| *** reborn has quit IRC | 01:17 | |
| *** nimish has joined #buildstream | 01:18 | |
| *** nimish has quit IRC | 01:40 | |
| *** nimish has joined #buildstream | 01:57 | |
| *** nimish has quit IRC | 02:22 | |
| *** nimish has joined #buildstream | 02:51 | |
| *** nimish has quit IRC | 03:01 | |
| *** toscalix has joined #buildstream | 08:41 | |
| *** alatiera has joined #buildstream | 09:04 | |
| *** benschubert has joined #buildstream | 09:17 | |
| *** sambishop has joined #buildstream | 09:19 | |
| gitlab-br-bot | juergbi approved MR !1095 (valentindavid/crash_in_scheduler_857->master: Fix crash when spawned job completes very fast) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1095 | 09:23 |
|---|---|---|
| gitlab-br-bot | juergbi approved MR !1099 (valentindavid/wrong_type_in_status_code->master: Fix type of error codes in CAS server) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1099 | 09:23 |
| gitlab-br-bot | juergbi approved MR !1092 (valentindavid/make_cache_dir->master: Make sure testing cache directory exists) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1092 | 09:27 |
| gitlab-br-bot | juergbi approved MR !1086 (aevri/bst_track_guidance->master: Fixup refs to 'bst track' and 'bst fetch') on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1086 | 09:33 |
| valentind | juergbi, I think we do not need !1092 anymore. The merge for centos fixed it in another way. | 09:46 |
| valentind | I will close it. | 09:46 |
| gitlab-br-bot | valentindavid closed MR !1092 (valentindavid/make_cache_dir->master: Make sure testing cache directory exists) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1092 | 09:47 |
| juergbi | ah, handled in .gitlab-ci.yml now | 09:47 |
| juergbi | valentind: couldn't this still be an issue when running integration tests locally? | 09:48 |
| valentind | juergbi, maybe. I was fixing an issue with the arm builder. | 09:48 |
| valentind | We can reopen it if someone has an issue locally. | 09:48 |
| juergbi | tox doesn't seem to like Ctrl+C, really annoying | 09:54 |
| juergbi | it keeps running in the background | 09:55 |
| juergbi | have to manually killall pytest | 09:55 |
| *** raoul has joined #buildstream | 09:56 | |
| gitlab-br-bot | juergbi reopened MR !1092 (valentindavid/make_cache_dir->master: Make sure testing cache directory exists) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1092 | 09:56 |
| *** tpollard has joined #buildstream | 10:06 | |
| *** jonathanmaw has joined #buildstream | 10:12 | |
| *** rdale has joined #buildstream | 10:15 | |
| *** SotK is now known as SotK_ | 10:45 | |
| *** SotK has joined #buildstream | 10:45 | |
| phildawson | Hi all, I've got a couple of MRs up at the moment that could do with some review. If anyone has time to take a look at !1057 and/or !1075 I'd much appreciate it. | 10:52 |
| gitlab-br-bot | MR !1057: plugin.py: Add API to allow plugins to raise deprecation warnings https://gitlab.com/BuildStream/buildstream/merge_requests/1057 | 10:52 |
| gitlab-br-bot | MR !1075: Expose basic api for testing external plugins. https://gitlab.com/BuildStream/buildstream/merge_requests/1075 | 10:52 |
| laurence | tristan, for this issue re local cache size calculation - https://gitlab.com/BuildStream/buildstream/issues/869 - i think it may be better to ask the original issue raiser to close the ticket once the fix has landed in master | 11:12 |
| laurence | i'm an advocate of this workflow - i think it makes more sense than closing off the tickets directly (as can sometimes happen with gitlab's ui) | 11:12 |
| laurence | do you mind if i re-open and ping benschubert on the ticket? | 11:13 |
| gitlab-br-bot | LaurenceUrhegyi reopened issue #869 (Local cache size calculation is wrong) on buildstream https://gitlab.com/BuildStream/buildstream/issues/869 | 11:14 |
