| *** tristan has quit IRC | 05:35 | |
| *** tristan has joined #buildstream | 05:52 | |
| *** ChanServ sets mode: +o tristan | 05:52 | |
| *** jonathanmaw has joined #buildstream | 08:22 | |
| *** ssam2 has joined #buildstream | 08:52 | |
| *** tlater has joined #buildstream | 09:02 | |
| gitlab-br-bot | push on buildstream@master (by Tristan Van Berkom): 2 commits (last: _yaml.py: Strengthen node_get() to handle TypeError in type conversions.) https://gitlab.com/BuildStream/buildstream/commit/2c79d0aa52e582a6553a2230407d1fde9f11ae4d | 09:04 |
|---|---|---|
| jonathanmaw | hmm, looking at what debian builds give me, artifact splitting would best be served by generating the list of files to add by listing the files inside the per-package subdir | 09:21 |
| jonathanmaw | which means calculating them after the build | 09:21 |
| jonathanmaw | which is pretty unfortunate, given that's usually included in the cache key | 09:26 |
| jonathanmaw | the alternative would be to install the packages within their subdir (i.e. /bsdgames/usr/bin), but that's not a usable artifact | 09:27 |
| *** persia has joined #buildstream | 10:27 | |
| *** tristan has quit IRC | 12:06 | |
| *** jonathanmaw has quit IRC | 12:24 | |
| *** jonathanmaw has joined #buildstream | 12:42 | |
| *** tristan has joined #buildstream | 12:47 | |
| *** ChanServ sets mode: +o tristan | 12:48 | |
| tlater | How could I modify the dependencies returned by _pipeline.dependencies()? I currently have a function which removes dependencies from a list after that function, but that doesn't show up in app.print_heading() | 14:14 |
| tristan | tlater, better to also do it for `bst show` btw | 14:17 |
| tristan | tlater, so print_heading() takes a list of dependencies (elements) | 14:18 |
| tristan | that is what will be printed | 14:18 |
| tlater | That list is usually taken from deps_elements, which in turn takes dependencies from _pipeline.dependencies() | 14:19 |
| tlater | I could modify the dependencies returned by that after it's done, but I feel like this should stay in pipeline. | 14:21 |
| tristan | tlater, yeah I see what you mean... it's a bit icky | 14:23 |
| tristan | tlater, I guess the best is the move the main.c app.deps_elements() logic to _pipeline() too | 14:24 |
| tristan | ideally | 14:24 |
| tristan | main.py | 14:24 |
| tristan | heh | 14:24 |
| tlater | Yeah, that could work | 14:25 |
| tlater | Just basically move that function to _pipeline.py? | 14:25 |
| jonathanmaw | tristan: I think your changes to sandboxing might have broken bst checkout. It seems to be creating a complete rootfs, instead of just extracting the target. | 14:35 |
| tlater | Oh. Well that explains why that keeps happening. | 14:37 |
| jonathanmaw | tristan: hmm, I'll have to confirm that, checking out the base-system-image didn't seem to have everything in it, might have to check some stuff first. | 14:57 |
| tristan | jonathanmaw, I did that intentionally after the sandboxing changes | 14:59 |
| tristan | as a separate patch, does it break anything ? | 14:59 |
| tristan | jonathanmaw, my rationale was, originally it only "checks out an artifact", which was interesting for things like, checking out a built system image | 15:00 |
| jonathanmaw | tristan: I was using `bst checkout` to check what the artifact contained | 15:00 |
| tristan | jonathanmaw, but as it can also be useful for say, checking out a gnu-toolchain and having something to chroot into... | 15:00 |
| tristan | my rationale was that staging and integrating Scope.RUN had the same effect anyway for a system image | 15:00 |
| tristan | I see | 15:00 |
