IRC logs for #buildstream for Monday, 2019-05-13

« Sunday, 2019-05-12 Index Tuesday, 2019-05-14 »
gitlab-br-botjjardon opened (was WIP) MR !1341 (jjardon/ostree_removal->master: Remove OSTree plugin; It lives now in the bst-plugins-experimental repo) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/134102:29
*** nimish2711 has joined #buildstream02:51
*** nimish2711 has quit IRC02:56
*** nimish2711 has joined #buildstream02:57
*** nimish2711 has quit IRC03:17
*** nimish2711 has joined #buildstream03:17
*** nimish2711 has quit IRC03:22
*** nimish2711 has joined #buildstream03:23
*** nimish2711 has quit IRC03:28
*** nimish2711 has joined #buildstream03:28
*** nimish2711 has quit IRC03:33
*** nimish2711 has joined #buildstream03:33
*** nimish2711 has quit IRC03:38
*** nimish2711 has joined #buildstream03:39
*** nimish2711 has quit IRC03:44
*** nimish2711 has joined #buildstream03:44
*** nimish2711 has quit IRC04:09
*** nimish2711 has joined #buildstream04:09
*** nimish2711 has quit IRC04:34
*** nimish2711 has joined #buildstream04:35
*** nimish2711 has quit IRC04:55
*** dftxbs3e has quit IRC04:59
*** nimish2711 has joined #buildstream05:29
*** nimish2711 has quit IRC05:39
*** nimish2711 has joined #buildstream05:39
*** nimish2711 has quit IRC05:42
*** dftxbs3e has joined #buildstream05:46
*** nimish2711 has joined #buildstream07:10
*** nimish2711 has quit IRC07:15
*** nimish2711 has joined #buildstream07:15
*** tristan has joined #buildstream07:27
*** nimish2711 has quit IRC07:27
*** bochecha has joined #buildstream07:40
*** toscalix has joined #buildstream07:48
gitlab-br-bottristanvb closed issue #1019 (Buildstream fails to spot bad project.conf) on buildstream https://gitlab.com/BuildStream/buildstream/issues/101908:05
*** rdale has joined #buildstream08:10
*** benschubert has joined #buildstream08:14
*** nimish2711 has joined #buildstream08:30
Kinnisonjennis: is today's report really a day late?08:34
Kinnisonjennis: also is there any obvious fingerpointing to be done for the drop in performance?08:35
*** nimish2711 has quit IRC08:40
*** nimish2711 has joined #buildstream08:40
juergbi!1332 might have had an impact08:41
gitlab-br-botMR !1332: Fix build track all no strict https://gitlab.com/BuildStream/buildstream/merge_requests/133208:41
KinnisonI'd not expect that to impact things too hard unless the cache key changes, and for the benchmarking I thought we'd fully resolved the refs etc so cache keys ought to be stable from the get-go08:43
KinnisonOr is the implication that the cache keys were never right first time through?08:44
juergbinot about cache key change, rather about initial cache key calculation now triggering reverse dependency updates as well08:46
*** raoul has joined #buildstream08:46
juergbidon't know whether that had a significant impact08:46
KinnisonThe cost on pre-scheduler (show) time was pretty low, the impact was more in build time so it's not initial calculation08:51
*** nimish2711 has quit IRC08:55
*** nimish2711 has joined #buildstream08:55
raoulLooks like marge broke on the inside after I passed over an MR to her on Friday...08:56
Kinnisonpoor marge08:59
jennisKinnison, it's not a day late, just forgot to remove that08:59
jennisand from the graphs, it looks like Tristan's workspaced artifacts MR made the build times a bit slower08:59
jennisBut I haven't looked at the MR to confirm09:00
*** jonathanmaw has joined #buildstream09:00
benschubertyep it's the fix for workspaced elements, I didn't expect it to be that bad09:01
benschubertIf someone sees an obvious way of separating completely the workspaced workflow from the non workspaced one...09:02
KinnisonI think the work jennis and jonathanmaw were doing was aiming to do that eventually09:02
*** tpollard has joined #buildstream09:02
benschubertperfect09:03
KinnisonNote, I *think*09:03
* Kinnison is known to be wrong from time to time09:03
Kinnison:D09:03
jennisThe original implementation was doing that quite explicitly, however, things have changed a lot now, jonathanmaw should have a better idea09:05
jonathanmawmy tentative implementation that split the workspaced and non-workspaced codepath completely involved a lot of duplicated work, so was a degradation in maintainability, so I had to take it back to figure out a more maintainable approach09:07
*** nimish2711 has quit IRC09:25
*** nimish2711 has joined #buildstream09:26
*** lachlan has joined #buildstream09:34
*** ChanServ sets mode: +o tristan09:34
tristanabderrahim[m], Around ?09:34
