| *** skullman has quit IRC | 00:59 | |
| *** skullman has joined #buildstream | 00:59 | |
| *** tristan has joined #buildstream | 04:14 | |
| *** catonano has joined #buildstream | 04:15 | |
| *** catonano has quit IRC | 04:18 | |
| *** catonano has joined #buildstream | 04:19 | |
| *** klaas has joined #buildstream | 05:01 | |
| *** paulsherwood has quit IRC | 06:13 | |
| *** laurence has quit IRC | 06:14 | |
| *** mablanch has quit IRC | 06:14 | |
| *** aiden has quit IRC | 06:14 | |
| *** valentind has quit IRC | 06:14 | |
| *** laurence has joined #buildstream | 06:15 | |
| *** mablanch has joined #buildstream | 06:16 | |
| *** valentind has joined #buildstream | 06:16 | |
| *** aiden has joined #buildstream | 06:16 | |
| *** paulsherwood has joined #buildstream | 06:19 | |
| *** adds68 has quit IRC | 06:19 | |
| *** coldtom has quit IRC | 06:19 | |
| *** coldtom has joined #buildstream | 06:19 | |
| *** adds68 has joined #buildstream | 06:19 | |
| *** tristan has quit IRC | 06:25 | |
| *** catonano has quit IRC | 06:45 | |
| *** mablanch has quit IRC | 06:49 | |
| *** mablanch has joined #buildstream | 06:50 | |
| *** tristan has joined #buildstream | 06:53 | |
| *** lor4x has joined #buildstream | 08:16 | |
| *** rdale has quit IRC | 08:32 | |
| *** rdale has joined #buildstream | 08:35 | |
| *** toscalix has joined #buildstream | 09:17 | |
| *** jonathanmaw has joined #buildstream | 09:43 | |
| *** tristan has quit IRC | 09:45 | |
| *** CTtpollard has joined #buildstream | 09:58 | |
| *** tpollard has quit IRC | 09:59 | |
| *** raoul has joined #buildstream | 10:04 | |
| *** alatiera_ has joined #buildstream | 10:06 | |
| *** lachlan has joined #buildstream | 10:29 | |
| *** jonathanmaw_ has joined #buildstream | 10:45 | |
| *** jonathanmaw has quit IRC | 10:45 | |
| laurence | Does the buildstream project have a YouTube channel? | 10:52 |
|---|---|---|
| laurence | toscalix, you're good with these things ? | 10:52 |
| laurence | I have edited the videos from the gathering, want to post them somewhere and link to them from the website / wiki / a ML post | 10:52 |
| toscalix | I would open a youtube channel or similar once we have several videos. How many do we have now? | 10:54 |
| *** CTtpollard is now known as tpollard | 10:55 | |
| laurence | There are three from the gathering | 10:55 |
| laurence | BuildStream 101, BuildStream 102, Architecture | 10:55 |
| laurence | There is also Vimeo | 10:56 |
| tpollard | somebody unsurprisingly has the username buildstream on youtube, but I'd still advocate youtube | 10:56 |
| laurence | OK, will make one called buildstream-tool or similar....will check what the twitter handle decision was...if any | 11:08 |
| toscalix | laurence: I am not familiar enough with any of them. I haven't admin any. I have added videos to both and they are reliable. License wise though, we might prefer a channel more CC friendly | 11:08 |
| toscalix | since we do not expect a huge audience | 11:08 |
| adds68 | laurence, what about under the gnome youtube channel? | 11:09 |
| laurence | toscalix, I think youtube is CC friendly? I have seen videos posted their with CC licence, anyway | 11:11 |
| laurence | adds68, I think it's better to have our own | 11:11 |
| tpollard | youtube is CC friendly | 11:11 |
| toscalix | then youtube seems fine | 11:11 |
| toscalix | let's create our own channel, I would say | 11:12 |
| laurence | agree | 11:14 |
| laurence | did we ever get a twitter handle, toscalix ? | 11:15 |
| toscalix | laurence: no | 11:15 |
| toscalix | it is in todo | 11:15 |
| laurence | Seem to have been able to create 'BuildStream' channel, using my personal account | 11:21 |
| laurence | Wonder if i can set up others as admins | 11:21 |
| laurence | firstly, reduce the bus factor ! | 11:21 |
| gitlab-br-bot | adds68 opened issue #764 (Stack trace when integrating a local plugin) on buildstream https://gitlab.com/BuildStream/buildstream/issues/764 | 11:51 |
| adds68 | could someone take a look that and see if it's something i am doing that is causing the stack trace? ^ thanks | 11:52 |
