| *** rdale has joined #buildstream | 08:10 | |
| *** slaf has quit IRC | 08:22 | |
| *** tristan has quit IRC | 08:24 | |
| *** slaf has joined #buildstream | 08:38 | |
| *** slaf has quit IRC | 09:13 | |
| *** slaf has joined #buildstream | 09:14 | |
| *** slaf has joined #buildstream | 09:15 | |
| *** tristan has joined #buildstream | 09:43 | |
| *** ChanServ sets mode: +o tristan | 10:17 | |
| tristan | Error instantiating artifact cache: Did not find 'buildbox-casd' in PATH: /codethink/GNOME/buildstream/.tox/py37/bin:/home/tristan/.cargo/bin:/usr/local/bin:/usr/bin:/bin:/usr/local/games:/usr/games:/home/tristan/.local/bin | 10:17 |
|---|---|---|
| tristan | Hmmm | 10:17 |
| tristan | So how does one run tests with tox now | 10:17 |
| tristan | CONTRIBUTING.rst seems to still say I only need gcc and python3-dev | 10:19 |
| tristan | Prebuilt docker images containing casd ? | 10:23 |
| tristan | 275d3f62c56de8a8ae3fc8156a70c656bd28052d | 10:23 |
| tristan | juergbi, any update on the developer story here ? I guess for the time being I need a local build of buildbox-casd ? | 10:24 |
| jjardon | Nice! Good catch tristan | 10:32 |
| tristan | Ummm | 10:39 |
| tristan | And what debian package provides FindgRPC.cmake ? | 10:39 |
| tristan | Building buildbox-common, which one does when blindly trying to build buildbox-casd without any INSTALL.md file, complains that that file doesnt exist :-/ | 10:40 |
| tristan | I could download these random files on the internet https://github.com/IvanSafonov/grpc-cmake-example/tree/master/cmake | 10:43 |
| tristan | but I doubt that I am meant to manually download some dude's example project things and place them manually into some CMake module directory | 10:43 |
| tristan | sounds a bit brute force | 10:43 |
| tristan | jjardon, yeah so I was just hoping to tidy up the branch at https://gitlab.com/BuildStream/buildstream/merge_requests/1542, and backport this to bst-1 before rolling 1.4 | 10:45 |
| tristan | only problem is that master has changed in such a way that we must somehow provide buildbox-casd... | 10:46 |
| tristan | turns out what I need is gRPCConfig.cmake, libgrpc-dev seems like the right package to provide that... | 10:59 |
| tristan | https://github.com/grpc/grpc/issues/11378 seems to imply the problem was fixed 2 years ago upstream | 11:00 |
| *** tristan has quit IRC | 11:02 | |
| *** tristan has joined #buildstream | 11:02 | |
| *** ChanServ sets mode: +o tristan | 11:02 | |
| juergbi | tristan: https://buildstream.build/buildbox_install.html | 11:10 |
| juergbi | there are instructions how to install the dependencies and there is also a link to a prebuilt static binary | 11:11 |
| gitlab-br-bot | tristanvb opened issue #1115 (No instructions to use BuildStream from master) on buildstream https://gitlab.com/BuildStream/buildstream/issues/1115 | 11:22 |
| tristan | juergbi, Ah thanks ! | 11:22 |
| tristan | juergbi, I think we still want to amend the existing docs which explain how everything is gonna be fine when you just type `tox` after installing `gcc` and `python3-dev` | 11:23 |
| tristan | and link to that | 11:23 |
| juergbi | tristan: I've added a link to these buildbox instructions to the install page, with a remark only for master as we don't have different install pages for different versions | 11:24 |
| juergbi | https://buildstream.build/source_install.html#installing_dependencies | 11:25 |
| juergbi | the instructions in CONTRIBUTING.md seem to be incomplete anyway, aren't they? | 11:25 |
| juergbi | I don't think we should have separate install instructions for CONTRIBUTING.md, except possibly extra dependencies for testing (but those are pretty much all installed by tox) | 11:26 |
| tristan | juergbi, I think that before casd, tox installs everything basically so long as you have python, tox and venv | 11:30 |
| juergbi | I think we still need bubblewrap, fuse, possibly pygobject | 11:30 |
| juergbi | as they aren't in pip | 11:30 |
