| *** tristan has quit IRC | 06:59 | |
| *** tristan has joined #buildstream | 07:09 | |
| gitlab-br-bot | push on buildstream@sam/compose-log-splits (by Tristan Van Berkom): 3 commits (last: Refactor: Move context.py -> _context.py) https://gitlab.com/BuildStream/buildstream/commit/213d9072b684d2dff78d8b4f1c7cfa9d6335b0d0 | 08:30 |
|---|---|---|
| gitlab-br-bot | buildstream: merge request (sam/compose-log-splits->master: Log details of artifact splitting when building 'compose' elements) #140 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/140 | 08:30 |
| gitlab-br-bot | push on buildstream@master (by Tristan Van Berkom): 1 commit (last: git.py source plugin: Prune remote-tracking branches when fetching) https://gitlab.com/BuildStream/buildstream/commit/794e73cb4469c30bd830b333597a6932da1254f1 | 09:14 |
| gitlab-br-bot | buildstream: merge request (sam/git-fetch-prune->master: git.py source plugin: Prune remote-tracking branches when fetching) #146 changed state ("merged"): https://gitlab.com/BuildStream/buildstream/merge_requests/146 | 09:14 |
| gitlab-br-bot | buildstream: Sam Thursfield deleted branch sam/git-fetch-prune | 09:14 |
| *** givascu has joined #buildstream | 09:31 | |
| *** valentind has joined #buildstream | 09:42 | |
| *** jonathanmaw has joined #buildstream | 09:48 | |
| *** ssam2 has joined #buildstream | 10:13 | |
| gitlab-br-bot | buildstream: merge request (97-apply-pep-3102-to-all-public-api-surfaces->master: WIP: Resolve "Apply pep 3102 to all public API surfaces") #127 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/127 | 10:17 |
| gitlab-br-bot | push on buildstream@97-apply-pep-3102-to-all-public-api-surfaces (by Jonathan Maw): 20 commits (last: Refactoring: Move exceptions module to be private) https://gitlab.com/BuildStream/buildstream/commit/27aac095bdb6b0025c989572789a1f98d8d89b3c | 10:17 |
| tristan | valentind, I'm going to step out early, I think the compose patch is great, and I've also blocked https://gitlab.com/BuildStream/buildstream/merge_requests/140 on it (because exposing the API you need is an all around better solution than exposing the other API proposed on 140) | 10:27 |
| tristan | But, I would *really* want to change that test for an integration test | 10:28 |
| tristan | tlater, I think I'm going to kick off early today | 10:29 |
| tristan | if you have thoughts on my comments on the regression fix... I think now is better | 10:29 |
| tlater | tristan: Alright, I'll need a few minutes, didn't notice them | 10:31 |
| tlater | tristan: Why can we assume that an element is recursing if it has been visited in the same scope before? | 10:36 |
| valentind | tristan, I am fine with your comments. | 10:37 |
| tlater | Ah, nevermind, I understand now. The last commend on not returning something that has been visited actually makes this work. | 11:01 |
| *** givascu has quit IRC | 11:03 | |
| tristan | tlater, great, yeah I *think* my proposed solution to this is going to work well | 11:03 |
| tristan | tlater, it is admittedly, a very tricky function; it's primary use case is stable walking of the tree; staging order | 11:04 |
| tristan | I dont think you'll achieve a combined staging order walk with your approach, but I dont think that is the goal either (build plan is where that counts most, and in any case we have the planner which does that) | 11:05 |
| tristan | i.e.: bst show will only show correct staging order when only one element is specified; but that is quite alright I think | 11:06 |
| * tristan prepares himself to press "Send" on the d-d-l announcement ! | 11:10 | |
| tlater | dotted duckbill lovers? | 11:12 |
| ssam2 | https://mail.gnome.org/mailman/listinfo/desktop-devel-list | 11:15 |
| tristan | https://mail.gnome.org/archives/desktop-devel-list/2017-November/msg00032.html | 11:22 |
| tristan | :D | 11:22 |