| gitlab-br-bot | aevri merged MR !1086 (aevri/bst_track_guidance->master: Fixup refs to 'bst track' and 'bst fetch') on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1086 | 11:36 |
| jjardon | laurence: that breaks how gitlab is meant to work: issues get closed when the MR gets merged; I think is better to reopen if the issue is seen again after testing, or even open a new issue (the root cause can be a completely different one) | 11:48 |
| laurence | jjardon, I thought it was only if you use the UI from within an issue and press 'create merge request' or due to our gitlab template on MRs (not sure if this template is even still there actually?) | 11:51 |
| laurence | i think we can still link an MR to an issue, without closing the issue, and ask the original submitter to verify (this depends on them using master branch or a dev branch) | 11:52 |
| laurence | not sure if we can automate it just yet | 11:52 |
| laurence | anyway, i think we'll discuss gitlab usage / governance / committer policy over the next week | 11:52 |
| jjardon | no; if you put in the merge request (or in the commit message in the commit of the git branch) "Fixes #12", Issue 12 will be fixed when the MR / commit gets merged. github works in the same way | 11:53 |
| gitlab-br-bot | Issue #12: installing buildstream on fresh jessie from instructions at http://buildstream.gitlab.io/buildstream/install.html#installing https://gitlab.com/BuildStream/buildstream/issues/12 | 11:53 |
| jjardon | This is if you use "Fixes" or "Closes" keywords; if not the issue/Mr are linked but issue doesn't get closed | 11:54 |
| jjardon | laurence: also, problem with ask to verify is that people forget and then you will have a long list of issues that are fixed but not closed because they are verify yet; I think is better to be proactive and close them and expect the reporter to reopen / open a new one if he sees the problem again | 11:56 |
| laurence | hmmm that's another angle, although is frustrating if someone closes an issue that doesn't fix my issue. | 11:59 |
| persia | My experience with systems that depend on reporter feedback is that most of the reporter feedback is not useful for development (sadly). I think a lot of reporters are used to either opening new issues (with a subtly different description and referencing the issue that was incorrectly diagnosed), or to reopen the issue with something like "Thanks for fixing that other bug, but the problem I described initially still occurs (with a test case or | 12:00 |
| persia | example output, etc.). | 12:00 |
| jjardon | laurence: sure, but you get a notification when it gets closed, so you can quickly check; although sometimes It doesn't get fixed, we should assume the dev knows what he is doing and the default is that it actually gets fixed | 12:01 |
| laurence | jjardon, if this were a community driven project, i'd be tempted to agree | 12:02 |
| laurence | anyway, it's not a very important point, not one to debate for a long time :) | 12:03 |
| jjardon | laurence: you have a point there, sure | 12:04 |
| *** cs-shadow has joined #buildstream | 12:09 | |
| laurence | Reading some back-scroll from this channel, I see another topic to bring for discussion this week, about disabling caches for a command | 12:12 |
| laurence | tpollard started some work on this a while back - https://gitlab.com/BuildStream/buildstream/merge_requests/950 | 12:12 |
| laurence | will raise it this week | 12:12 |
| jjardon | jennis: hey, seems https://gitlab.com/BuildStream/buildstream/tree/jennis/add_new_profile_topic doesn't exist anymore; Did it got merged? | 12:15 |
| jennis | jjardon, yeah pretty sure it did | 12:16 |
| jennis | one sec | 12:16 |
| jennis | jjardon, yes it exists in master, I think the source branch just got removed upon merge | 12:16 |
| jennis | Do you want me to repush? | 12:17 |
| jjardon | jennis: no need, thanks | 12:19 |
