| *** jsgrant has joined #buildstream | 01:47 | |
| *** jsgrant has quit IRC | 02:34 | |
| *** jsgrant has joined #buildstream | 02:56 | |
| *** jsgrant has joined #buildstream | 02:57 | |
| *** jsgrant has quit IRC | 03:21 | |
| *** tristan has joined #buildstream | 05:17 | |
| *** tristan has quit IRC | 05:41 | |
| *** tristan has joined #buildstream | 05:55 | |
| *** jsgrant has joined #buildstream | 06:16 | |
| *** jsgrant has quit IRC | 06:53 | |
| *** tristan has quit IRC | 06:53 | |
| *** mohan43u has quit IRC | 07:06 | |
| *** mohan43u has joined #buildstream | 07:09 | |
| *** tristan has joined #buildstream | 07:17 | |
| *** noisecell has joined #buildstream | 07:23 | |
| *** toscalix has joined #buildstream | 07:57 | |
| *** bethw has joined #buildstream | 07:58 | |
| *** jonathanmaw has joined #buildstream | 08:02 | |
| *** tiago has joined #buildstream | 09:05 | |
| *** mohan43u has quit IRC | 09:05 | |
| *** tiago has quit IRC | 09:06 | |
| jennis | tristan, hi, renaming these rst files, do you think it's a good idea to have a directory structure in doc/source/ that reflects the ToC? | 09:08 |
|---|---|---|
| *** mohan43u has joined #buildstream | 09:09 | |
| jennis | So for example, all that we have in doc/source/ is index.rst and index/ | 09:09 |
| jennis | Or do you think it's better to keep all of the rst files in doc/source/ and give them long names reflecting where they belong | 09:10 |
| *** tiago has joined #buildstream | 09:16 | |
| *** dominic has joined #buildstream | 09:24 | |
| *** toscalix has quit IRC | 09:54 | |
| *** valentind has quit IRC | 10:07 | |
| *** bethw has quit IRC | 11:04 | |
| *** toscalix has joined #buildstream | 11:29 | |
| *** aday has joined #buildstream | 11:35 | |
| *** toscalix has quit IRC | 11:41 | |
| *** tristan has quit IRC | 11:43 | |
| *** tristan has joined #buildstream | 11:46 | |
| *** bethw has joined #buildstream | 11:47 | |
| tristan | jennis, I dont really mind either way, consistency would be nice to have; such that any section has a "word" which is used to prefix subsections of that section | 11:48 |
| tristan | directories, or words with underscores, doesnt make much difference... maybe directories will be nicer | 11:48 |
| tristan | considering addition of examples | 11:49 |
| jennis | yeah and maybe directories will encourage devs to think about structure more | 11:50 |
| jennis | My only concern about this directories approach: It's nice to simply have index.rst and index/ in the doc/source folder. However, if you're hacking and execute `make -C doc` you'll get all the buildstream.* files and a couple of extra directories | 11:52 |
| *** toscalix has joined #buildstream | 11:53 | |
| tristan | jennis, I dont understand what it means "you'll get all the buildstream.* files and a couple of extra directories" | 12:01 |
| tristan | jennis, if you build the docs, the docs will be built, into doc/build/html or smth, you will not "get" ... I dont know what the issue is | 12:03 |
| *** toscalix has quit IRC | 12:41 | |
| jmac | If I'm making an artifact cache that serves over HTTP, should it be necessary to create a GPG key? | 12:42 |
| jmac | My server is being repeatedly asked for /artifacts/summary.sig which doesn't exist atm | 12:43 |
| tlater | jmac: Pretty sure it should be, if it wasn't you'd have a wide-open security hole | 12:45 |
| tristan | jmac, ostree will probably ask for it indeed, but we currently dont do anything for that | 12:45 |
| tristan | tlater, no, we dont use that feature yet, it's known | 12:45 |
| tristan | tlater, that's why we currently only recommend serving over https | 12:46 |
| tlater | Ah, fair enough | 12:46 |
| tristan | we are looking into artifact signing, but will need to do so in an ostree agnostic way | 12:46 |
| tristan | (we *do* support GPG verification in the ostree source plugin, though) | 12:46 |
| tristan | It's plausible that we need to file an issue and fix upstream to avoid requesting the summary.sig in the case that you are pulling from an ostree repo which is not configured to do GPG ostree-built-in validation | 12:47 |