abderrahim[m]yes09:35
tristanAh good timing09:35
tristanabderrahim[m], So I'd like to discuss a potential 1.4, since you have expressed interest in that09:36
tristanabderrahim[m], From my side, (A): I don't think I can justify a lot of my own effort towards backporting things... (B): I don't want to backport any of the major rearchitecturing things at all09:37
tristanabderrahim[m], we could however start by defining what we think it should be (personally I am mostly interested in an option to control 'max-jobs' from user config / maybe CLI option)09:38
tristanAnd from there we could milestone some things, and if people have time to work on them, a 1.4 could materialize09:38
*** lachlan has quit IRC09:42
tristanabderrahim[m], I am thinking of 2 categories of candidate for backports to 1.4... As you suggested, it would be good to backport some of the CLI changes and adding deprecation warnings to the commands which will change in 2.0 (i.e. deprecate, not obsolete, the existing commands should not be disabled in 1.x)09:42
abderrahim[m]and I'd like to have easier porting to bst 2, so I'd like the new UX and new APIs such as virtual directory and whatever is the final API for the new world order09:42
abderrahim[m]And there are the nifty small features I'd be content with having in 1.2 but I guess you don't want new features in a stable release09:42
tristanMaybe that would mean that we could have at least the new `bst artifact log` and `bst artifact delete`, those are simple enough features09:42
abderrahim[m]tristan: off-topic, but I'm going to send a GNOME OS talk to guadec. Do you want to do something together or is it okay to do it by myself?09:42
abderrahim[m]of course09:42
abderrahim[m]the goal is to ease transition, not to break people's workflow09:42
tristanabderrahim[m], Ah seems all  your messages came in at once :)09:43
tristanabderrahim[m], I have submitted a talk which is about describing how GNOME integrates with the OS, maybe that is related, maybe we could do that or a GNOME OS talk together instead09:44
abderrahim[m]I'd also like to discuss small incompatibilities one by one, whether to allow them or not09:44
tristanabderrahim[m], Ok so lets not switch topics just yet ... :)09:44
tristanabderrahim[m], 1.x is `bst 1`, that means API stable, which means no incompatibilities are allowed at all09:45
abderrahim[m]ok, let's finish the bst 1.4 topic first09:45
tristan1.2.x gets no new features, and 1.4 can have features (we never planned to make a 1.4 but there are some justifications now)09:45
abderrahim[m]well, there were between 1.0 and 1.209:45
tristanabderrahim[m], if there were, they were purely accidental and should never have happened09:46
abderrahim[m]let me try to compile a list of things I'd like to have, and let's see together09:46
tristanabderrahim[m], for instance... people did really CRAZY things like referring to out-of-project files by sneaking `../../../` paths into their projects09:46
*** lachlan has joined #buildstream09:46
tristanAnd we didnt have an error message for it09:46
tristanThat is not really an incompatibility, but a missing roadblock for API abuses09:46
tristanAnyway, we are not perfect, but that is how API works, stable never changes unless mistakes are made, and we try our best to not make mistakes09:47
abderrahim[m]tristan: would !1140 be considered in the same way?09:47
gitlab-br-botMR !1140: Do not resolve or mangle symlinks during staging https://gitlab.com/BuildStream/buildstream/merge_requests/114009:47
tristanif we change API, we bump the major point version and make the new thing parallel installable09:48
abderrahim[m]brb09:48
tristanabderrahim[m], I don't know, that one look like it might be a breaking change, maybe it needs to be "buy-in" semantically somehow in that case09:49
*** lachlan has quit IRC09:49
tristanabderrahim[m], The *other* category of thing that I would think would be helpful to have in 1.4, would be some of the convenience in how dependencies are expressed in YAML (i.e. 2.x now has "build-depends" and "runtime-depends", and allows cross junction elements to be expressed as "junction.bst:element.bst")09:50
tristanabderrahim[m], That said, we don't want to diverge from 2.x, any feature additions we backport, we should have a high level of confidence that they will be the same in 2.x09:50
tristan(I think that applies to the CLI changes and YAML conveniences, though)09:51
