github.com/moarvm/moarvm | IRC logs at colabti.org/irclogger/irclogger_logs/moarvm
Set by AlexDaniel on 12 June 2018.
MasterDuke how about saturday brunch time, anybody up for some git submoduling? 11:17
nine has never done anything with submodules 11:30
MasterDuke gonna be afk for a while, but i'll leave an explanation of what i'm trying to do 12:40
in github.com/MoarVM/MoarVM/pull/1221 i've switched our libtommath submodule to the upstream v1.2.0 tag
(and added a bunch of commits to moarvm) 12:41
MasterDuke however, to build with clang or msvc and no optimization like our CI environments do (Travis clang jobs and Appveyor), i had to add a commit on top of the v1.2.0 tag 12:43
now tags are immutable, so i'm not sure how/where to then push that commit or how to have our submodule reference it 12:46
i did make a commit in 3rdparty/libtommath, but when i tried to push it git complained i wasn't on a branch, so i then did `git push origin v1.2.0`, which succeeded, but i don't know what it actually did 12:48
the commit id i have locally can't be browsed to in github.com/MoarVM/libtommath 12:49
if i `git checkout <that_id>` in 3rdparty/libtommath, a `git status` back in the moarvm root says `Your branch is ahead of 'origin/use_libtommath_v1.2.0_for_real_maybe' by 1 commit. (use "git push" to publish your local commits)` 12:51
however, if i push, it errors with `fatal: src refspec 'refs/heads/use_libtommath_v1.2.0_for_real_maybe' must name a reffatal: process for submodule '3rdparty/libtommath' failed` 12:52
i've googled around a bit, but haven't found anything that's fixed it for me
hm, maybe i should create a branch in our submodule from the v1.2.0 tag (e.g., something like `git checkout -b v1.2.0_plus_changes_for_building_without_optimization`) which just has my one extra commit and try to point to that? 12:57
timotimo MasterDuke:i can try to help 13:13
is v1.2.0_for_real_maybe a tag? is use_libtommath_v1.2.0_for_real_maybe a moarvm branch or a "our own 3rdparty repository" tag or branch?
MasterDukeMobile timotimo: not at my computer and on a poor connection 13:49
But that’s a MoarVM branch
timotimo OK 14:07
timotimo if we want libtommath version 1.2.0 + a custom commit or two to be available, we'll have to create that as a branch or tag and push it to github, then we'd change the URL of the submodule to point at our fork, then we can push a version of moarvm's repo that has our stuff checked out 14:08
MasterDukeMobile The submodule does already point to our fork I think 14:11
I think I set a new remote to point upstream and pulled the v1.2.0 tag and set my branch to it 14:15
But maybe I haven’t yet pushed those commits to our fork
So maybe I should do that, then creating a branch in our fork with my extra commit seems like it might work 14:16
MasterDuke ah ha! 16:25
now it builds!
ci shows fails because of the 2 bad nqp expmod tests, but once we resolve how those should be handled we're good to go with libtommath v1.2.0
lizmat MasterDuke++ 16:31