systemd@pid1] has joined #dev
[21:21:52] -!- mechanicjay [mechanicjay!~mechanicj@Soylent/Staff/Sysop/mechanicjay] has joined #dev
[21:21:52] -!- mode/#dev [+v mechanicjay] by Hera
[21:22:42] <kolie> Hey man
[21:22:44] <kolie> Uhm
[21:22:53] <kolie> So the db - we have some versioning issues.
[21:23:14] <mechanicjay> heh
[21:23:35] <kolie> I made some sql updates - you mentioned we should probably put that in a folder - one file per update - wrapped in a transaction block.
[21:23:40] <mechanicjay> kolie and I got into a 1:1 discussion about database versioning and updates and relalized it needs to me moved in here
[21:24:00] <kolie> +1 from me on that one.
[21:24:28] <kolie> Big decision right now is - 15_07 is in the git, whatever that means, live has uhh... 17_04 or something?
[21:24:49] <kolie> And they are actually the same database wise rn.
[21:25:12] <kolie> So i'm not sure what to set my current code commits to
[21:25:16] <kolie> And what to consider the "before"
[21:26:01] <kolie> But we all agree YEAR MONTH sorts well but is a shitty schema sequence.
[21:26:11] <mechanicjay> right, I'm going to come back to the idea of makeing code version and schema version their own things, an update to either doesnt' need to necessetate an update to the other
[21:43:54] <kolie> Ok. Also apparently we need a production/not production flag to do stuff like email crap to people.
[21:59:08] <kolie> I think part of the start_rehash process needs to check a flag like DISABLE_PRODUCTION_SIDEEFFECTS=1
[21:59:08] <kolie> If that's present, it sets a var in vars, once at startup.
[21:59:08] <kolie> It's checked per execution or cahced for a very short period.
[21:59:08] <kolie> That will be checked in things like update_twitter.pl not running.