| *** raoul has quit IRC | 12:29 | |
| *** sambishop has quit IRC | 13:01 | |
| *** bochecha has joined #buildstream | 13:01 | |
| gitlab-br-bot | jennis opened MR !1112 (jennis/filter-docs->master: Improve our filter documentation) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1112 | 13:07 |
| jennis | jonathanmaw, would you be able to spare some time to review https://gitlab.com/BuildStream/buildstream/merge_requests/1112 ? It's an improvement (I hope...) to our filter documentation | 13:08 |
| jonathanmaw | jennis: I'll have a look | 13:09 |
| *** raoul has joined #buildstream | 13:14 | |
| *** bochecha has quit IRC | 13:16 | |
| *** sambishop has joined #buildstream | 13:22 | |
| abderrahim[m] | laurence : case in point, looking through the bug list, I found #179 and #706 which are fixed but are waiting for an ack | 13:24 |
| gitlab-br-bot | Issue #179: Be smarter about querying summaries from remote caches https://gitlab.com/BuildStream/buildstream/issues/179 | 13:24 |
| gitlab-br-bot | Issue #706: Information when a circular dependency is detected is not good enough https://gitlab.com/BuildStream/buildstream/issues/706 | 13:24 |
| Kinnison | abderrahim[m]: Hi! Thanks for the data sets you provided. | 13:24 |
| abderrahim[m] | jjardon: adds68 ^ | 13:25 |
| abderrahim[m] | Kinnison: :) | 13:25 |
| juergbi | laurence: regarding disabling artifact servers, my proposal for artifact server configuration was supposed to cover this as well | 13:28 |
| juergbi | i.e., I think we need an overall discussion on artifact server configuration | 13:28 |
| abderrahim[m] | speaking of artifact servers, I have a branch to fix #401 | 13:29 |
| gitlab-br-bot | Issue #401: If a project B use project A as a junction, B should not try to use the bst remote cache of A (at least by default) https://gitlab.com/BuildStream/buildstream/issues/401 | 13:29 |
| jonathanmaw | jennis: I'm finished reviewing !1112 | 13:32 |
| gitlab-br-bot | abderrahimk opened MR !1113 (abderrahim/artifact-cache-junction->master: Use artifact cache specs from the parent project before those defined in junctions) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1113 | 13:37 |
| tpollard | interesting | 13:41 |
| gitlab-br-bot | tpollard closed issue #829 (Download buildtrees on demand for bst shell --build) on buildstream https://gitlab.com/BuildStream/buildstream/issues/829 | 13:44 |
| gitlab-br-bot | tpollard merged MR !1050 (tpollard/829->master: Download buildtrees on demand for bst shell --use-buildtree) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1050 | 13:44 |
| *** nimish has joined #buildstream | 13:52 | |
| *** lachlan has joined #buildstream | 13:56 | |
| *** sambishop has quit IRC | 13:56 | |
| *** nimish has quit IRC | 13:56 | |
| *** nimish has joined #buildstream | 13:57 | |
| *** nimish has joined #buildstream | 13:57 | |
| *** raoul has quit IRC | 14:01 | |
| *** sambishop has joined #buildstream | 14:01 | |
| *** lachlan has quit IRC | 14:03 | |
| *** nimish_ has joined #buildstream | 14:03 | |
| *** nimish has quit IRC | 14:03 | |
| *** nimish_ is now known as nimish | 14:04 | |
| *** lachlan has joined #buildstream | 14:06 | |
| paulsherwood | 12:02 < laurence> jjardon, if this were a community driven project, i'd be tempted to agree | 14:06 |
| paulsherwood | laurence: what makes you think it isn't? | 14:06 |
| *** nimish has quit IRC | 14:13 | |
| *** nimish has joined #buildstream | 14:13 | |
| paulsherwood | well, in any case, i believe it is | 14:19 |
| gitlab-br-bot | jjardon opened issue #884 (Traceback with current master (cd9b8e54aad6aa885afd28cefbbd40ea77c9353d)) on buildstream https://gitlab.com/BuildStream/buildstream/issues/884 | 14:23 |
| *** nimish has quit IRC | 14:23 | |
| *** nimish has joined #buildstream | 14:24 | |
| jjardon | Does anyone have the same problem as ^ with current master? | 14:26 |
