| *** tristan has joined #buildstream | 02:18 | |
| *** jude has quit IRC | 06:04 | |
| *** bochecha has joined #buildstream | 06:45 | |
| *** bochecha has joined #buildstream | 06:56 | |
| *** bochecha has quit IRC | 07:10 | |
| *** bochecha has joined #buildstream | 07:10 | |
| *** bochecha has quit IRC | 07:14 | |
| *** jude has joined #buildstream | 07:17 | |
| *** bochecha has joined #buildstream | 07:33 | |
| *** bochecha has quit IRC | 07:34 | |
| *** bochecha has joined #buildstream | 07:35 | |
| *** bochecha has quit IRC | 07:40 | |
| *** bochecha has joined #buildstream | 07:40 | |
| *** bochecha has quit IRC | 07:49 | |
| *** bochecha_ has joined #buildstream | 07:50 | |
| *** jonathanmaw has joined #buildstream | 08:32 | |
| *** tlater has joined #buildstream | 09:42 | |
| *** xjuan has joined #buildstream | 12:24 | |
| *** xjuan has quit IRC | 12:41 | |
| *** xjuan has joined #buildstream | 13:00 | |
| *** tristan has quit IRC | 13:22 | |
| gitlab-br-bot | buildstream: merge request (artifact-push->master: Construct pipeline without push queue if cannot push to artifact cache) #89 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/89 | 13:57 |
|---|---|---|
| gitlab-br-bot | push on buildstream@70-third-party-plugin-sharing (by Tristan Maat): 3 commits (last: _plugincontext.py: Add support for pip-installed plugins) https://gitlab.com/BuildStream/buildstream/commit/0a7d5691635857c17e27e73f144d410d314cd905 | 13:58 |
| gitlab-br-bot | push on buildstream@70-third-party-plugin-sharing (by Tristan Maat): 2 commits (last: Add support for YAML default config loading) https://gitlab.com/BuildStream/buildstream/commit/fcc671e479e8909d5795e1d47b09f30c3942b6f8 | 14:02 |
| gitlab-br-bot | push on buildstream@70-third-party-plugin-sharing (by Tristan Maat): 3 commits (last: _plugincontext.py: Add support for pip-installed plugins) https://gitlab.com/BuildStream/buildstream/commit/385d77f07c3e1d5a136a1b1d06fb01449333bca7 | 14:04 |
| gitlab-br-bot | push on buildstream@70-third-party-plugin-sharing (by Tristan Maat): 3 commits (last: _plugincontext.py: Add support for pip-installed plugins) https://gitlab.com/BuildStream/buildstream/commit/c49d7fa4552caa905397c501184d904db56bb28a | 14:06 |
| gitlab-br-bot | buildstream: merge request (artifact-push->master: Construct pipeline without push queue if cannot push to artifact cache) #89 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/89 | 14:07 |
| *** tristan has joined #buildstream | 14:27 | |
| *** ChanServ sets mode: +o tristan | 14:27 | |
| tristan | Any thoughts on https://gitlab.com/BuildStream/buildstream/issues/84 ? | 14:31 |
| tlater | tristan: Should I add a documentation page on writing third party plugins? | 14:31 |
| tristan | it's pretty bad and hoping I can resolve it swiftly | 14:31 |
| tristan | tlater, I want the documentation about actually writing plugins to be in plugin.py, element.py and source.py... but it's worth adding a page about distributing the plugins ? | 14:32 |
| tristan | then also maybe if we add other methods of sharing/discovering plugins, we can amend that part; separately from the actual plugin writing docs | 14:33 |
| tlater | tristan: So a page on how to get setuptools to actually load the module? | 14:33 |
| tristan | ummm, get setuptools to load the module ? | 14:34 |
| * tristan hopes we're not going full circle :) | 14:34 | |
| tlater | entry_points, MANIFEST.in | 14:34 |
| tlater | (well, setuptools has to *find* it, something else loads it) | 14:35 |
| tristan | a doc about how to setup a project with a setup.py which installs plugins in such a way that buildstream will find them yeah | 14:35 |
| tristan | whether it results in one being able to run a shell and 'import mypackage.myplugin' is I guess an orthogonal side effect | 14:36 |
| gitlab-br-bot | push on buildstream@master (by Tristan Van Berkom): 1 commit (last: Construct pipeline without push queue if cannot push to artifact cache) https://gitlab.com/BuildStream/buildstream/commit/f9e6fadee05c14b83fe75cf107f271dc09f60044 | 14:41 |
| gitlab-br-bot | buildstream: issue #90 ("bst build bails if it cannot push to the artifact cache") changed state ("closed") https://gitlab.com/BuildStream/buildstream/issues/90 | 14:41 |
| gitlab-br-bot | buildstream: merge request (artifact-push->master: Construct pipeline without push queue if cannot push to artifact cache) #89 changed state ("merged"): https://gitlab.com/BuildStream/buildstream/merge_requests/89 | 14:41 |
| tristan | Right so issue 84 | 14:43 |
| tristan | The problem is that local sources do not consider file attributes in their cache keys | 14:43 |
| tristan | So if you chmod +x ${project}/files/usr/bin/myscript.sh, the local source that uses that wont rebuild, and you wont get something with an executable myscript.sh (presuming you already built it non-exec) | 14:44 |
