| juergbi | robjh: in the sandbox environment there is currently only a single uid and a single gid. changing owner or group is thus impossible | 04:38 |
|---|---|---|
| juergbi | such changes are normally not permitted for unprivileged users outside the sandbox either, though, so it's typically optional in package build systems (at least those that work as non-root) | 04:41 |
| juergbi | we are planning to support a whole range of uids and gids in the linux sandbox in the future (using subuid/subgid), however, that's not implemented yet. and it will require setuid helpers (newuidmap/newgidmap) and suitable config on the host OS, which some distros provide out of the box but not all | 04:43 |
| juergbi | also see issue #38 | 04:43 |
| *** tristan has quit IRC | 05:40 | |
| *** benschubert has joined #buildstream | 07:44 | |
| *** santi has joined #buildstream | 08:29 | |
| *** jonathanmaw has joined #buildstream | 08:52 | |
| robjh | juergbi, thanks, i'll see if i can poke it to not change groups then | 09:02 |
| *** tristan has joined #buildstream | 09:10 | |
| *** ChanServ sets mode: +o tristan | 09:10 | |
| *** pointswaves has joined #buildstream | 09:21 | |
| tristan | morning benschubert, could you give https://gitlab.com/BuildStream/buildstream/-/merge_requests/1973 a lookover ? | 09:58 |
| tristan | it now has the fast-recursive/less-fast-iterative try/fallback approach, and the fast approach is slightly faster than master | 09:59 |
| benschubert | tristan: monring! Sure I'll do this | 09:59 |
| benschubert | amazing :) I'm looking forward ot review | 09:59 |
| tristan | thanks :) | 09:59 |
| * tristan relieved to get on with other things | 09:59 | |
| tristan | first, backport simpler original version of lazy resolution to 1.6 | 10:00 |
| robjh | juergbi, i got it sorted. That problem meant i didnt get my weekend project finished in the weekend. but its close now. thanks for the tip | 10:30 |
| tristan | benschubert, I think we have a race condition issue with ostree, I've been seeing these intermittent errors: https://gitlab.com/BuildStream/buildstream/-/jobs/646498214 | 10:40 |
| tristan | we might need an flock() in place after moving to the CLI approach I wonder ? | 10:41 |
| benschubert | tristan: possible, I wonder why we didn't have htat before? Because we didn't have any such checks | 10:41 |
| tristan | Not sure if test_track_recurse uses the same repo more than once | 10:42 |
| tristan | Yeah I don't know | 10:42 |
| tristan | I did see this twice today, though | 10:43 |
| benschubert | Mind opening an issue? I don't have the bandwidth right now :) | 10:43 |
| tristan | Ok | 10:44 |
| benschubert | thanks! | 10:44 |
| * tristan files https://gitlab.com/BuildStream/buildstream/-/issues/1368 | 10:49 | |
| benschubert | thanks | 10:49 |
| tristan | instead of backporting just valentind's patch to bst-1, I've copied _variables.pyx from the branch of master to _variables.py in bst-1 and un-cythonized it, and it's working well as a drop-in replacement | 11:57 |
| tristan | regardless of cythonization, should still bring a small performance boost and ensure some richer error reporting | 11:57 |
| *** tristan has quit IRC | 16:01 | |
| *** tristan has joined #buildstream | 16:02 | |
| *** ChanServ sets mode: +o tristan | 16:02 | |
| *** jonathanmaw has quit IRC | 16:52 | |
| *** santi has quit IRC | 18:36 | |
| *** benschubert has quit IRC | 19:40 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!