| tristan | jonathanmaw, I suppose we could add the synonymous --deps (and --except args we're adding to other commands), to get that done too, but then integrating would have to be optional too | 15:01 |
| tristan | ok coffee shop closing time... I'm out ! | 15:02 |
| *** tristan has quit IRC | 15:05 | |
| ssam2 | i agree about `bst checkout`, at least there's a use case for seeing what files an artifact contains | 15:14 |
| *** tristan has joined #buildstream | 15:14 | |
| tlater | I've been using the sandbox as a semi-temporary directory for the source-bundle command as well. | 15:15 |
| tristan | ssam2, nod, agree | 15:18 |
| tristan | jonathanmaw, ok so in that case... while I'm away... you could either do that, or wait until tlater submits that MR he was talking about with the --except stuff as it will probably avoid a conflict | 15:19 |
| tristan | it should be a piece of cake to implement anyway | 15:19 |
| tlater | I'm merging after I run the tests one last time | 15:19 |
| jonathanmaw | tristan: it's fine, I just wasn't sure what was going on. | 15:19 |
| tristan | but in the meantime you can always look at the ostree repo or extract directories :) | 15:20 |
| tristan | an invocation to `bst shell --scope run <target>` will provoke an extract directory to exist | 15:20 |
| ssam2 | i seem to have cross compiled a powerpc64le sysroot with buildstream :-) | 15:21 |
| * tristan heads out for a midnight gym | 15:21 | |
| tristan | ssam2, nice ! \o/ | 15:21 |
| tristan | better try chrooting it on a powerpc64le system :) | 15:21 |
| ssam2 | yep | 15:22 |
| ssam2 | i think we have one! | 15:22 |
| ssam2 | need to disable integration commands for cross built stuff though so `bst checkout` stops trying to execute the powerpc64le `ldconfig` :-) | 15:22 |
| tristan | ahhh interesting, indeed | 15:23 |
| tristan | for bst checkout | 15:23 |
| tristan | ssam2, another reason why integration must be optional at checkout time ! | 15:23 |
| tristan | ssam2, in the context of the build of gnu-toolchain though, I think we dont have to care; i.e. the ldconfig we run must be the one of the host runnable cross toolchain | 15:24 |
| ssam2 | i think it's actually the shell that's causing issues | 15:24 |
| tristan | I guess in terms of building, the gnu-toolchain.bst stack element should provide the ldconfig integration command right there | 15:25 |
| tristan | so building gnu-toolchain.bst will never execute it | 15:25 |
| tristan | only running it will | 15:25 |
| tlater | My merge request is up | 15:27 |
| *** jonathanmaw has quit IRC | 16:19 | |
| ssam2 | urgh, the powerpc64le sysroot contains /lib64/ld-linux-x86-64.so.2 | 16:19 |
| ssam2 | and no other ld.so | 16:20 |
| ssam2 | this most probably explains why the compiler produces binaries that don't work | 16:20 |
| ssam2 | ah, just another place where I need to s/%{bst-arch}/%{bst-target-arch}/ | 16:21 |
| *** tlater has quit IRC | 16:54 | |
| *** ssam2 has quit IRC | 16:59 | |
| *** ssam2 has joined #buildstream | 17:01 | |
| *** ChanServ sets mode: +o tristan | 17:42 | |
| *** ssam2 has quit IRC | 18:08 | |
| *** tristan has quit IRC | 18:12 | |
| *** albfan[m] has quit IRC | 18:12 | |
| *** waltervargas[m] has quit IRC | 18:12 | |
| *** mattiasb has quit IRC | 18:12 | |
| *** tristan has joined #buildstream | 18:24 | |
| *** albfan[m] has joined #buildstream | 19:07 | |
| *** mattiasb has joined #buildstream | 19:10 | |
| gitlab-br-bot | buildstream: merge request (except->master: Add a --except flag to show, fetch and track) #26 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/26 | 19:31 |
| *** waltervargas[m] has joined #buildstream | 20:33 | |
| *** tristan has quit IRC | 21:01 | |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!