| tlater | \o/ | 11:23 |
| *** valentind has quit IRC | 11:29 | |
| *** tristan has quit IRC | 11:37 | |
| gitlab-br-bot | push on buildstream@97-apply-pep-3102-to-all-public-api-surfaces (by Jonathan Maw): 1 commit (last: element.py: Change node_subst_member's 'default_value' field to 'default') https://gitlab.com/BuildStream/buildstream/commit/a776d94f1c9d2478e29f9dbb811389049bf6500e | 11:42 |
| gitlab-br-bot | buildstream: merge request (97-apply-pep-3102-to-all-public-api-surfaces->master: WIP: Resolve "Apply pep 3102 to all public API surfaces") #127 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/127 | 11:42 |
| gitlab-br-bot | buildstream: merge request (git-element-origin->master: WIP: Change origin to point at source repo) #143 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/143 | 11:53 |
| gitlab-br-bot | buildstream: merge request (git-element-origin->master: Change origin to point at source repo) #143 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/143 | 11:53 |
| *** adds68_ has joined #buildstream | 12:55 | |
| *** adds68 has quit IRC | 12:56 | |
| ssam2 | seems the `local` element doesn't consider file permissions part of the cache key | 13:04 |
| ssam2 | that must be a bug, if the only way to set permissions is by setting the equivalent permissions on the source files | 13:04 |
| tlater | ssam2: https://gitlab.com/BuildStream/buildstream/issues/84 | 13:14 |
| ssam2 | ah, thanks | 13:15 |
| tlater | iirc the discussion ended with 'We'll support execute permissions at some point' | 13:16 |
| ssam2 | i'm still not convinced that Git remembers any permissions other than +x | 13:16 |
| ssam2 | i guess this needs to be researched fully | 13:16 |
| ssam2 | bah, i really thought i was going to get a bootable VM today | 13:17 |
| tlater | :( | 13:17 |
| ssam2 | but it fails to boot because the /init script isn't +x | 13:17 |
| ssam2 | (obviously i could work around this, but that's not the point :-) | 13:17 |
| ssam2 | this could be quite tricky to do in fact | 13:21 |
| ssam2 | i am more or less against Tristan's idea to rely on the filesystem for permissions in the first place | 13:25 |
| ssam2 | as not all filesystems preserve modes correctly | 13:25 |
| ssam2 | so putting project files on a FAT-formatted USB stick would break everything | 13:26 |
| ssam2 | and yes, Git only stores either 0755 or 0644 | 13:31 |
| tlater | ssam2: It's hard to know what permissions mean exactly then. Linux permissions just don't make sense everywhere. | 13:31 |
| ssam2 | I think we can assume BuildStream is always producing UNIX-like systems | 13:32 |
| ssam2 | that doesn't limit us from producing filesystems that are less expressive, e.g. FAT | 13:32 |
| ssam2 | it would limit us from producing a type of system that has a richer permissions model stored at the filesystem level, but i don't know of one that would matter | 13:33 |
| ssam2 | xattrs are an interesting one of course; and are another argument in favour of not relying on the filesystem to provide such info | 13:33 |
| *** WSalmon has joined #buildstream | 13:34 | |
| tlater | ssam2: On the other hand, would it be enough to just store 0755/0644 like git does? | 13:35 |
| ssam2 | for a first cut, I think that's OK | 13:35 |
| ssam2 | it does mean that if you want to set permissions like 0600 then you need to do it manually with `chmod` as a build or integration command | 13:36 |
| ssam2 | but that can be second issue | 13:36 |
| ssam2 | *a separate issue | 13:36 |
| * ssam2 finally boots into his VM | 13:53 | |
| ssam2 | which is 17MB in size! that's good going, at least | 13:53 |
| ssam2 | except the disk image is 1.4GB, which seems a bit wrong | 13:54 |
| ssam2 | swap-size defaults to 1G, I see | 13:55 |
| * ssam2 forsees some patches to the x86image element | 13:57 | |