| tristan | if that is any significant overhead | 12:48 |
| jmac | Right, I'll stick with https then | 12:48 |
| jennis | tristan, this is what I meant: http://termbin.com/xefi | 13:08 |
| tristan | jennis, myeah... so ? | 13:10 |
| tristan | jennis, if you want something to inform your choice, an idea might be: "What is the easier thing to specify in .gitignore ?" | 13:10 |
| jennis | ok thanks | 13:11 |
| tristan | jennis, I dont really have a specific preference, that's why I'm reaching for a practical reason :) | 13:14 |
| *** xjuan has joined #buildstream | 13:24 | |
| *** sstriker has joined #buildstream | 13:53 | |
| *** tristan has quit IRC | 13:53 | |
| *** sstriker has quit IRC | 13:54 | |
| *** sstriker has joined #buildstream | 13:54 | |
| *** tristan has joined #buildstream | 14:02 | |
| *** ernestask has joined #buildstream | 14:02 | |
| *** dplayle has joined #buildstream | 14:03 | |
| *** dplayle has quit IRC | 14:04 | |
| *** toscalix has joined #buildstream | 14:06 | |
| Nexus | tristan: could i get a quick update from you about the workspace changes you've been making over the weekend pls? | 14:51 |
| Nexus | as i just rebased to find that the section of code i've been working on has been moved :) | 14:52 |
| tristan | Nexus, in a bit of a meeting here... it has not changed drastically; it has moved from Pipeline -> App mostly | 14:52 |
| tristan | Nexus, it was a lot of bloat for Pipeline, and some workspace operations do not require an instantiated Pipeline, hence the change | 14:53 |
| * tlater hates how gitlab caches pipeline results on MR pages | 15:08 | |
| *** Prince781 has joined #buildstream | 15:11 | |
| *** Prince781 has quit IRC | 15:36 | |
| *** toscalix has quit IRC | 15:45 | |
| tristan | Nexus, regarding changes related to workspaces, asides from moving the higher level code to a new venue, there is only one change in the API contract | 15:55 |
| tristan | Which is Workspaces() object, in _workspaces.py, now takes an element name, instead of an element, for looking up and deleting a workspace | 15:55 |
| tristan | This is necessary to work with workspaces without instantiated elements | 15:56 |
| *** toscalix has joined #buildstream | 15:56 | |
| tlater | Ah, that does indeed sound nicer | 15:56 |
| tristan | ironfoot, any chance you could slap the gitlab bot ? I think he's been sleeping :) | 15:56 |
| ironfoot | tough easter :/ | 15:56 |
| tristan | tlater, yeah, it was a part of fixing that issue where you cant close a workspace if you're project doesnt have the element anymore | 15:57 |
| *** Prince781 has joined #buildstream | 15:57 | |
| Nexus | tristan: i'm mostly having trouble finding and replacing all of the points where calls were changed. E.G. raise pipelineError() -> raise appError() | 15:58 |
| Nexus | and alike | 15:58 |
| tristan | actually the weekend refactors went really nicely, the two branches I landed, A.) improved things... B.) deleted roughly ~70 or ~80 LOC, and C.) Added a lot of comments (meaning, a lot more LLOC were actually removed) | 15:58 |
| tlater | Nexus: Shouldn't tristan's commit tell you that? | 15:58 |
| Nexus | tlater: i have to go through line by line and compare them, because it was all done in a single commit | 15:59 |
| Nexus | as far as git knows, the entire thing is new code, not just the changes | 15:59 |
| tlater | Ah, too bad, are you sure git's interactive rebase doesn't help? | 16:00 |
| tlater | Usually it's really good figuring this stuff out | 16:00 |
| tristan | 19e31adb00f19f0789a3c4622150ca8f6e9c8901 is not immensely huge, but was done this way to retain the policy that every commit should build and pass tests | 16:00 |
| tristan | it is separated from earlier supporting commits which change how initialization of the application works | 16:01 |
| tristan | I suspect interactive rebase is not very helpful, I moved main.py -> app.py in an earlier commit conform with regular naming policy | 16:01 |