| abderrahim[m] | jjardon: I believe you need to update grpcio | 14:29 |
| jonathanmaw | benschubert: would you like to have a look at https://gitlab.com/BuildStream/buildstream/merge_requests/1108 ? | 14:29 |
| juergbi | yes, can you try with: pip3 install --user -U --upgrade-strategy eager . | 14:29 |
| juergbi | (optionally also use -e) | 14:30 |
| juergbi | pip doesn't upgrade dependencies automatically otherwise, iirc | 14:30 |
| abderrahim[m] | I never found the time to search for the minimum version that works, but it definitely won't work with 1.10 | 14:30 |
| benschubert | jonathanmaw: thanks I'll have a look | 14:30 |
| abderrahim[m] | juergbi: well it does if the minimum requirement isn't met | 14:30 |
| juergbi | right | 14:31 |
| juergbi | so if it doesn't work right now, we may need to bump our minimum reqs | 14:31 |
| abderrahim[m] | jjardon: what version of grpcio do you have? | 14:32 |
| jjardon | 1.17.1 | 14:32 |
| *** nimish has quit IRC | 14:34 | |
| *** nimish has joined #buildstream | 14:35 | |
| jonathanmaw | ta benschubert | 14:35 |
| abderrahim[m] | jjardon: works fine here with 1.16.1 and 1.17.1, maybe it's protobuf then | 14:43 |
| abderrahim[m] | (and make sure you don't have more than one version installed, that may be also the problem) | 14:44 |
| *** nimish has quit IRC | 14:45 | |
| *** sambishop has quit IRC | 14:45 | |
| *** raoul has joined #buildstream | 14:46 | |
| *** nimish has joined #buildstream | 14:48 | |
| *** nimish has joined #buildstream | 14:49 | |
| *** lachlan has quit IRC | 14:50 | |
| *** nimish has quit IRC | 14:50 | |
| abderrahim[m] | jjardon: yeah, it's protobuf. You need at least 3.6.0 | 14:53 |
| jennis | thanks for the review jonathanmaw | 14:56 |
| jjardon | abderrahim[m]: mmm, I wonder how come the required version didn't got install with "pip install --user ." | 14:57 |
| jjardon | ah, the requires is >=3.5 | 14:59 |
| gitlab-br-bot | abderrahimk opened MR !1114 (abderrahim/protobuf-version->master: requirements/requirements.in: require protobuf >= 3.6) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1114 | 14:59 |
| jjardon | with protobuf 3.6 I got another error: | 15:00 |
| jjardon | https://www.irccloud.com/pastebin/iyWI6qJR/ | 15:00 |
| *** sambishop has joined #buildstream | 15:01 | |
| juergbi | jjardon: that would be too old system click | 15:03 |
| tpollard | yup | 15:05 |
| jjardon | mmm | 15:06 |
| jjardon | Is "pip3 install --user ." not the recommended way to install buildstream anymore? | 15:06 |
| abderrahim[m] | jjardon: if you have an older pip, it works | 15:07 |
| abderrahim[m] | with a newer version, you can use --upgrade-strategy eager | 15:07 |
| abderrahim[m] | but I agree that this should be fixed | 15:07 |
| jjardon | Are we not exercising the default installation anywhere? Maybe we can add some CI job somewhere | 15:08 |
| jjardon | and yeah we should update the docs | 15:08 |
| abderrahim[m] | jjardon: by updating the minimum version, it should work | 15:09 |
| abderrahim[m] | it would be nice to have a CI job to test things with the oldest supported version | 15:09 |
| *** lachlan has joined #buildstream | 15:09 | |
| tpollard | the requirements file is pinned to 7.0 | 15:10 |
| jjardon | ok, so if I use requirements/requirements.txt instead requirements/requirements.in in setup.py, everything works as expected | 15:10 |
| jjardon | is this expected or a bug? | 15:10 |
| jennis | Would appreciate a review of https://gitlab.com/BuildStream/buildstream/merge_requests/1112 if anyone has the time | 15:11 |
| abderrahim[m] | jjardon: requirements.txt is used for reproducibility in the CI, so it always works | 15:11 |
| abderrahim[m] | the requirements.in gives the versions that should work, but isn't tested by the CI | 15:12 |
| jjardon | rigth | 15:12 |
| abderrahim[m] | is Click 7.0 the required version? I seem to recall it based on a discussion on the ML | 15:13 |