*** lachlan has joined #buildstream09:55
abderrahim[m]tristan: the {build,runtime}-depends junction.bst:element.bst are the reason I started this discussion09:55
*** nimish2711 has quit IRC09:56
*** nimish2711 has joined #buildstream09:56
tristanabderrahim[m], Ah interesting, sure yes those are dead easy to backport and would be nice to have in 1.409:57
abderrahim[m]so dead easy to backport I wished to add them in a future 1.2.x :p09:58
abderrahim[m]anyway, I'm making a list09:58
*** lachlan has quit IRC10:01
*** WSalmon__ has joined #buildstream10:01
*** nimish2711 has quit IRC10:01
*** nimish2711 has joined #buildstream10:01
*** phil has joined #buildstream10:01
*** CTtpollard has joined #buildstream10:02
*** tpollard has quit IRC10:03
*** phildawson_ has quit IRC10:03
*** phil has quit IRC10:06
*** phil has joined #buildstream10:06
*** nimish2711 has quit IRC10:06
*** nimish2711 has joined #buildstream10:07
*** WSalmon__ has quit IRC10:10
*** cs-shadow has joined #buildstream10:12
*** phil has quit IRC10:17
*** phil has joined #buildstream10:17
*** nimish2711 has quit IRC10:52
*** nimish2711 has joined #buildstream10:52
*** phil has quit IRC11:01
*** nimish2711 has quit IRC11:02
*** nimish2711 has joined #buildstream11:02
*** phildawson has joined #buildstream11:04
*** nimish2711 has quit IRC11:12
*** nimish2711 has joined #buildstream11:13
*** lachlan has joined #buildstream11:16
*** nimish2711 has quit IRC11:23
*** nimish2711 has joined #buildstream11:23
*** lachlan has quit IRC11:25
*** nimish2711 has quit IRC11:33
*** nimish2711 has joined #buildstream11:33
*** lachlan has joined #buildstream11:42
*** toscalix has quit IRC11:43
*** lachlan has quit IRC11:48
*** lachlan has joined #buildstream11:49
*** lachlan has quit IRC11:55
*** albfan has quit IRC12:06
*** lachlan has joined #buildstream12:13
*** lachlan has quit IRC12:20
*** tristan has quit IRC12:21
*** nimish2711 has quit IRC12:33
*** nimish2711 has joined #buildstream12:34
*** jjardon has quit IRC12:57
*** benschubert has quit IRC12:58
*** benschubert has joined #buildstream12:58
*** jjardon has joined #buildstream12:58
*** ChanServ sets mode: +o jjardon12:58
*** nimish2711 has quit IRC13:09
*** nimish2711 has joined #buildstream13:09
*** nimish2711 has quit IRC13:14
*** nimish2711 has joined #buildstream13:14
*** CTtpollard has quit IRC13:30
*** CTtpollard has joined #buildstream13:30
*** CTtpollard is now known as tpollard13:30
*** nimish2711 has quit IRC14:04
*** nimish2711 has joined #buildstream14:05
*** lachlan has joined #buildstream14:05
*** lachlan has quit IRC14:12
*** lachlan has joined #buildstream14:29
*** tristan has joined #buildstream14:34
*** nimish2711 has quit IRC14:35
*** nimish2711 has joined #buildstream14:35
*** lachlan has quit IRC14:39
*** nimish2711 has quit IRC14:50
*** nimish2711 has joined #buildstream14:50
gitlab-br-botjonathanmaw opened MR !1342 (shared/split_out_update_state_2->master: Split Element._update_state into cache key and artifact state methods) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/134214:52
jonathanmawtristan: I've created !1342 to split up _update_state(). It's the changes at https://gitlab.com/BuildStream/buildstream/commit/ef117177d90fe8e74102ee22133f8188c878bf67 with some extra comments added.14:56
*** nimish2711 has quit IRC15:11
*** lachlan has joined #buildstream15:14
*** nimish2711 has joined #buildstream15:22
*** lachlan has quit IRC15:22
*** bochecha has quit IRC15:39
*** lachlan has joined #buildstream16:02
*** lachlan has quit IRC16:09
*** lachlan has joined #buildstream16:09
*** lachlan has quit IRC16:12
*** lachlan has joined #buildstream16:29
*** lachlan has quit IRC16:37
*** lachlan has joined #buildstream16:57
*** nimish2711 has quit IRC17:02
*** raoul has quit IRC17:05
*** lachlan has quit IRC17:07
*** lachlan has joined #buildstream17:25
*** lachlan has quit IRC17:28
*** jonathanmaw has quit IRC17:29
*** lachlan has joined #buildstream17:45
*** jonathanmaw has joined #buildstream17:48
*** jonathanmaw has quit IRC17:50
*** lachlan has quit IRC18:13
*** nimish2711 has joined #buildstream18:34
*** nimish2711 has quit IRC18:39
*** nimish2711 has joined #buildstream18:40
*** nimish2711 has quit IRC18:41
*** nimish2711 has joined #buildstream18:43
*** lachlan has joined #buildstream18:56
*** lachlan has quit IRC19:10
*** lachlan has joined #buildstream19:49
*** lachlan has quit IRC19:53
« Sunday, 2019-05-12 Index Tuesday, 2019-05-14 »

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