| gitlab-br-bot | push on buildstream@97-apply-pep-3102-to-all-public-api-surfaces (by Jonathan Maw): 8 commits (last: git.py source plugin: Prune remote-tracking branches when fetching) https://gitlab.com/BuildStream/buildstream/commit/794e73cb4469c30bd830b333597a6932da1254f1 | 14:19 |
| gitlab-br-bot | buildstream: merge request (97-apply-pep-3102-to-all-public-api-surfaces->master: WIP: Resolve "Apply pep 3102 to all public API surfaces") #127 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/127 | 14:19 |
| gitlab-br-bot | buildstream: merge request (incremental-build->master: Add support for doing incremental builds) #126 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/126 | 14:25 |
| * tlater seems to be too stupid to implement this, unless tristan overlooked that we have the special use case of not recursing the full tree. | 14:39 | |
| gitlab-br-bot | push on buildstream@except_intersections (by Tristan Maat): 2 commits (last: load.py: Migrate to new test style) https://gitlab.com/BuildStream/buildstream/commit/f8f3db8661c06f5f63142664d4e4bfa3a5047303 | 14:53 |
| gitlab-br-bot | push on buildstream@except_intersections (by Tristan Maat): 8 commits (last: load.py: Migrate to new test style) https://gitlab.com/BuildStream/buildstream/commit/d9a876eb2565497b463f0df876a6d48bf18e95e6 | 15:13 |
| gitlab-br-bot | push on buildstream@except_intersections (by Tristan Maat): 5 commits (last: main.py: Fix app initialization) https://gitlab.com/BuildStream/buildstream/commit/633b25b500cbb51e93d48cf7e272ae036a0528c1 | 15:14 |
| *** anahuelamo has quit IRC | 15:21 | |
| *** anahuelamo has joined #buildstream | 15:22 | |
| *** tristan has joined #buildstream | 15:45 | |
| *** jonathanmaw has quit IRC | 15:45 | |
| paulsherwood | how do changes to https://wiki.gnome.org/Newcomers/BuildSystemComponentBst happen, please? | 15:47 |
| ssam2 | the gnome wiki has this rather annoying anti-spam mechanism in place | 15:47 |
| ssam2 | once you have an account, you need to ask someone to add your username to https://wiki.gnome.org/TrustedEditorGroup so that you can edit | 15:47 |
| * paulsherwood has been triggered by the tristan apostrophe | 15:47 | |
| ssam2 | creating an account is easy, and then anyone who already has write access to the wiki can edit that page for you | 15:48 |
| paulsherwood | perhaps tristrophe should be a word :) | 15:48 |
| tristan | what what, people getting picky about how it's written again, with all it's tristophracies ? | 15:50 |
| paulsherwood | quite | 15:50 |
| * tristan updates, but reflects that the world will eventually have to accept that it's just not a problem :) | 15:51 | |
| paulsherwood | tristan: would you mind adding paulsherwood to the list of folks who can edit that page? | 15:52 |
| *** anahuelamo has quit IRC | 15:53 | |
| * tristan doesnt think he has the rights | 15:53 | |
| paulsherwood | also, is there a good reason that https://gnome7.codethink.co.uk/gnome-modulesets.git is at codethink, rather than gitlab? | 15:53 |
| tristan | its not a page local thing, I think jjardon[m] gave me access originally | 15:54 |
| tristan | as I hadnt modified the wiki in years | 15:54 |
| ssam2 | i have added you | 15:55 |
| ssam2 | anyone with edit permissions for the wiki can edit that page | 15:56 |
| *** tristan has quit IRC | 15:57 | |
| jjardon[m] | I've suggested to use https://git.gnome.org//browse/gnome-modulesets instead | 16:00 |
| paulsherwood | ssam2: tvm | 16:01 |
| *** tristan has joined #buildstream | 16:01 | |
| tristan | oops, /me seems to have had an internet hiccup, not sure that the relevant part about modulesets on gnome7 got through | 16:02 |
| paulsherwood | it did not | 16:02 |
| tristan | <tristan> paulsherwood, yes there is a good reason; it cannot be an editable repository, and it's generated as a part of a process of migrating jhbuild modulesets | 16:03 |