| jjardon | abderrahim[m]: looking at requirements.txt, seems you are rigth | 15:13 |
| abderrahim[m] | So it works with 7.0, and doesn't with 6.7 | 15:15 |
| tpollard | yeh I think hidden got added with 7.0 | 15:15 |
| abderrahim[m] | and there is no version in between | 15:16 |
| benschubert | jonathanmaw: I'm not sure about https://gitlab.com/BuildStream/buildstream/merge_requests/1108/diffs?commit_id=2d7460c832f9251c7f27e1c5698759fda01b6bfa#ba11c2672dd9b17f396563914c616d753b3b69e9_59_59 . Why is the error only available with sockets and not the rest? | 15:16 |
| * jjardon creates MR | 15:17 | |
| jennis | abderrahim[m], yes Click 7.0 is required as it includes a feature to 'hide' commands, which we're using for various deprecations | 15:17 |
| abderrahim[m] | jjardon: I'll add it to mine | 15:17 |
| jennis | e.g. bst fetch | 15:17 |
| *** raoul_ has joined #buildstream | 15:18 | |
| *** raoul has quit IRC | 15:19 | |
| gitlab-br-bot | jjardon opened MR !1115 (jjardon/fix_minimum_deps->master: requirements/requirements.in: Update minimum dependencies) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1115 | 15:19 |
| *** lachlan has quit IRC | 15:21 | |
| jjardon | abderrahim[m]: ah, you already created one :); please feel free to close mine | 15:21 |
| *** kapil___ has joined #buildstream | 15:21 | |
| gitlab-br-bot | abderrahimk closed MR !1115 (jjardon/fix_minimum_deps->master: requirements/requirements.in: Update minimum dependencies) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1115 | 15:23 |
| jonathanmaw | benschubert: afaict, the problem is only apparent when binding to a unix socket, where the allowed path length is surprisingly short | 15:24 |
| gitlab-br-bot | jjardon approved MR !1114 (abderrahim/protobuf-version->master: requirements/requirements.in: update minimum versions) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1114 | 15:25 |
| gitlab-br-bot | jjardon opened issue #885 (Add CI to check current master works) on buildstream https://gitlab.com/BuildStream/buildstream/issues/885 | 15:29 |
| jonathanmaw | AIUI the unix specification is 108 characters, but linux normally permits longer paths, and WSL's implementation did not. | 15:29 |
| skullman | nah, it's that short on Linux too | 15:30 |
| skullman | the socket api is not good | 15:30 |
| gitlab-br-bot | jjardon closed issue #706 (Information when a circular dependency is detected is not good enough) on buildstream https://gitlab.com/BuildStream/buildstream/issues/706 | 15:32 |
| benschubert | jonathanmaw: ok I see, thanks! | 15:34 |
| jonathanmaw | skullman: hmm, weird we haven't encountered it before, then. Maybe python is usually responsible for working around that and the WSL version took it down an unexpected code path | 15:35 |
| jonathanmaw | either that or the gitlab runner in docker mode's working directory is shorter than the runner in shell mode. | 15:36 |
| *** lachlan has joined #buildstream | 15:37 | |
| skullman | python doesn't work around it, I've had to stick a symlink in a tempdir as a work-around before in python | 15:37 |
| gitlab-br-bot | jjardon closed issue #611 (Using a modern PEP8 linter detects errors/warnings) on buildstream https://gitlab.com/BuildStream/buildstream/issues/611 | 15:42 |
| Kinnison | Should we have our workspace be something like C:\bsws then? | 15:47 |
| jonathanmaw | I could configure my gitlab-runner to use a shorter directory | 15:54 |
| jonathanmaw | though the change I put in there would make the tests work for a casual user, so it offers a bit more | 15:54 |
| *** raoul_ has quit IRC | 15:55 | |
| jonathanmaw | though I'll need to change the comment because it's not actually a WSL-specific problem | 15:55 |
| Kinnison | The safest thing might be to also ensure that we use a shorter path for the gitlab runner (maybe even bind a drive name to a path if you can do that) | 15:55 |
