IRC logs for #buildstream for Monday, 2019-01-07

« Sunday, 2019-01-06 Index Tuesday, 2019-01-08 »
*** tristan has joined #buildstream01:52
*** tristan has quit IRC01:58
*** tristan has joined #buildstream01:59
*** tristan has quit IRC02:11
*** tristan has joined #buildstream02:18
*** alatiera has joined #buildstream05:51
*** tristan has quit IRC06:58
*** alatiera has quit IRC07:04
*** alatiera has joined #buildstream07:44
*** toscalix has joined #buildstream08:18
*** alatiera has quit IRC08:36
*** mohan43u has quit IRC09:04
*** WSalmon_ has joined #buildstream09:16
*** benschubert has joined #buildstream09:25
*** rdale has joined #buildstream09:26
*** kapil___ has joined #buildstream10:01
*** raoul has joined #buildstream10:04
*** tpollard has joined #buildstream10:05
*** jonathanmaw has joined #buildstream10:34
*** bochecha has joined #buildstream10:54
gitlab-br-botphildawson opened MR !1041 (phil/remove-source-bundle-reference->master: element.py: remove reference to source bundle command) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/104111:24
phildawson_^ one liner documentation change for review11:24
*** alatiera has joined #buildstream11:36
*** alatiera has joined #buildstream11:44
juergbijmac: commented on !1023 again. however, my question whether it's still needed before moving to a pure CAS workflow still stands12:06
gitlab-br-botMR !1023: Stop resolving symlinks in _relative_symlink_target https://gitlab.com/BuildStream/buildstream/merge_requests/102312:06
juergbii.e., if we agree that this becomes irrelevant in the future (pure CAS, no symlink conversion), do we really want/need to touch the current code path?12:08
jmacI agree it becomes irrelevant in the future, but we cannot do !991 without it12:18
gitlab-br-botMR !991: Cache artifacts with virtual directories instead of filesystem. https://gitlab.com/BuildStream/buildstream/merge_requests/99112:18
*** mohan43u has joined #buildstream12:30
*** raoul has quit IRC13:17
*** mohan43u has quit IRC13:18
gitlab-br-botjonathanmaw opened MR !1042 (jonathan/test-missing-workspace-guessing->master: Add tests to cover reinstated support for guessing targets) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/104213:27
gitlab-br-botvalentindavid closed MR !995 (valentindavid/overnight_build_arch->master: .gitlab-ci.yml: Fix arch options for overnight freedesktop-sdk build) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/99513:37
*** raoul has joined #buildstream13:52
*** raoul has quit IRC14:02
*** phildawson_ has quit IRC14:02
*** phildawson_ has joined #buildstream14:02
*** kapil___ has quit IRC14:50
*** raoul has joined #buildstream14:55
*** nimish has joined #buildstream14:58
*** nimish has quit IRC15:03
*** nimish has joined #buildstream15:03
*** tristan has joined #buildstream15:06
jennisCould someone add me as a developer to the benchmarking repo? pls15:40
jennisoh nvm, already there, probably something my end15:41
ikerperezhi, I am running a script like root (it needs to be run like root) that executes buildstream, the script finds the bst command but it can't find the buildstream module, how can I fix that? Where is the buildstream module stored?15:46
*** raoul_ has joined #buildstream15:53
*** raoul has quit IRC15:54
*** raoul__ has joined #buildstream15:56
*** raoul_ has quit IRC15:57
gitlab-br-botjonathanmaw opened issue #846 (Elements with an open workspace cannot be fetched) on buildstream https://gitlab.com/BuildStream/buildstream/issues/84616:01
*** raoul has joined #buildstream16:17
*** bochecha has quit IRC16:17
*** raoul__ has quit IRC16:18
*** nimish has quit IRC16:18
*** nimish has joined #buildstream16:18
*** tristan has quit IRC16:28
tpollardis it expected that we can't run the tests outside of tox?16:32
tpollardor am I just having issues locally16:32
tpollardas I can't run the integration tests standalone anymore16:33
*** tristan has joined #buildstream16:40
*** ChanServ sets mode: +o tristan16:41
gitlab-br-botphildawson opened issue #847 (Expose testing api for use by external plugins) on buildstream https://gitlab.com/BuildStream/buildstream/issues/84716:46
tpollardlooks like it might just be me16:47
*** nimish has quit IRC16:53
*** nimish has joined #buildstream16:54
gitlab-br-botphildawson opened issue #848 (Create api for plugins to give deprecation warnings) on buildstream https://gitlab.com/BuildStream/buildstream/issues/84816:55
*** nimish has quit IRC17:01
*** nimish has joined #buildstream17:02
benschuberttpollard: it should run locally still17:02
benschubertwhat problem are you seeing?17:02
*** nimish has joined #buildstream17:02
*** nimish has joined #buildstream17:03
tpollard'AttributeError: 'Function' object has no attribute 'get_marker'' on near enough every tests, I've tried a 'pip3 install --user -e .' again too17:03
tpollardwork fine under tox17:04