| tristan | now that that file actually just implements App() | 16:01 |
| tristan | history for git blame is retained, but I suspect it will fool git rebase; probably looking at the diffs you had before and applying them by hand to the new HEAD is safer | 16:02 |
| tlater | Hm, I'm getting random failures where files in about-to-be cached artifacts disappear before I can `stat` them | 16:04 |
| tlater | This CI has an example of that: https://gitlab.com/BuildStream/buildstream/-/jobs/60887678 | 16:04 |
| *** gitlab-br-bot has joined #buildstream | 16:06 | |
| juergbi | tristan: what's the reason for not using double underscores for local private symbols in private classes? | 16:06 |
| *** Prince781 has quit IRC | 16:06 | |
| * juergbi thinks this would be more consistent | 16:07 | |
| juergbi | and Python would enforce that it's not accessed from outside the same file | 16:07 |
| tristan | juergbi, my feeling was that those are generally the outlier, python developers dont generally expect them, nor do they recommend their use | 16:07 |
| tristan | principal of least surprise I guess | 16:07 |
| jjardon[m] | Hi, is there a plugin/extension somewhere to generate rootfs? something similar to the tar extension in baserock? https://gitlab.com/baserock/definitions/blob/morphs/clusters/build-system-x86_64-chroot-deploy.morph | 16:08 |
| *** dominic has quit IRC | 16:08 | |
| tristan | juergbi, they are not exactly pretty either, I think we're better off only using them where needed | 16:08 |
| tristan | juergbi, note that while I have documented it as "a means of disambiguation", the underlying reason is actually different | 16:09 |
| juergbi | tristan: ah, if that's what's considered best practice in Python, then it's fine with me | 16:09 |
| tristan | Mostly, for public classes which can be derived in plugins/extensions, we want to free up namespace, so there is a technical reason | 16:09 |
| tristan | However, the way I have documented it, happens to satisfy that need at the same time as disambiguating | 16:10 |
| * tristan is waiting for pipeline to complete to send the email, so that I can link to the page in the docs | 16:10 | |
| *** valentind has joined #buildstream | 16:11 | |
| tristan | the HACKING.rst is now literally included into the docs, thanks to jjardon[m] :) | 16:11 |
| juergbi | right, one leading underscore for non-public methods and two leading underscores to avoid name clashes with subclasses matches PEP8 | 16:12 |
| tristan | So, it means Sandbox also gets that treatment, even though it is not publicly derived, it is still nice to disambiguate A.) What is hidden from public API from B.) What will really only be used in this file | 16:14 |
| tristan | I've handled the most hectic of files so far, which are my fault (i.e. making Project and Context private at the last moment before 1.0 stable) | 16:15 |
| tristan | and Pipeline(), which was just getting really messy | 16:15 |
| tristan | And also improved doc comments for methods all around | 16:15 |
| tristan | a lot of house cleaning | 16:16 |
| tristan | mail sent | 16:20 |
| gitlab-br-bot | buildstream: merge request (135-expire-artifacts-in-local-cache->master: WIP: Resolve "Expire artifacts in local cache") #347 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/347 | 16:29 |
| tristan | jjardon[m], not per se, there is an issue report lying around somewhere to implement a tar streaming output from `bst checkout`, though. | 16:31 |
| *** noisecell has quit IRC | 16:31 | |
| gitlab-br-bot | buildstream: merge request (135-expire-artifacts-in-local-cache->master: WIP: Resolve "Expire artifacts in local cache") #347 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/347 | 16:32 |
| jjardon[m] | tristan: what would be the correct thing to do if I want to simply generate a rootfs? write and element that do a tarball of everything or a plugin to deploy? I'm trying to understand what would be the equivalent of baserock clusters in buildstream | 16:36 |