| tristan | <tristan> which happens on gnome7 | 16:03 |
| tristan | <tristan> so basically, until we hit "switch", it will remain something that is generated from modifying jhbuild modulesets | 16:03 |
| tristan | :) | 16:03 |
| tristan | so basically how it works is we keep modifying the 3.28 modulesets at https://git.gnome.org/browse/jhbuild/tree/modulesets | 16:04 |
| tristan | and every 10 minutes, https://gnome7.codethink.co.uk/gnome-modulesets.git is updated | 16:04 |
| tristan | there is a blog post linked in the announcement email which describes the process in detail | 16:04 |
| tristan | that will continue to happen so that the transition is not aggressive, until at least end of 2017, but we will create a permanent repo in advance of 3.28 stable release | 16:05 |
| paulsherwood | ok... maybe update the wiki page to say this is special modulesets temporarily at codethink? (just seems a bit odd to see a codethink service in gnome instructions) | 16:06 |
| tristan | Yeah, the wiki page is also a work in progress but understood | 16:06 |
| tristan | I'll add some notes about that first thing tomorrow | 16:07 |
| tristan | maybe a small note with a link to that blog post will be good | 16:08 |
| * ssam2 wonders why modifying a .yaml file in the bst-external repo seems to have no effect on my build ... | 16:15 | |
| ssam2 | i have done `pip3 install --user .` in the bst-external repo, so would expect this to just work | 16:15 |
| ssam2 | oh | 16:15 |
| ssam2 | i meant to add `-e` :-) | 16:15 |
| * tristan seems to recall there was an issue with `pip install --user -e` | 16:16 | |
| ssam2 | yeah... it still doesn't work | 16:16 |
| tristan | but maybe it's fixed, you'd have to check with jonathan and tlater | 16:16 |
| tlater | Hi | 16:16 |
| tristan | tlater, using plugins - installed with `pip3 install --user -e .` | 16:17 |
| ssam2 | argh, turns out one of my changes that was merged is totally broken | 16:17 |
| ssam2 | and i never noticed because of this other issue | 16:17 |
| * tlater is fairly sure that is supposed to have been fixed | 16:17 | |
| ssam2 | it appears not | 16:18 |
| tlater | I'll check the issue for that again, can't quite remember what that was about | 16:18 |
| tlater | ssam2: That's a new issue, the old one was a stacktrace | 16:20 |
| ssam2 | ah right | 16:20 |
| tlater | For reference: https://gitlab.com/BuildStream/buildstream/issues/110 | 16:21 |
| tlater | Though I doubt it's related at all | 16:21 |
| ssam2 | an integration test for the x86image plugin would be pretty useful, i guess it has so many dependencies though that its tricky | 16:22 |
| ssam2 | although we are going to start using it in Baserock really soon, which will at least catch future breakages | 16:22 |
| ssam2 | presumably whatever the problem is with plugin loading, it's in that same code that !107 touched | 16:23 |
| * tlater hopes that python doesn't garbage collect loaded files now | 16:24 | |
| gitlab-br-bot | push on buildstream@migrate_pipeline_load (by Tristan Maat): 1 commit (last: Remove DummyElement hack) https://gitlab.com/BuildStream/buildstream/commit/b35bd94bd2b01853b0096dea6e3bf5404d79b1e9 | 16:49 |
| gitlab-br-bot | buildstream: merge request (migrate_pipeline_load->master: Migrate `tests/pipeline/load.py`) #145 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/145 | 16:49 |
| gitlab-br-bot | push on buildstream@except_intersections (by Tristan Maat): 6 commits (last: Remove DummyElement hack) https://gitlab.com/BuildStream/buildstream/commit/b35bd94bd2b01853b0096dea6e3bf5404d79b1e9 | 17:11 |
| *** tpollard has quit IRC | 17:59 | |
| gitlab-br-bot | push on buildstream@tracking-changes (by Tristan Maat): 20 commits (last: Refactoring: Move exceptions module to be private) https://gitlab.com/BuildStream/buildstream/commit/27aac095bdb6b0025c989572789a1f98d8d89b3c | 18:01 |
