| *** vmeson [~quassel@128.224.252.2] has quit [Ping timeout: 260 seconds] | 06:15 | |
| *** vmeson [~quassel@128.224.252.2] has joined #baserock | 06:25 | |
| *** straycat [~straycat@vortis.xen.tardis.ed.ac.uk] has quit [] | 10:21 | |
| richard_maw | paulsherwood: you're on to talk :) | 12:14 |
|---|---|---|
| paulsherwood | :) | 16:46 |
| richard_maw | Are you sure it's not building the branches where you have a detached HEAD? I've had a look at the code, and it should work, but I haven't gotten around to deep testing of that yet | 16:49 |
| paulsherwood | i haven't checked that in the last few weeks, but i can do so now... | 16:57 |
| paulsherwood | i'll try morph master | 16:57 |
| richard_maw | the current behaviour is older than that | 16:57 |
| paulsherwood | yes | 17:00 |
| richard_maw | yes it's still not building from detached HEADs? | 17:01 |
| richard_maw | hm, it must be more complicated than I thought then. | 17:02 |
| paulsherwood | richard_maw: i'm not 100% sure. checking now | 17:04 |
| paulsherwood | actually, i fear it's weirder than i said in my email | 17:11 |
| paulsherwood | (but it may be me, of course) | 17:11 |
| paulsherwood | http://fpaste.org/124500/40769087/ | 17:15 |
| paulsherwood | first thing i notice is that [Build 484/554] [linux-x86-64-generic-devel] Starting actual build: linux-x86-64-generic-devel 98007cb | 17:16 |
| paulsherwood | does not correspond with | 17:16 |
| paulsherwood | /src2/ws-10aug/foo/upstream/linux # git log | 17:16 |
| paulsherwood | commit d67a0e110187abd560a1de63fa172894a52839d5 | 17:16 |
| paulsherwood | but then when i check out a detached head, build says ERROR: Ref foo is an invalid reference for repo git://git.baserock.org/delta/linux | 17:18 |
| paulsherwood | which is in line with my broad understanding - morph is trying to build the current branch, even though i'm not on it | 17:18 |
| paulsherwood | s/current branch/system branch/ | 17:21 |
| richard_maw | hm, can you try building it with --verbose? | 17:26 |
| richard_maw | AHA! I think I found the problem | 17:30 |
| paulsherwood | http://fpaste.org/124507/40769186/ | 17:30 |
| * paulsherwood cheers! | 17:31 | |
| richard_maw | it's made the temporary build branch correctly, so it includes the contents, but it checks HEAD against the value in the morphology, and doesn't include it if they differ | 17:31 |
| paulsherwood | richard_maw: reading your email just now, you have a patch to make it remove the existing branch and re-create it with | 17:36 |
| paulsherwood | the right commit | 17:36 |
| paulsherwood | ? | 17:36 |
| paulsherwood | wouldn't that be dangerous? the branch may mean something to someone else | 17:37 |
| richard_maw | I don't think so. This is done by `morph edit`, which is the one to put the repository there in the first place. Nobody should have had time to put the branch in yet. | 17:38 |
| paulsherwood | i mean 'morph checkout b:b/definitions master; morph edit <chunk>' then accidentally push to master? | 17:46 |
| richard_maw | paulsherwood: my current thought for fixing your "checkout another branch" use-case is to have morph edit set "morph.chunk" in the config when it clones the chunk repository, and check that against the "name" field of the chunks list in the stratum | 17:46 |
| richard_maw | paulsherwood: ah, well "master" isn't going to be a problem for most repos, as you won't be allowed to push to most of those beacuse of gitano's rules | 17:47 |
| paulsherwood | ok s/master/any name for an existing tag or branch/ | 17:48 |
| richard_maw | yeah, in that case you need either a policy that the branch of a chunk repository is the same as the system branch, so you should name your system branches such that there won't be conflicts | 17:49 |
| paulsherwood | but actually, maybe i'm not using gitano - maybe i checked out something from github | 17:49 |
| richard_maw | or have morph edit fabricate a chunk repository branch name that is prefixed by your system branch's name | 17:49 |
| paulsherwood | what's wrong with my suggestion of just getting the chunk repo, in the right state in the directory, and thereafter building whatever the user does in that directory? | 17:50 |
| paulsherwood | (ie no need to create a branch at all, until morp publish) | 17:51 |
| * paulsherwood knows morph publish doesn't exist :) | 17:51 | |
| richard_maw | currently edit conflates making the repository available, setting up a branch that you may push and making sure you can build your local changes. I take it your suggestion is to cut the making of branches out? | 17:52 |
| paulsherwood | yes | 17:53 |
| paulsherwood | as a user i just want to be building my changes | 17:53 |
| richard_maw | that doesn't prevent the pushing issue unless `morph edit` leaves you with a detached HEAD or something | 17:54 |
| paulsherwood | i always have to create a different branch when i decide i'm ready to push in any case | 17:54 |
| paulsherwood | true. but i still think it's better, since the chunk branch i'm on is the chunk branch that would be pushed, rather than a branch that happens to be based on my system branch name | 17:56 |
| richard_maw | you would need to update your morphologies to point to the branch you made when you come to push your system branch though | 17:58 |
| richard_maw | but I suppose those are details for morph publish | 17:58 |
| paulsherwood | yes, that's currently my preference - but i appreciate you arguing it out :) | 17:58 |
| paulsherwood | this stuff is quite complicated :-) | 17:59 |
| richard_maw | it's software, we live in a world of constantly moving parts | 17:59 |
| paulsherwood | imagine how dull it would be if nothing moved | 18:00 |
| richard_maw | well, it's not boring, which is why I opted to work on it a bit this weekend | 18:04 |
| paulsherwood | :) | 18:05 |
| richard_maw | but I've just discovered my branch for avoiding making temporary branches has broken some old tests, and I need to work out whether the tests are bad and doing something weird, or if I need to rework that | 18:09 |
| richard_maw | and it's stopped being fun, but would be a waste of time if it doesn't work | 18:10 |
| * richard_maw leaves to eat something | 18:11 | |
| * paulsherwood does the same | 18:13 | |
| richard_maw | bah, the test failures for those old tests were because I lost a patch somewhere, I've got an extra patch to add to my branch for avoiding creating temporary build branches when they aren't needed, then there's a fix for number 1 of your workflow issues, and making `morph show-branch-root` useful | 18:49 |
| Kinnison | You lot do much more useful stuff during your w/e than I do | 18:55 |
| richard_maw | 1. vacuum floor 2. laundry 3. opted out of the open register https://secure.manchester.gov.uk/forms/form/960/en/opt_out_of_or_in_to_the_open_register 4. shopping 5. Baserock | 19:05 |
| richard_maw | I guess so | 19:05 |
| Kinnison | painted pub, mopped behind bar, played minecraft, decided to buy a corn snake, researched corn snakes, decided to buy after being in the US | 19:05 |
| richard_maw | I still need to get around to doing something useful with my Microserver and do the thing so I can access machines on my home network via my microserver | 19:07 |
| richard_maw | why are you buying a Snake? | 19:07 |
| Kinnison | They're cool | 19:07 |
| richard_maw | Kinnison: https://www.youtube.com/watch?v=Bxc_55ur-J4 ? | 19:08 |
| Kinnison | yeah, so that is larger than a corn snake | 19:09 |
| Kinnison | corn snakes grow to maybe 1.5m or so in length | 19:10 |
| Kinnison | and last ca 15-20 years | 19:10 |
| richard_maw | well, that's them patches away | 19:31 |
| richard_maw | night all | 19:32 |
| Kinnison | :-) | 19:34 |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!