| gitlab-br-bot | buildstream: merge request (135-expire-artifacts-in-local-cache->master: WIP: Resolve "Expire artifacts in local cache") #347 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/347 | 16:43 |
| gitlab-br-bot | buildstream: merge request (135-expire-artifacts-in-local-cache->master: WIP: Resolve "Expire artifacts in local cache") #347 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/347 | 16:52 |
| gitlab-br-bot | buildstream: merge request (135-expire-artifacts-in-local-cache->master: WIP: Resolve "Expire artifacts in local cache") #347 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/347 | 16:52 |
| gitlab-br-bot | buildstream: merge request (135-expire-artifacts-in-local-cache->master: WIP: Resolve "Expire artifacts in local cache") #347 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/347 | 16:52 |
| gitlab-br-bot | buildstream: merge request (135-expire-artifacts-in-local-cache->master: WIP: Resolve "Expire artifacts in local cache") #347 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/347 | 16:52 |
| gitlab-br-bot | buildstream: merge request (135-expire-artifacts-in-local-cache->master: WIP: Resolve "Expire artifacts in local cache") #347 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/347 | 16:53 |
| gitlab-br-bot | buildstream: merge request (jmac/artifact_cache_error_message->master: Artifact cache: Mention the remote URL when we fail to fetch remote refs) #354 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/354 | 16:57 |
| *** dominic has joined #buildstream | 16:57 | |
| *** gitlab-br-bot has quit IRC | 16:59 | |
| *** gitlab-br-bot has joined #buildstream | 16:59 | |
| *** brlogger has joined #buildstream | 17:03 | |
| *** gitlab-br-bot has joined #buildstream | 17:03 | |
| *** bethw has quit IRC | 17:03 | |
| tristan | jjardon[m], sorry it's late, just obtained food from the last closing restaurant (damn I wanted chicken ! but unavailable) | 17:05 |
| tristan | jjardon[m], it would be better IMO to implement that checkout feature, and avoid an extra step of generating a tarball inside BuildStream, to be honest | 17:06 |
| tristan | but that could be a workaround for not having the feature | 17:06 |
| tristan | the problem with using `bst checkout` directly and tarring the result, is you have the filesystem and user permissions to worry about (files in the tarball will not be owned by root, etc) | 17:07 |
| *** jonathanmaw has quit IRC | 17:07 | |
| tristan | if we have a proper `bst checkout --tar` or such, then we can generate the tar with exactly the right file permissions and attributes | 17:07 |
| jjardon[m] | rigth | 17:08 |
| *** jsgrant has joined #buildstream | 17:11 | |
| *** solid_black has joined #buildstream | 17:27 | |
| *** aday has quit IRC | 17:27 | |
| *** jsgrant has quit IRC | 17:27 | |
| *** aday has joined #buildstream | 17:28 | |
| *** aday has quit IRC | 17:32 | |
| *** jsgrant has joined #buildstream | 17:32 | |
| *** aday has joined #buildstream | 17:33 | |
| *** solid_black has quit IRC | 17:34 | |
| *** dominic has quit IRC | 18:01 | |
| *** jsgrant has quit IRC | 18:01 | |
| *** aday has quit IRC | 18:12 | |
| *** toscalix has quit IRC | 18:22 | |
| *** slaf- has joined #buildstream | 18:28 | |
| *** slaf has quit IRC | 18:30 | |
| *** slaf- is now known as slaf | 18:30 | |
| *** xjuan has quit IRC | 20:01 | |
| *** Prince781 has joined #buildstream | 21:02 | |
| *** ernestask has quit IRC | 21:06 | |
| *** valentind has quit IRC | 21:40 | |
| *** Prince781 has quit IRC | 22:09 | |
| *** sstriker has quit IRC | 22:26 | |
| gitlab-br-bot | buildstream: issue #336 ("Error when running bst --track in big projects: "BlockingIOError: [Errno 11] Resource temporarily unavailable: Exception ignored when trying to write to the signal wakeup fd:"") changed state ("opened") https://gitlab.com/BuildStream/buildstream/issues/336 | 23:25 |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!