| juergbi | WSalmon: I commented on !897 | 11:53 |
| gitlab-br-bot | MR !897: Updated Workspace CLI https://gitlab.com/BuildStream/buildstream/merge_requests/897 | 11:53 |
| tpollard | is https://testcache.codethink.co.uk:11001 down? | 11:57 |
| *** ctolentino has joined #buildstream | 11:58 | |
| adds68 | tpollard, it was | 11:58 |
| adds68 | tpollard, back up now | 11:58 |
| tpollard | cheers adds68 | 11:59 |
| benschubert | Anyone having used bst on windows (WSL)? Running bst build my_element.bst, stays stuck at "loading elements" for multiple minutes (even though I only have one dependency on my element, which then has none) | 12:02 |
| benschubert | 4:44 minutes last invocation | 12:03 |
| Kinnison | benschubert: if you invoke bst with debugging turned up, does it tell you anything useful? | 12:05 |
| Kinnison | benschubert: bst --verbose --debug build my_element.bst | 12:05 |
| tpollard | adds68: hmm, I'm still getting failures to initialise on it | 12:11 |
| benschubert | https://gitlab.com/snippets/1777685 just this Kinnison :/ | 12:12 |
| *** jonathanmaw_ is now known as jonathanmaw | 12:19 | |
| juergbi | benschubert: 404 on snippet. might strace give any hints? (I think WSL supports strace) | 12:23 |
| benschubert | juergbi: the 404 should be fixed, I will try after lunch, thanks! | 12:24 |
| juergbi | ah, it actually does succeed, not a complete hang | 12:26 |
| benschubert | Yeah, which is weird, a complete hang might have been better to debug | 12:26 |
| juergbi | filesystem I/O is a lot slower in WSL, afaik. however, it shouldn't do much with a single dependency, so that can't explain it | 12:27 |
| benschubert | And would there be a way of getting the line where an exception is thrown in the logs? I just get a "Failed to link", which is hard to trace back given the number of file operations we do | 12:27 |
| benschubert | Yeah, and staging a 1 Gig tar archive is a lot faster | 12:27 |
| benschubert | (Off for food) | 12:29 |
| juergbi | benschubert: you need strace -f to also trace the child processes | 12:47 |
| *** jonathanmaw_ has joined #buildstream | 12:51 | |
| *** jonathanmaw has quit IRC | 12:52 | |
| adds68 | does anyone have any idea what this error means? | 13:13 |
| adds68 | provenance.elements = [e.clone for e in self.elements] | 13:13 |
| adds68 | AttributeError: 'function' object has no attribute 'clone' | 13:13 |
| Kinnison | That suggests something in self.elements is a function rather than an element object | 13:14 |
| adds68 | The stack trace is coming from: File "/home/adds68/.local/lib/python3.7/site-packages/buildstream/_yaml.py", | 13:14 |
| *** rdale has quit IRC | 13:15 | |
| *** rdale has joined #buildstream | 13:19 | |
| benschubert | juergbi: it seems like it is doing loads of file creation. Why would it do that for loading the elements? | 13:40 |
| juergbi | benschubert: do you see the filenames? what kind of files is it creating? | 13:41 |
| juergbi | also, I assume there is no junction involved, is there? | 13:41 |
| benschubert | There is a junction yes, and it's file relative to the junction that are being created | 13:41 |
| benschubert | (it's a huge junction) | 13:41 |
| juergbi | ah, so it's not just a single dependency after all | 13:43 |
| juergbi | the project pointed to by the junction has to be fetched (if not cached) and then checked out into a temporary directory to be able to access it | 13:43 |
| benschubert | True, sorry, it is a single file in another junction. However, the junction is already cached. Is it checked out everytime? | 13:44 |
| juergbi | the temp directory itself is not reused across invocations | 13:44 |
| juergbi | the git clone is cached but as you could use different revisions we can't easily cache the checkout | 13:44 |
| * Kinnison wonders if adding the capability to buildstream to use pygit2 might help in the case of a junctioned project from git | 13:45 | |
| Kinnison | bzrlib is already python so could be used directly too as an optimisation path | 13:45 |