| tristan | Obviously | 14:44 |
| tristan | The tricky things is that file attributes are not exactly the same on all platforms | 14:44 |
| tristan | So what is the solution; what *should* we consider in the cache key ? | 14:45 |
| tristan | It feels tempting to say that, say standing on tlater's cross platform work, we delegate that decision to the Platform implementation in use, and encode the chosen Platform implementation into the base cache key too | 14:47 |
| tristan | But that is wrong | 14:47 |
| tristan | Because it ties the host system with what you are building, in fact; you want to be able to observe linux file attributes when you are building a linux system; if there is a problem with your native buildstream installation on windows, such that it cannot observe linux file attributes when building a linux system; that would be a bug in a hypothetical windows platform | 14:48 |
| tristan | So perhaps this is more related to the sandbox/artifact cache area of code (still in the cross platform areas... but on the target side of things) | 14:49 |
| * tristan was hoping that after saying all that an elegant solution would pop into mind allowing him to close this bug 5 or 10 minutes later | 14:49 | |
| tristan | juergbi, any idea ? | 14:49 |
| juergbi | tristan: maybe we should go with the git approach for now and only consider a very limited set of file attributes? | 15:32 |
| tristan | that seems fair yeah | 15:32 |
| tristan | good at least for 'for now' and possibly just 'good enough' | 15:33 |
| tristan | because when I think about it; even if you want to build linux on native windows; you are only ever going to be able to read the file attributes from the checked out project (i.e. the local files) from the OS host filesys | 15:33 |
| tristan | So if you really needed the attributes to be perfect (lets say that it's a matter of staging sources into a virtual machine sandbox where the target attributes can be applied), you would need to provide additional metadata | 15:34 |
| gitlab-br-bot | push on buildstream@70-third-party-plugin-sharing (by Tristan Maat): 1 commit (last: Add documentation) https://gitlab.com/BuildStream/buildstream/commit/f72668aa79bf624d3d2dfc7ace69295965ae6b3a | 15:35 |
| gitlab-br-bot | buildstream: merge request (70-third-party-plugin-sharing->master: Resolve "Third party plugin sharing") #90 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/90 | 15:36 |
| gitlab-br-bot | push on buildstream@master (by Tristan Van Berkom): 6 commits (last: utils.py: Added sha256sum utility) https://gitlab.com/BuildStream/buildstream/commit/08da7cc715c50e808c99d1cfdcef9b1cadcaa59a | 15:39 |
| tristan | apparently git only stores "executable or not" when it comes to regular files | 15:42 |
| tristan | I suppose that is still much better than what we have | 15:42 |
| tlater | tristan: Hm, I'm not sure how to test this without defining an entry point in buildstream's setup.py. | 16:25 |
| *** tristan has quit IRC | 16:26 | |
| tlater | ^ Clearly so abhorred by the idea of adding something for testing to setup.py that he had to leave. | 16:27 |
| gitlab-br-bot | push on buildstream@70-third-party-plugin-sharing (by Tristan Maat): 4 commits (last: Add support for YAML default config loading) https://gitlab.com/BuildStream/buildstream/commit/78fd7b5309e0979dbc51f0f24d765c3024db8d41 | 16:31 |
| gitlab-br-bot | buildstream: merge request (70-third-party-plugin-sharing->master: Resolve "Third party plugin sharing") #90 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/90 | 16:31 |
| *** jude has quit IRC | 16:42 | |
| *** bochecha_ has quit IRC | 16:51 | |
| *** bochecha_ has joined #buildstream | 16:51 | |
| *** bochecha has joined #buildstream | 16:55 | |
| *** bochecha_ has quit IRC | 16:56 | |
| *** tristan has joined #buildstream | 16:57 | |
| tlater | tristan: Did you see my message from earlier? | 17:00 |
| *** jonathanmaw has quit IRC | 17:01 | |
| *** bochecha has quit IRC | 17:05 | |
| *** bochecha has joined #buildstream | 17:05 | |
| *** tristan has quit IRC | 17:09 | |
| *** bochecha has quit IRC | 17:09 | |
| *** bochecha has joined #buildstream | 17:10 | |
| *** tlater has quit IRC | 17:14 | |
| *** bochecha has quit IRC | 17:39 | |
| *** bochecha has joined #buildstream | 19:02 | |
| *** bochecha has quit IRC | 19:05 | |
| *** bochecha has joined #buildstream | 19:05 | |
| *** bochecha has quit IRC | 19:09 | |
| *** bochecha has joined #buildstream | 19:10 | |
| *** bochecha has quit IRC | 19:34 | |
| *** bochecha has joined #buildstream | 19:46 | |
| *** bochecha has quit IRC | 20:03 | |
| *** xjuan has quit IRC | 20:52 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!