| *** dylan-m has joined #buildstream | 00:14 | |
| *** dylan-m has quit IRC | 00:35 | |
| *** dylan-m has joined #buildstream | 01:19 | |
| *** dylan-m has quit IRC | 01:36 | |
| *** dylan-m has joined #buildstream | 02:20 | |
| *** dylan-m has quit IRC | 03:13 | |
| *** traveltissues has joined #buildstream | 08:44 | |
| gitlab-br-bot | coldtom opened MR !1811 (coldtom/fix-overnight-test->master: .gitlab-ci.yml: Bump bst-plugins-experimental) on buildstream https://gitlab.com/BuildStream/buildstream/-/merge_requests/1811 | 09:15 |
|---|---|---|
| *** benschubert has quit IRC | 09:20 | |
| *** benschubert has joined #buildstream | 09:20 | |
| *** jonathanmaw has joined #buildstream | 09:36 | |
| gitlab-br-bot | BenjaminSchubert approved MR !1811 (coldtom/fix-overnight-test->master: .gitlab-ci.yml: Bump bst-plugins-experimental) on buildstream https://gitlab.com/BuildStream/buildstream/-/merge_requests/1811 | 09:41 |
| *** tme5 has joined #buildstream | 09:44 | |
| tme5 | is it possible for !1808 to be merged today? pretty sure it's passed review | 09:45 |
| gitlab-br-bot | MR !1808: Improvements to _GitSourceBase and _GitMirror https://gitlab.com/BuildStream/buildstream/-/merge_requests/1808 | 09:45 |
| benschubert | tme5: sure, can I go ahead and merge it now? | 09:48 |
| tme5 | yeah, it's ready | 09:50 |
| gitlab-br-bot | BenjaminSchubert merged MR !1808 (tmewett/git-source->master: Improvements to _GitSourceBase and _GitMirror) on buildstream https://gitlab.com/BuildStream/buildstream/-/merge_requests/1808 | 09:50 |
| benschubert | thanks :) | 09:50 |
| *** santi has joined #buildstream | 09:54 | |
| *** slaf has quit IRC | 09:58 | |
| *** slaf has joined #buildstream | 09:59 | |
| *** slaf has joined #buildstream | 10:00 | |
| tme5 | np, thanks for merge | 10:06 |
| robjh | ironfoot, you were right. the '/' at the beginning of the path was the issue. | 10:22 |
| robjh | there's a new problem now though, it looks like overlap-whitelist follows symlinks /o\ | 10:23 |
| ironfoot | oh, ouch | 10:24 |
| robjh | not the whitelist itsself, but the thing that detects changes in overlapped files, so the symlink is being flagged as an overlap when the destination is what's actually overlapping | 10:26 |
| robjh | i should probably raise a bug report | 10:26 |
| juergbi | robjh: please do, we should definitely not be following symlinks. if you've seen this issue on bst 1.x, it's possible it's already fixed in master, though | 10:29 |
| juergbi | (we fixed a lot of symlink issues a while ago but I don't remember whether those were applied before branching / to bst 1.x) | 10:30 |
| robjh | my bst is from the debian repo, so its probably ancient | 10:31 |
| *** santi has quit IRC | 10:33 | |
| *** santi has joined #buildstream | 10:33 | |
| juergbi | ok, so it might be fixed already | 10:33 |
| coldtom | hi, i'm trying to figure out exactly how proxying through buildbox-casd to push artifacts works, but I can't for the life of me see how buildbox-casd is informed of the remote url, does anyone have any pointers on where this happens? | 10:36 |
| coldtom | s/artifacts/blobs from artifacts/ | 10:37 |
| juergbi | coldtom: GetInstanceNameForRemoteRequest() is used to get an identifier (instance name) for the URL and that identifier is then used in actual blob-related requests | 10:47 |
| coldtom | ah, thanks juergbi, i think i was missing the fact that the instance name is sufficient to push/pull things | 10:51 |
| *** santi has quit IRC | 10:51 | |
| *** santi has joined #buildstream | 10:52 | |
| juergbi | it may be a bit confusing but this approach allows us to work with existing CAS protocol methods that we couldn't extend | 10:52 |
| *** phildawson-ct has quit IRC | 11:12 | |
| *** narispo has joined #buildstream | 11:16 | |
| *** santi has quit IRC | 11:52 | |
| *** santi has joined #buildstream | 11:52 | |