| juergbi | Kinnison: would this allow reading files without actual checkout? | 13:45 |
| benschubert | So that means operations will get slower as the junction gets a bigger history? | 13:45 |
| benschubert | juergbi: pygit2 would allow this yes | 13:45 |
| juergbi | benschubert: the history doesn't really matter, but the current working size | 13:46 |
| benschubert | Ouch ok | 13:46 |
| juergbi | Kinnison: could theoretically be interesting, but the source API doesn't allow access to individual files | 13:46 |
| Kinnison | juergbi: Mmm :( | 13:46 |
| juergbi | however, the planned source cache in CAS should actually help for this | 13:46 |
| Kinnison | nod. | 13:47 |
| benschubert | Any way to get where an exception was thrown? :/ The log just gives me that a file copy failed | 13:48 |
| WSalmon | tlater[m], valentind ^ have you any good tips for this? | 15:01 |
| *** lachlan has quit IRC | 15:07 | |
| *** lachlan has joined #buildstream | 15:13 | |
| *** lachlan has quit IRC | 15:20 | |
| *** tristan has joined #buildstream | 15:22 | |
| *** lachlan has joined #buildstream | 15:26 | |
| *** alatiera_ has quit IRC | 15:58 | |
| *** alatiera_ has joined #buildstream | 15:58 | |
| *** alatiera_ has quit IRC | 16:05 | |
| *** jonathanmaw has joined #buildstream | 16:21 | |
| *** jonathanmaw_ has quit IRC | 16:21 | |
| valentind | WSalmon, benschubert, you want to log every place an exception is raised? | 16:39 |
| WSalmon | no, he basicly wants the full stack trace so he can work out what is happening, is there a option like --debug that can turn this on? | 16:41 |
| benschubert | valentind: I want, when I get an exception in Buildstream that is caught and printed nicely, get the full stack trace so that I can actually debug it :) (tried with --debug and --verbose but without luck) | 16:41 |
| *** adds68 has quit IRC | 16:42 | |
| valentind | I am not aware of any way to do that. | 16:44 |
| valentind | benschubert, is your issue reproducible? If so just modify buildstream to not catch the exception the place it prints the error. | 16:45 |
| valentind | benschubert, could I see the error message? | 16:46 |
| valentind | Because an error copying file sounds like an issue I am currently fixing. | 16:46 |
| WSalmon | so the way myself and ben deal with this is to change the error type so it acts like a bug and give you the full stack but that is ugly as sin | 16:46 |
| WSalmon | do you do it that way too ben? | 16:48 |
| valentind | WSalmon, yes, I know it is ugly. The problem of issue copying files is that usually it is a problem from the file system. But sometimes it is also a bug. We should not catch bugs. Unfortunately, for this case, it is hard to decide if we need to not catch. | 16:48 |
| benschubert | valentind: https://gitlab.com/snippets/1777794 | 16:48 |
| benschubert | Yeah, I'd like a real error :/ | 16:48 |
| laurence | BuildStream YouTube channel - https://www.youtube.com/channel/UCU8gOKOs1C_DFBaSvd5HitQ | 16:48 |
| valentind | symbolic links and compose elements... | 16:49 |
| valentind | benschubert, It is this issue: #647 | 16:50 |
| gitlab-br-bot | Issue #647: BuildStream misorders elements in subdirectories of symlinks when link_files called in compose elements https://gitlab.com/BuildStream/buildstream/issues/647 | 16:50 |
| valentind | I think. | 16:50 |
| valentind | If you look at the logs | 16:50 |
| valentind | [00:00:02] FAILURE vm/minimal-systemd-vm.bst: Failed to link '/builds/freedesktop-sdk/freedesktop-sdk/cache/buildstream/build/vm-minimal-systemd-vm-uftjaonx/root -> /builds/freedesktop-sdk/freedesktop-sdk/cache/buildstream/build/vm-minimal-systemd-vm-uftjaonx/root/buildstream/install': [Errno 2] No such file or directory: '/builds/freedesktop-sdk/freedesktop-sdk/cache/buildstream/build/vm-minimal-systemd-vm-uftjaonx/root/buildstream/install/etc/in | 16:50 |
| valentind | it.d' | 16:50 |
| * WSalmon wonders if it would be hard/tristan blocked to have a --died-ugly option to get bst that made every error more like a "bug" | 16:50 | |