| juergbi | the https://buildstream.build/source_install.html#installing_dependencies lists instructions for a few distros | 11:30 |
| tristan | True bubblewrap, pygobject we worked around and comes through pip as I understand | 11:30 |
| tristan | juergbi, I don't think that the dependencies listed on the *website* however are related to experimental master | 11:31 |
| juergbi | there was a discussion before the website update was merged where we essentially concluded that we should have per-version install pages | 11:32 |
| tristan | as a separate issue, we're going to need to split that up yeah | 11:32 |
| juergbi | but I didn't want to block casd migration for this | 11:32 |
| tristan | those should probably be a part of the actual docs | 11:32 |
| juergbi | so for now I added it to the shared website | 11:32 |
| juergbi | with a remark | 11:32 |
| juergbi | I'm surprised that bubblewrap and co. are automatically installed via pip, have to give this a try | 11:34 |
| juergbi | there was also some discussion about automatically installing buildbox-casd via a wheel | 11:34 |
| tristan | juergbi, not bubblewrap; pygobject | 11:34 |
| juergbi | ah ok | 11:35 |
| tristan | there was an ugly trick for that years ago which I couldnt get work, I think it was cs-shadow who got it working | 11:35 |
| juergbi | in any case, it seems the CONTRIBUTING.md instructions are incomplete. we should fix this up but I don't think we should duplicate the instructions | 11:36 |
| juergbi | but different instructions for master and 1.x branches definitely sounds sensible | 11:36 |
| tristan | Sure, I do think that looking for an INSTALL or in the CONTRIBUTING as a developer is the first place to look | 11:40 |
| tristan | I think if we're going to have an explanation of how to run tests, build docs, etc etc... we should be linking to the right instructions from there if that's not where they go | 11:41 |
| tristan | Now just have to make this raise an error when the user explicitly sets `strict` to False... I wonder if I can give `None` as the default return to ScalarNode.get_bool() | 11:43 |
| juergbi | tristan: I don't think that works. might have to handle it with a separate check before get_bool, i.e., if 'strict' in node | 11:48 |
| tristan | Let's see | 11:51 |
| * tristan writing test case at this point anyway | 11:51 | |
| tristan | Gah | 11:51 |
| tristan | you're right | 11:51 |
| tristan | juergbi, I'm going to put the "Make explicit strict = False assignments trigger an error" part a separate patch (plus test case) at the end of the branch also for added clarity (as I split out the change of recursive element names for weak cache keys to the beginning as you suggested) | 11:57 |
| tristan | juergbi, Sounds good for you ? | 11:57 |
| juergbi | sounds good | 11:57 |
| tristan | juergbi, new branch is up fwiw | 12:11 |
| gitlab-br-bot | juergbi approved MR !1542 (tristan/strict-rebuild->master: Support strict build dependencies) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1542 | 12:22 |
| gitlab-br-bot | tristanvb opened MR !1574 (tristan/bst-1/strict-rebuild->bst-1: Support strict build dependencies (bst 1)) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1574 | 13:11 |
| gitlab-br-bot | marge-bot123 closed issue #254 (Strict rebuild needs to be a dependency semantic) on buildstream https://gitlab.com/BuildStream/buildstream/issues/254 | 13:18 |
| gitlab-br-bot | marge-bot123 merged MR !1542 (tristan/strict-rebuild->master: Support strict build dependencies) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1542 | 13:18 |
| gitlab-br-bot | LaurenceUrhegyi opened issue #1116 (Split up the CONTRIBUTING doc) on buildstream https://gitlab.com/BuildStream/buildstream/issues/1116 | 13:24 |
| *** tristan has quit IRC | 13:55 | |
| *** phoenix has joined #buildstream | 20:01 | |
| *** phoenix has quit IRC | 20:15 | |
| *** rdale has quit IRC | 21:17 | |
| *** tristan has joined #buildstream | 22:48 | |
| *** tristan has quit IRC | 23:37 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!