| coldtom | so, i've written a patch to allow bst-artifact-server to optionally proxy artifact/reference storage to a remote CAS, however while implementing said feature i realised that this might actually be better as an external tool altogether. does this sound like something people would be interested in? | 13:15 |
| coldtom | the reasoning for writing this being mainly "i want to store my artifact protos with some other backend" and "i don't want to add more storage for artifact protos when i have a perfectly good volume attached to my CAS" | 13:17 |
| juergbi | coldtom: how exactly do you store the artifact protos? standard CAS alone does not suffice | 13:32 |
| juergbi | as discussed yesterday, the plan is to switch to the Remote Asset API | 13:33 |
| juergbi | not sure yet how bst-artifact-server should be changed at that point | 13:35 |
| *** xjuan has joined #buildstream | 13:37 | |
| *** tpollard has quit IRC | 13:42 | |
| *** tpollard has joined #buildstream | 13:46 | |
| *** phildawson-ct has joined #buildstream | 13:48 | |
| *** phildawson-ct has quit IRC | 14:00 | |
| coldtom | juergbi: i'm just serialising them and using bytestream to push them, is this approach naïve? i agree that the remote asset api is the best way to go in the future, which is part of the reason why i think an external tool for this would be more useful | 14:00 |
| juergbi | coldtom: ah, you have a CAS server that accepts bytestreams with arbitrary names instead of strictly just CAS objects? | 14:01 |
| juergbi | (i.e., not content addressed) | 14:01 |
| coldtom | ah, i hadn't thought of that, i might need to do some more digging | 14:02 |
| *** phildawson-ct has joined #buildstream | 14:02 | |
| juergbi | with strict CAS, it's not possible to store artifact protos/refs | 14:02 |
| juergbi | (you can store the artifact protos, of course, but you couldn't get them back without knowing the hash) | 14:03 |
| juergbi | buildgrid and buildbox casd don't accept non-CAS objects, afaik | 14:03 |
| juergbi | don't know about buildbarn/buildgrid but I wouldn't expect those to accept non-CAS objects either | 14:03 |
| coldtom | i _think_ bb-storage (which is what i'm intending to use) does handle arbitrary bytestreams with some caveats, so it should be able to store them safely | 14:12 |
| coldtom | but other CAS implementations not accepting this confirms to me that this should be an external tool for sure | 14:12 |
| coldtom | ack, i was wrong, i think i'll just have to accept that bst-artifact-server is unavoidable :/ | 14:48 |
| *** phildawson-ct has quit IRC | 14:57 | |
| *** rdale has quit IRC | 15:03 | |
| *** rdale has joined #buildstream | 15:03 | |
| *** phildawson has joined #buildstream | 15:05 | |
| *** tme5 has quit IRC | 15:29 | |
| *** tme5 has joined #buildstream | 15:29 | |
| *** rdale has quit IRC | 15:42 | |
| *** rdale has joined #buildstream | 15:45 | |
| *** xjuan has quit IRC | 16:12 | |
| *** phildawson-ct has joined #buildstream | 16:21 | |
| *** phildawson has quit IRC | 16:22 | |
| *** traveltissues has quit IRC | 16:48 | |
| *** xjuan has joined #buildstream | 17:07 | |
| *** dylan-m has joined #buildstream | 17:23 | |
| *** jonathanmaw has quit IRC | 17:29 | |
| *** dylan-m_ has joined #buildstream | 17:29 | |
| *** dylan-m has quit IRC | 17:30 | |
| *** dylan-m_ is now known as dylan-m | 17:30 | |
| *** narispo has quit IRC | 17:33 | |
| *** tme5 has quit IRC | 17:38 | |
| *** santi has quit IRC | 18:10 | |
| *** phildawson-ct has quit IRC | 18:11 | |
| *** phildawson-ct has joined #buildstream | 18:11 | |
| *** dylan-m has quit IRC | 19:07 | |
| *** dylan-m has joined #buildstream | 19:17 | |
| *** dylan-m has quit IRC | 19:29 | |
| *** phildawson-ct has quit IRC | 19:40 | |
| *** toscalix has joined #buildstream | 20:48 | |
| *** dylan-m has joined #buildstream | 21:32 | |
| *** xjuan has quit IRC | 23:11 | |
| *** xjuan has joined #buildstream | 23:25 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!