| valentind | [00:01:15] FAILURE desktop-platform-image.bst: Failed to link '/builds/freedesktop-sdk/freedesktop-sdk/cache/buildstream/build/desktop-platform-image-8snp170t/root -> /builds/freedesktop-sdk/freedesktop-sdk/cache/buildstream/build/desktop-platform-image-8snp170t/root/buildstream/install': [Errno 2] No such file or directory: '/builds/freedesktop-sdk/freedesktop-sdk/cache/buildstream/build/desktop-platform-image-8snp170t/root/buildstream/install/us | 16:50 |
| valentind | r/etc/OpenCL/vendors' | 16:50 |
| valentind | laurence, great! I smash that subscribe button! | 16:51 |
| Kinnison | WSalmon: surely that should just be under --debug ? | 16:51 |
| valentind | benschubert, but in your case it is invalid argument. | 16:52 |
| laurence | valentind, :) | 16:54 |
| Kinnison | laurence: to share an account between users you have to make it a "brand" | 16:54 |
| Kinnison | laurence: https://support.google.com/youtube/answer/3056283 | 16:54 |
| valentind | benschubert, it looks like an issue staging the artifact. Have you looked at /home/bschubert15/.cache/buildstream/build/test-2iujds_s/root/opt/XXXX/XXXXX/inventory/v17n1? | 16:55 |
| valentind | Is that a directory? | 16:55 |
| valentind | Is that a symlink, does the target directory exist? | 16:55 |
| benschubert | let me check :) And I +1, this should be done by --debug :D | 16:56 |
| benschubert | valentind: it's a file with mode "755" that is not owned by my user | 16:59 |
| benschubert | in artifacts/pwd | 16:59 |
| valentind | benschubert, Ok so I see, you have an artifact that creates it as a file, and another as a directory. | 16:59 |
| benschubert | in artifacts/extract/XXX/bootstrap/..... | 16:59 |
| benschubert | But, I have a single platform.bst under my element (which doesn't have any source) | 16:59 |
| valentind | benschubert, no I was talking about the build directory. | 17:00 |
| benschubert | is not existing anymore so I can't check | 17:01 |
| valentind | The "v17n1" directory | 17:01 |
| valentind | Annoying. I was not aware build directories were removed when there is an error. | 17:01 |
| benschubert | let me retry | 17:02 |
| laurence | Kinnison, ta, will look into it | 17:02 |
| valentind | benschubert, what file system do you use there? ext4? | 17:10 |
| benschubert | ntfs exported as samba to a ubuntu machine (WSL with Docker sandbox (a POC I'm working on)) | 17:11 |
| valentind | benschubert, WSL? So not using the linux kernel? | 17:12 |
| valentind | I was trying to find in the linux kernel where it could return EINVAL for a link syscall. But was not finding it. | 17:12 |
| benschubert | No, except for the build part which runs on ubuntu (the docker vm) | 17:12 |
| benschubert | Sorry should have mentionned that before | 17:13 |
| valentind | benschubert, Is bst running on a WSL or on Linux? | 17:13 |
| benschubert | it definitely works on linux with bwrap | 17:13 |
| benschubert | bst is running on wsl | 17:13 |
| valentind | Oh ok. | 17:13 |
| valentind | Makes more sense. | 17:14 |
| benschubert | Sorry :) My current work is doing a docker POC so we could have bst working natively on Windows and build for linux via docker :) | 17:14 |
| valentind | Though I do not know when WSL returns EINVAL on link. | 17:14 |
| valentind | So it could be still the same as #647, just that WSL does not show the same error. | 17:15 |
| gitlab-br-bot | Issue #647: BuildStream misorders elements in subdirectories of symlinks when link_files called in compose elements https://gitlab.com/BuildStream/buildstream/issues/647 | 17:15 |
| benschubert | Yeah, I'm not exactly sure, I'm trying to investigate but getting the stack traces was not that nice | 17:15 |
| valentind | But did you say the same project worked correctly with bwrap? | 17:15 |
| benschubert | Yes! | 17:16 |
| benschubert | on a real linux machine, and bwrap it works perfectly on master | 17:16 |
| benschubert | so it's a linux/windows compatibility problem I guess | 17:16 |