*** nimish_ has joined #buildstream17:05
benschubertCan you paste the complete traceback somewhere? :)17:05
*** nimish has quit IRC17:06
*** nimish_ is now known as nimish17:06
tpollardsure17:08
*** tristan has quit IRC17:09
tpollardbenschubert: that's the snipped https://paste.gnome.org/ppljhsxin17:09
*** Kinnison has joined #buildstream17:10
tpollardfrom './setup.py test --addopts '--integration''17:10
benschubertNever seen that :/ Seems unlikely this comes from tox though :)17:10
tpollardnp!17:12
*** xjuan has joined #buildstream17:12
*** tristan has joined #buildstream17:23
*** ChanServ sets mode: +o tristan17:23
*** nimish has quit IRC17:25
jmacikerperez: It's a bit difficult to say as I usually install in user mode, but I'd expect the modules to be in one of the 'site-packages' directories17:25
*** nimish has joined #buildstream17:25
jmacikerperez: Either in /usr/lib/python3XX/site-packages or /root/.local/lib/python3XX/site-packages17:25
jmacDepending on how it was installed17:26
gitlab-br-bottristanvb opened MR !1043 (tristan/keyboard-interrupt-stack-trace->master: Fix stack traces discovered with ^C forceful termination.) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/104317:30
gitlab-br-botraoul.hidalgocharman opened (was WIP) MR !1013 (raoul/802-refactor-artifactcache->master: Raoul/802 refactor artifactcache) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/101317:31
benschuberttristan: I was looking at the Pipeline, and was wondering. What is the reason behind creating a new Process for every job? Couldn't we reuse them?17:32
tristanbenschubert, reuse the data model; which is very expensive to load17:34
tristanbenschubert, the processes are fork() without execve()17:34
benschubertOk, but wouldn't it be faster to spawn processes and then send them jobs by a queue? If I recall correclty, we are not modifying the datamodel in the jobs right?17:35
benschubertI'm just asking since on my computer, this leads to a horrendus amount of exceptions logged because of too short-lived processes17:36
benschubert(a "bug" (or is it a feature?) in asyncio)17:36
KinnisonWe'd have to replicate model updates to each subprocess if they were to live longer and run more than one job17:36
* Kinnison is surprised those asyncio signalfd things haven't been silenced upstream yet17:37
benschubertjust to make sure we are on the same page, when you mean model updates, you mean the elements correct?17:38
Kinnison]yes17:38
KinnisonWell, I do :-)17:38
benschubertand that shouldn't be needed if we have self contained jobs, correct?17:39
KinnisonIf a job doesn't need to know any of the data model beyond what is packaged up for it, then perhaps not17:39
Kinnisonme will let you and tristan discuss this though, I have to run away now,17:39
ikerperezjmac thanks!17:39
benschubertKinnison: perfect thanks! I was wondering if something else was behind it or not :) I've been looking into optimisations a bit17:40
tristanbenschubert, the plugin code is allowed to read the data model17:53
tristanbenschubert, a plausible change would be to use threads instead17:54
tristanbut right now it's processes17:54
benschubertthreads were problematic with grpc correct? And the plugin code shouldn't be modifying the code during the run right?17:54
*** nimish has quit IRC17:55
*** nimish has joined #buildstream17:56
gitlab-br-bottristanvb merged MR !1043 (tristan/keyboard-interrupt-stack-trace->master: Fix stack traces discovered with ^C forceful termination.) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/104318:00
*** nimish has quit IRC18:03
*** toscalix has quit IRC18:04
*** mohan43u has joined #buildstream18:08
*** xjuan has quit IRC18:18
*** raoul has quit IRC18:19
*** kapil___ has joined #buildstream18:22
*** tristan has quit IRC18:23
*** xjuan has joined #buildstream18:23
*** jonathanmaw has quit IRC18:38
gitlab-br-botgokcennurlu opened issue #849 (BuildStream installs using the git repo without tags, but doesn't work) on buildstream https://gitlab.com/BuildStream/buildstream/issues/84918:42
gitlab-br-bottristanvb closed issue #753 (Cache size logging is very spammy) on buildstream https://gitlab.com/BuildStream/buildstream/issues/75318:50
gitlab-br-bottristanvb merged MR !1040 (tristan/one-cache-size-job->master: Only queue one cache size job) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/104018:50
*** dbscoach has joined #buildstream19:12
*** dbscoach has joined #buildstream19:13
*** dbscoach has quit IRC19:24
*** alatiera has quit IRC20:22
*** kapil___ has quit IRC20:32
*** mohan43u has quit IRC21:05
*** mohan43u has joined #buildstream21:09
*** mohan43u has quit IRC21:13
*** mohan43u has joined #buildstream21:16
*** mohan43u has quit IRC21:47
« Sunday, 2019-01-06 Index Tuesday, 2019-01-08 »

Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!