| Kinnison | But yes, the generic improvements are always good | 15:55 |
| * Kinnison ran into an issue of socket name lengths when ansibling things a while back | 15:55 | |
| *** raoul_ has joined #buildstream | 15:58 | |
| *** nimish has joined #buildstream | 16:18 | |
| jjardon | Kinnison: Is it possible to put those performance test in the CI somewhere? | 16:23 |
| jjardon | or bst have already something to detect performance regression? | 16:24 |
| Kinnison | jjardon: We have a benchmarking suite which is being developed to monitor this kind of hting | 16:25 |
| jjardon | ah cool | 16:25 |
| Kinnison | the stuff I came up with is not really suited to CI because it needs control over its environment and a human to interpret the results. | 16:25 |
| * Kinnison will be going over it all in the first session tomorrow at the gathering | 16:25 | |
| Kinnison | if you're going to be there, you'll learn more then | 16:25 |
| jjardon | unfortunately I can not be there; any idea if it will be recorded? | 16:26 |
| Kinnison | I don't know, sorry | 16:26 |
| *** nimish has quit IRC | 16:28 | |
| *** nimish has joined #buildstream | 16:28 | |
| benschubert | jonathanmaw: thanks for the wsl effort :) | 16:29 |
| *** nimish has quit IRC | 16:38 | |
| *** nimish has joined #buildstream | 16:39 | |
| *** sambishop has quit IRC | 16:45 | |
| *** nimish has quit IRC | 16:54 | |
| *** nimish has joined #buildstream | 16:54 | |
| *** nimish has joined #buildstream | 16:55 | |
| jennis | lachlan, jonathanmaw, this MR has been set for automatic merge but requires one more approval: https://gitlab.com/BuildStream/benchmarks/merge_requests/12 | 17:08 |
| jennis | jonathanmaw, I guess you're happy to approve seeing as you set it to automatic merge | 17:08 |
| jennis | ? | 17:08 |
| * jonathanmaw just mashed the button | 17:08 | |
| jennis | heh, thanks | 17:08 |
| *** nimish has quit IRC | 17:10 | |
| *** nimish has joined #buildstream | 17:10 | |
| *** toscalix has quit IRC | 17:15 | |
| *** nimish has quit IRC | 17:15 | |
| *** nimish has joined #buildstream | 17:16 | |
| abderrahim[m] | valentind: What's the status of #734? If you're not going to work on it, I'll try to do it | 17:20 |
| gitlab-br-bot | Issue #734: Cleaning the internal cache is extremely slow https://gitlab.com/BuildStream/buildstream/issues/734 | 17:20 |
| *** nimish has quit IRC | 17:20 | |
| *** nimish has joined #buildstream | 17:21 | |
| abderrahim[m] | (I interrupted a cleanup after 13 hours, and it cleaned almost nothing. It didn't even clean all the artifacts from the performance testing) | 17:22 |
| juergbi | abderrahim[m]: I suspect the slowest part is determining what should be removed, not the actual removal | 17:26 |
| juergbi | more than 13 hours is horrible in any case | 17:27 |
| abderrahim[m] | it didn't finish cleaning up 8000+ 1~2kB artifacts, it's clear it's not the removal that takes long | 17:29 |
| abderrahim[m] | it may be fixable in the same way as for remote cache | 17:29 |
| *** tpollard has quit IRC | 17:31 | |
| *** nimish has quit IRC | 17:31 | |
| *** nimish has joined #buildstream | 17:31 | |
| *** nimish has joined #buildstream | 17:32 | |
| *** phildawson has quit IRC | 17:34 | |
| *** jonathanmaw has quit IRC | 17:40 | |
| gitlab-br-bot | jennis approved MR !1113 (abderrahim/artifact-cache-junction->master: Use artifact cache specs from the parent project before those defined in junctions) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1113 | 17:42 |
| juergbi | abderrahim[m]: yes. the issue is that BuildStream currently assumes that the whole artifact is cached locally if the artifact ref exists. cas object-based expiry breaks that assumption. we need to break that assumption anyway for 'partial cas' but that's what makes it less straight forward | 17:50 |