| valentind | benschubert, my recommendation here is to try to rebuild it again, and make sure you keep the build directory. Then take the two last paths from the error message, and try to us "ln" manually to see if you can reproduce manually (not symbolic). Then figure out why. | 17:17 |
| benschubert | It's building :) | 17:18 |
| valentind | By the way, how does WSL deal with max_path being 260 in Windows? | 17:18 |
| benschubert | badly | 17:19 |
| benschubert | very badly | 17:19 |
| benschubert | It's one of the next bugs I need to see how to fix :) | 17:19 |
| valentind | benschubert, is there some translation of linux paths to windows that adds a path in front? | 17:19 |
| benschubert | Not sure what you mean there | 17:20 |
| juergbi | hm, I thought that was only a Win32 limitation, not an NT kernel limitation | 17:20 |
| juergbi | so I wouldn't have expected issues in WSL | 17:20 |
| valentind | juergbi, it is not an NT kernel limitation. | 17:21 |
| valentind | But most API still suffers from it. | 17:21 |
| juergbi | sure but WSL shouldn't be related to the Win32 API at all | 17:21 |
| skullman | jonathanmaw: I reviewed https://gitlab.com/BuildStream/buildstream/merge_requests/924 | 17:21 |
| jonathanmaw | skullman: ta | 17:21 |
| valentind | Unless you expectly use NT path. I think it was something prefixed by \\? . or something. But then I am not sure that all API works. | 17:22 |
| benschubert | juergbi: we definitely do not work with that, the only tests that are failing (if my PRs get merged) will be the PATH problem | 17:22 |
| jonathanmaw | I'll respond to it when I've scraped together enough brain | 17:22 |
| *** abderrahim4 has joined #buildstream | 17:30 | |
| *** abderrahim3 has quit IRC | 17:32 | |
| benschubert | valentind: still trying to reproduce. I had wiped my cache before and it seems to... work now -_-' | 17:38 |
| benschubert | Can't reproduce -_-' | 17:50 |
| benschubert | So I guess busted cache while playing with my sandbox | 17:50 |
| valentind | benschubert, ok | 17:51 |
| *** lachlan has quit IRC | 17:56 | |
| *** lachlan has joined #buildstream | 18:04 | |
| *** jonathanmaw has quit IRC | 18:08 | |
| *** jmac has quit IRC | 18:15 | |
| *** benbrown has quit IRC | 18:16 | |
| *** Nexus has quit IRC | 18:16 | |
| *** johnward has quit IRC | 18:16 | |
| *** laurence has quit IRC | 18:16 | |
| *** paulsherwood has quit IRC | 18:16 | |
| *** bethw has quit IRC | 18:16 | |
| *** jmac has joined #buildstream | 18:16 | |
| *** ikerperez has quit IRC | 18:17 | |
| *** benbrown has joined #buildstream | 18:17 | |
| *** Nexus has joined #buildstream | 18:17 | |
| *** phildawson has quit IRC | 18:17 | |
| *** jennis has quit IRC | 18:17 | |
| *** bethw has joined #buildstream | 18:17 | |
| *** coldtom has quit IRC | 18:17 | |
| *** paulsherwood has joined #buildstream | 18:17 | |
| *** johnward has joined #buildstream | 18:19 | |
| *** laurence has joined #buildstream | 18:21 | |
| jjardon | Hi, is there any issue to track the "bst artifact" group of commands introduction? | 18:37 |
| jjardon | nevermind, found it: https://gitlab.com/BuildStream/buildstream/issues/234 | 19:00 |
| *** xjuan has joined #buildstream | 19:16 | |
| *** raoul has quit IRC | 19:18 | |
| *** xjuan has quit IRC | 19:56 | |
| *** tristan has quit IRC | 20:03 | |
| *** benschubert has quit IRC | 21:10 | |
| *** xjuan has joined #buildstream | 21:57 | |
| *** wayne has joined #buildstream | 22:05 | |
| *** catonano has joined #buildstream | 22:37 | |
| *** rdale has quit IRC | 22:40 | |
| *** catonano has quit IRC | 22:47 | |
| *** catonano has joined #buildstream | 22:47 | |
| *** catonano has quit IRC | 22:56 | |
| *** catonano has joined #buildstream | 22:56 | |
| *** catonano has quit IRC | 23:03 | |
| *** xjuan has quit IRC | 23:08 | |
| *** toscalix has quit IRC | 23:25 | |
| *** mohan43u has quit IRC | 23:41 | |
| *** mohan43u has joined #buildstream | 23:44 | |
| *** mohan43u has quit IRC | 23:49 | |
| *** mohan43u has joined #buildstream | 23:53 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!