IRC logs for #baserock for Monday, 2017-02-06

*** gtristan has joined #baserock06:55
*** fay has joined #baserock08:27
*** fay is now known as Guest3409908:27
*** paulwaters_ has joined #baserock09:06
*** tiagogomes has joined #baserock09:13
*** ctbruce has joined #baserock09:36
*** jonathanmaw has joined #baserock09:40
paulsherwood"Paul, you are right that there is a bit of a reality gap between the formal verification people, who tend to be hard-core mathematicians, and the grimy, bitter husks of human beings (such as you and me) that do automotive software. "09:48
*** paulwaters_ has joined #baserock10:05
*** vgrade12 has joined #baserock10:18
*** locallycompact has joined #baserock10:33
*** gtristan has quit IRC10:48
*** gtristan has joined #baserock11:02
*** franred has joined #baserock11:04
*** franred has quit IRC11:58
*** franred has joined #baserock12:11
*** inara has quit IRC12:12
*** inara has joined #baserock12:26
*** locallycompact has quit IRC12:34
*** locallycompact has joined #baserock12:40
*** franred has quit IRC12:46
*** locallycompact has quit IRC12:51
*** locallycompact has joined #baserock12:54
*** franred has joined #baserock13:01
*** paulwaters_ has joined #baserock13:11
gtristanright, so MR 300; only comment was that the commit message does not really tell me why we're setting the git user/email:
gtristanpresuming it must have a purpose, +1 :)13:39
paulsherwoodgtristan: yes, actually that MR still combines two real changes, sorry:13:40
paulsherwood- the headline change is to export all definitions as a single $target.yml file, which can also be used as an input instead of the normal set-of-morph-files-in-a directory-tree13:41
paulsherwood- in the process of doing the above, i encountered some issues with the release-note functionality, so fixed them along the way13:42
paulsherwood  - notably, release-note needs to explore some revious ref of definitions, to generate the diff between then and now13:42
paulsherwood  - so ybd needs to do git stash while exploring, then git stash pop at the end. these fail unless git username/email is configured13:43
* paulsherwood should have documented the commits better13:43
gtristannod, that makes reviews much easier :)13:48
paulsherwoodi can fix them up later today13:48
gtristanactually I dont seem to have review power on gitlab/baserock13:49
* paulsherwood notes that this is a prime example of why cache-key is better resolving to tree, rather than commit :)13:49
gtristanpaulsherwood, I'm still not seeing it13:51
gtristanI mean, I dont see where the tree sha thing is relevant or how there is an example of that13:51
paulsherwoodif i fix up the commit messages, the tree stays unchanged, but the commit history will be completely different13:52
paulsherwoodas a result your algorithm will rebuild, mine wont13:52
gtristanIf you change the commit message on definitions, yes13:53
paulsherwoodif i change the commit message in any component, including the build tool13:53
gtristanI see what you mean, that shouldnt be such a frequent, not worth crippleing cache key calculation with a network connection requirement13:54
gtristanIf you change the commit message on definitions, nothing will change13:55
gtristanif you point definitions to a new revision of something it builds, cache key will change13:55
gtristanif you change the commit message of the build tool, that should never change anything, either13:57
gtristanonly bumping the tools' internal cache key revision (meaning it's calculating things differently now also), would change keys13:57
* paulsherwood will wait and see :)14:20
*** toscalix has joined #baserock14:22
gtristanone alternative optimization, if you're worried about superficial commits in projects we build; is to have the automatic updating mechanism say "I refuse to update the commit sha to the latest on the branch, IF that commit sha points to the same tree I already have a commit sha for"14:26
paulsherwoodor just put tree into the definitions, instead of sha :)14:26
gtristanyou cant checkout a tree actually14:26
gtristanI recall checking that14:27
paulsherwoodbecause it may be non-unique, maybe. but setting a ref to a tree and checking that out would always work i think. anyway... this is probably too far into the weeds14:28
jonathanmawIs there a helper method in ybd to get the commit sha1 for a given ref? I can't find one in repos.py14:52
jonathanmawbut I'd be surprised if this is the first time ybd's needed to get the commit sha1 for a ref14:52
gtristanjonathanmaw, I think it's entirely new, the MR above adds something in so that get_tree() returns _also_ the sha14:53
gtristan(which is basically just a rev-parse)14:53
jonathanmawaha, that's handy!14:53
* paulsherwood thinks it's about time MR300 landed :)15:05
pedroalvarezNote that is a real email address15:09
pedroalvarezNot a problem of course :)15:10
pedroalvarezI don't think anybody will be sending emails to that one15:11
*** locallycompact has quit IRC16:06
*** locallycompact has joined #baserock16:19
*** locallycompact has joined #baserock16:20
*** paulwaters_ has joined #baserock16:58
*** paulwaters_ has quit IRC16:58
*** jude_ has quit IRC17:11
*** ctbruce has quit IRC17:12
*** jonathanmaw has quit IRC17:17
*** jude_ has joined #baserock17:30
*** franred has quit IRC17:35
*** tiagogomes has quit IRC17:58
*** locallycompact has quit IRC18:05
*** toscalix has quit IRC18:05
*** jude_ has quit IRC18:25
*** paulwaters_ has joined #baserock20:28
*** lachlan_ has joined #baserock20:42
*** jude_ has joined #baserock20:51
*** lachlan_ has quit IRC21:19
*** gtristan has quit IRC22:07
*** jude_ has quit IRC22:16
*** lchlan_ has joined #baserock23:27
*** edcragg_ has joined #baserock23:29
*** edcragg has quit IRC23:34
*** lchlan has quit IRC23:34
*** lchlan_ is now known as lchlan23:37

Generated by 2.15.3 by Marius Gedminas - find it at!