| *** nimish has quit IRC | 17:52 | |
| *** nimish has joined #buildstream | 17:53 | |
| gitlab-br-bot | jjardon closed issue #884 (Traceback with current master (564cb2450a8657762e16c1d26d1373987dc4a6c5)) on buildstream https://gitlab.com/BuildStream/buildstream/issues/884 | 17:54 |
| gitlab-br-bot | jjardon merged MR !1114 (abderrahim/protobuf-version->master: requirements/requirements.in: update minimum versions) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1114 | 17:54 |
| *** nimish has quit IRC | 17:58 | |
| *** nimish has joined #buildstream | 17:58 | |
| *** raoul_ has quit IRC | 17:58 | |
| *** nimish has quit IRC | 18:03 | |
| *** nimish has joined #buildstream | 18:03 | |
| *** nimish has quit IRC | 18:18 | |
| *** nimish has joined #buildstream | 18:19 | |
| *** nimish has quit IRC | 18:34 | |
| *** nimish has joined #buildstream | 18:34 | |
| *** nimish has quit IRC | 18:39 | |
| *** nimish has joined #buildstream | 18:40 | |
| *** nimish has quit IRC | 18:45 | |
| *** nimish has joined #buildstream | 18:45 | |
| *** BILSTEIN has joined #buildstream | 18:46 | |
| *** BILSTEIN has left #buildstream | 18:48 | |
| *** nimish has quit IRC | 18:50 | |
| *** nimish has joined #buildstream | 18:51 | |
| *** alatiera has quit IRC | 19:07 | |
| *** nimish has quit IRC | 19:10 | |
| *** nimish has joined #buildstream | 19:11 | |
| *** nimish has joined #buildstream | 19:11 | |
| *** alatiera has joined #buildstream | 19:17 | |
| abderrahim[m] | juergbi: my idea is to (1) use object-based expiry (2) go through all the refs and remove those that are missing objects (3) possibly do a last pass removing dangling objects | 19:19 |
| abderrahim[m] | would that work? | 19:19 |
| abderrahim[m] | my understanding is that now, bst removes the refs one by one, and for each ref does a (3). Is this correct? | 19:20 |
| *** nimish has quit IRC | 19:21 | |
| *** nimish has joined #buildstream | 19:22 | |
| *** nimish has quit IRC | 19:27 | |
| *** nimish has joined #buildstream | 19:27 | |
| *** nimish has quit IRC | 19:37 | |
| *** nimish has joined #buildstream | 19:37 | |
| doras[m] | jjardon: I'm actually modifying the freedesktop-sdk vm code, with hope to get it closer to running GNOME. So far I split systemd to libsystemd and "the rest", got systemd's default daemons to all load and run, and got dbus-daemon to host the system bus. All of this got me networking, DNS resolution and session management working (systemd-networkd, systemd-resolved and systemd-logind). I'll obviously share my changes as well once I turn | 19:42 |
| doras[m] | all the random hacks into actual integration code. | 19:42 |
| doras[m] | Since I rarely have time during weekdays, I think I'll start submitting changes incrementally. Or at least in a WIP MR. | 19:44 |
| jjardon | doras[m]: simply updating to systemd 240 gave us networking | 19:51 |
| jjardon | see https://gitlab.com/freedesktop-sdk/freedesktop-sdk/issues/619 | 19:52 |
| jjardon | the first MR we were trying to add the needed users manually, but seems they are automatically added in 240 :) | 19:52 |
| *** nimish has quit IRC | 19:52 | |
| *** nimish has joined #buildstream | 19:53 | |
| jjardon | doras[m]: BTW, very cool project! please keep up posted (we basically do not work on that because lack of time) | 19:53 |
| jjardon | keep us* | 19:53 |
| jjardon | doras[m]: some of us hang in #freedesktop-sdk in freenode as well | 19:54 |
| jjardon | doras[m]: Also, there is another MR to fix another stuff, take a look to the MR attached at https://gitlab.com/freedesktop-sdk/freedesktop-sdk/issues/618 | 19:57 |
| doras[m] | jjardon: yeah, those are the two systemd bugs I mentioned yesterday. I saw they were fixed in 240 but postponed the update to fix a few other issues. I see you and valentind had made similar progress. | 20:18 |
| doras[m] | I joined #freedesktop-sdk | 20:18 |
