*** xjuan has quit IRC | 01:41 | |
*** tristan has quit IRC | 02:24 | |
*** tristan has joined #buildstream | 05:11 | |
*** ChanServ sets mode: +o tristan | 05:11 | |
tristan | Great, crazy job pickles are broken | 05:11 |
---|---|---|
tristan | https://gitlab.com/BuildStream/buildstream/-/jobs/589968432, by branch doesnt break it, it adds tests which reveal that pickled jobs are insane | 05:12 |
* tristan doesnt understand `AttributeError: 'ElementFactory' object has no attribute '_types'`, clearly according to pluginfactory.py, that exists; __getstate() wipes it clean and makes it an empty dict, but doesn't cause it to not exist | 05:13 | |
*** tristan has quit IRC | 05:52 | |
*** hasebastian has joined #buildstream | 07:31 | |
*** traveltissues has joined #buildstream | 07:33 | |
*** cphang has joined #buildstream | 08:15 | |
*** tristan has joined #buildstream | 08:38 | |
*** ChanServ sets mode: +o tristan | 08:38 | |
*** santi has joined #buildstream | 08:54 | |
*** phildawson has quit IRC | 09:24 | |
*** phildawson has joined #buildstream | 09:29 | |
tristan | juergbi, do you prefer to have both Project.get_default_target() and also Project.invoked_from_workspace_element() ? | 09:31 |
* tristan sees that the latter was added a long while back by jonathanmaw, and that juergbi added the former afterwards, and wonders why we have both | 09:32 | |
tristan | I suppose the cli and the stream are using this for completely different purposes | 09:33 |
*** phildawson has quit IRC | 09:34 | |
tristan | oddly, it looks like the only justification for this extra stuff, is to give the user a message informing them that they can no longer run commands in the directory where a closed workspace once was (while closing it) | 09:36 |
tristan | https://gitlab.com/BuildStream/buildstream/-/blob/master/src/buildstream/_frontend/cli.py#L1079 | 09:37 |
tristan | I think that is a very silly feature to have :-S | 09:38 |
tristan | when I type "rm file.txt", I don't get a message saying "Removed file.txt, therefore you can no longer access it" | 09:38 |
juergbi | tristan: get_default_target() is used in more places | 09:38 |
tristan | yeah, that appears to be the right API | 09:39 |
juergbi | ah, you were talking about the other one | 09:39 |
tristan | Also it seems like a sensible contract, the cli doesnt need to know *why* the target is default | 09:39 |
juergbi | exactly, I'd prefer keeping get_default_target() | 09:40 |
tristan | I'd like to remove Project.invoked_from_workspace_element(), Stream.workspace_is_required() (what makes this workspace "required" anyway ?) and the silly echo() statement in cli.py | 09:40 |
juergbi | I'm ambivalent on that hint to the user | 09:40 |
juergbi | it is somewhat of special situation that you can issue a bst command that makes it impossible to issue any further bst commands for the same project (while in the same directory) | 09:42 |
juergbi | however, if the user thinks about what they're doing, it shouldn't be a surprise at all | 09:42 |
juergbi | so I don't mind removing it | 09:43 |
tristan | That does make the assumption that the user was actually in that directory also, rather than typing `bst -C directory...` | 09:43 |
juergbi | yes but bst -C into the workspace directory seems rather odd | 09:43 |
tristan | I'm not sure it does | 09:44 |
juergbi | the point of allowing bst commands from the workspace directory is convenience to avoid the need to `cd` or `-C` | 09:44 |
juergbi | well, it's also a shortcut to deal with the workspace element but still | 09:44 |
tristan | I mean, I first of all think the expectation that bst commands work in a workspace at all is odd, but once it does work, I'm not sure -C is far fetched | 09:45 |
* tristan uses -C frequently enough, but mostly from the perspective of running tests on projects in buildstream/tests/ after `tox -e venv /bin/bash` | 09:46 | |
*** hasebastian has quit IRC | 09:48 | |
tristan | juergbi, interestingly, if you follow the usage pattern which I recall sander and cs-shadow thought was normal (but I thought was strange), then I think commands would even keep working | 09:51 |
tristan | i.e. if you open the workspace as a subdirectory of the project itself, bst searches higher level directories for the project.conf | 09:52 |
tristan | Anyway, I'll throw a patch up and try to get back into these confusing duplicates | 09:52 |
tristan | I'm still unclear on how exactly I should manage these data structures | 09:53 |
* tristan throws up https://gitlab.com/BuildStream/buildstream/-/merge_requests/1963 | 10:05 | |
tristan | and adds some pings | 10:05 |
*** tristan has quit IRC | 11:40 | |
*** tristan has joined #buildstream | 11:42 | |
*** ChanServ sets mode: +o tristan | 11:42 | |
WSalmon | now that https://gitlab.com/BuildStream/buildstream/-/issues/1318 is merged can we have a dev tag? | 11:44 |
WSalmon | is it just as simple as pushing one? i think i can do that, or do you have to do some other magic too? | 11:45 |
WSalmon | also as 1.93.3 needs a diffrent project.conf to 1.93.3 i am confused as to why we have not moved to 1.95.1? | 11:46 |
WSalmon | also as 1.93.3 needs a diffrent project.conf to 1.93.2/1 i am confused as to why we have not moved to 1.95.1? | 11:46 |
WSalmon | tristan, etc ^ | 11:47 |
*** toscalix has joined #buildstream | 12:05 | |
*** Trevinho has quit IRC | 12:31 | |
*** Trevinho has joined #buildstream | 12:37 | |
*** Trevinho has quit IRC | 12:38 | |
*** Trevinho has joined #buildstream | 12:42 | |
*** xjuan has joined #buildstream | 14:04 | |
*** toscalix has quit IRC | 14:57 | |
*** xjuan has quit IRC | 17:20 | |
*** toscalix has joined #buildstream | 17:52 | |
*** toscalix has quit IRC | 18:00 | |
*** hasebastian has joined #buildstream | 18:26 | |
*** hasebastian has quit IRC | 18:43 | |
*** santi has quit IRC | 19:09 | |
*** toscalix has joined #buildstream | 20:07 | |
*** toscalix has quit IRC | 20:19 |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!