| gitlab-br-bot | buildstream: merge request (tracking-changes->master: WIP: Tracking changes) #119 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/119 | 18:01 |
| *** ssam2 has quit IRC | 19:05 | |
| *** adds68__ has joined #buildstream | 19:27 | |
| *** adds68_ has quit IRC | 19:28 | |
| *** tlater has quit IRC | 20:17 | |
| *** benbrown has quit IRC | 20:17 | |
| *** laurenceurhegyi has quit IRC | 20:17 | |
| *** paulsherwood has quit IRC | 20:17 | |
| *** laurenceurhegyi has joined #buildstream | 20:19 | |
| *** tlater has joined #buildstream | 20:20 | |
| *** benbrown has joined #buildstream | 20:22 | |
| *** paulsherwood has joined #buildstream | 20:22 | |
| *** valentind has joined #buildstream | 21:16 | |
| valentind | tristan, mocks are common practice in testing. | 21:20 |
| valentind | For contributors it is nice to have lots of simple tests to make sure we do not break anything before submitting a PR. | 21:21 |
| valentind | You do what you want of course. But see the point of view of contributors who do not have a full overview of the project. I would have not felt comfortable submitting if I did not write those tests. | 21:22 |
| tristan | valentind, you happen to catch me at 6 am before sleeping - yes they are common, the integration test is better, it runs automatically on gitlab, and it's less stuff to maintain | 21:22 |
| valentind | Sorr | 21:22 |
| valentind | I just wanted to explain my point of view. | 21:22 |
| valentind | All I wanted to say. | 21:22 |
| valentind | Have a good sleep. | 21:23 |
| hergertme | is there a wiki page i can look at with how to get started using the buildstream stuff for building modules? | 21:23 |
| hergertme | i should start playing with it soon so i can get it integrated into Builder at some point | 21:23 |
| hergertme | (as a side note, > .75 cycles notification from my side would have been appreciated, being that now i have to change my development timeline for 3.28 to compensate and doing buildstream work) | 21:24 |
| tristan | hergertme, right now there is only https://wiki.gnome.org/Newcomers/BuildSystemComponentBst and http://buildstream.gitlab.io/buildstream/ | 21:25 |
| hergertme | basically, this leaves our contributors using Builder out to hang unless i can somehow find a way to fit this into 3.28, which is already going to be tight. | 21:26 |
| * tristan is burning at 200% and approaching nervous breakdown territory | 21:26 | |
| hergertme | im sympathetic, ive been running on that since 2014. maybe choose a less aggressive timeline? | 21:27 |
| tristan | this is the wrong time for me to be on irc :-/ | 21:27 |
| tristan | hergertme, this is something we've planned since GUADEC, I'm sorry we didnt announce before it was really, really ready to test | 21:28 |
| tristan | If we push back, that means flatpak and booting images push back 6 more months, or crazy shit happens with autoconverting modulesets | 21:28 |
| tristan | this timeline fwiw, also does not effect flatpacks | 21:29 |
| hergertme | people use Builder for more than just flatpaks though, we have full jhbuild integration too | 21:29 |
| tristan | I cant even *start* make that stuff happens, until we cut off jhbuild. | 21:29 |
| hergertme | we all knew "build stream was coming", but most of us plan our cycles ... you know, before the cycle starts | 21:29 |
| hergertme | so by doing this (we're already 1 month into development, and 2 months before freezes start) | 21:30 |
| tristan | Ok, I wont discuss this at 6:30 am, sorry I appeared at all | 21:30 |
| tristan | this is exactly the wrong kind of discussion for me to have in the middle of the night | 21:30 |
| tristan | sorry | 21:30 |
| hergertme | i can email, its fine | 21:30 |
| *** tristan has quit IRC | 21:34 | |
| *** valentind has quit IRC | 23:36 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!