| *** kapil___ has quit IRC | 20:41 | |
| *** nimish has quit IRC | 20:43 | |
| *** nimish has joined #buildstream | 20:43 | |
| gitlab-br-bot | jjardon opened issue #886 (Handshake timed out in the cache server) on buildstream https://gitlab.com/BuildStream/buildstream/issues/886 | 20:47 |
| juergbi | abderrahim[m]: determining reachable objects from all refs would again be expensive/slow | 20:54 |
| *** lachlan has quit IRC | 20:55 | |
| juergbi | and there would be a big race condition between removing objects and removing the corresponding refs | 20:55 |
| abderrahim[m] | isn't the cleanup job supposed to have exclusive access to the cache? | 20:56 |
| juergbi | possibly, but ideally it's not needed | 20:58 |
| juergbi | we should make sure BuildStream core can deal with partial artifacts in the local cache | 20:58 |
| juergbi | we anyway want this to support reducing network bandwidth and local resources for remote execution | 20:59 |
| juergbi | (intermediate build results don't have to be completely downloaded with remote execution) | 20:59 |
| abderrahim[m] | I'm also thinking about 1.2. If the plan of dropping 1.4 proceeds, 1.2 still needs bugfixes | 21:00 |
| jjardon | abderrahim[m]: sorry it took so long to merge https://gitlab.com/freedesktop-sdk/infrastructure/freedesktop-sdk-docker-images/merge_requests/6 | 21:01 |
| jjardon | wrong channel | 21:01 |
| juergbi | yes, finding an acceptable but not too invasive solution for 1.2 might not be straight forward | 21:02 |
| alatiera | what plan about 1.4 | 21:14 |
| abderrahim[m] | alatiera: not releasing 1.4, and going directly to 2.0 when it's ready | 21:15 |
| alatiera | oh | 21:16 |
| alatiera | any timeframe for 2.0? say a year? | 21:16 |
| juergbi | see also https://mail.gnome.org/archives/buildstream-list/2019-January/msg00010.html | 21:17 |
| juergbi | no timeframe has been decided yet | 21:18 |
| *** nimish has quit IRC | 21:18 | |
| *** nimish has joined #buildstream | 21:19 | |
| juergbi | I would definitely hope that it still happens in 2019, though, but we'll likely discuss this at the gathering this week | 21:19 |
| *** nimish has quit IRC | 21:44 | |
| *** nimish has joined #buildstream | 21:44 | |
| *** nimish has quit IRC | 21:49 | |
| *** nimish has joined #buildstream | 21:49 | |
| *** nimish has quit IRC | 21:54 | |
| *** nimish has joined #buildstream | 21:55 | |
| doras[m] | juergbi, jennis: I've added a test for https://gitlab.com/BuildStream/buildstream/merge_requests/1110/ | 21:59 |
| doras[m] | Nice CI setup, by the way. Writing tests is fairly easy as well. | 22:02 |
| juergbi | great, thanks | 22:07 |
| *** mohan43u has quit IRC | 22:23 | |
| *** mohan43u has joined #buildstream | 22:27 | |
| *** alatiera has quit IRC | 22:32 | |
| gitlab-br-bot | juergbi closed issue #857 (test_pull_missing_blob spuriously crashes) on buildstream https://gitlab.com/BuildStream/buildstream/issues/857 | 22:36 |
| gitlab-br-bot | juergbi merged MR !1095 (valentindavid/crash_in_scheduler_857->master: Fix crash when spawned job completes very fast) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1095 | 22:36 |
| doras[m] | yikes, got a linting error. | 22:40 |
| doras[m] | Didn't expect that. | 22:41 |
| doras[m] | For some reason the linter lints stuff in .egg when run with "tox -e lint". That doesn't seem right. | 22:48 |
| doras[m] | ".eggs"* | 22:50 |
| *** mohan43u has quit IRC | 22:56 | |
| *** mohan43u has joined #buildstream | 23:00 | |
| gitlab-br-bot | doraskayo opened MR !1116 (doraskayo/exclude-eggs-from-linting->master: setup.cfg: exclude .eggs/** from pycodestyle linting) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1116 | 23:25 |
| doras[m] | Quite a silly MR. | 23:27 |
| *** mohan43u has quit IRC | 23:37 | |
| *** mohan43u has joined #buildstream | 23:41 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!