It looks like https://git.llvm.org/git/clang.git/ is two days behind trunk in svn.
If I checkout from https://git.llvm.org/git/clang.git/ the last commit is https://llvm.org/svn/llvm-project/cfe/trunk@319007
But if I checkout from SVN directly the last commit is at 319145.
Anyone who knows something more about this? Is it just a temporary glitch?
Regards,
Björn Pettersson
I've been wondering the same.
Adding Ben, can this be related to the automatic CC'ing of cfe-commits in Phabricator?
The Clang repository is still importing in there: https://reviews.llvm.org/source/clang/
I don't know how git-svn is implemented...
Jonas
Hi Jonas! I only changed settings in Phabricator; I didn’t knowingly change anything related to the git mirrors.
However, it’s definitely possible the git mirror process is blocked waiting for the Phabricator SVN import process to complete. If that’s the case, we just have to wait until Phabricator finishes importing SVN revisions.
Does anyone know who maintains the git-svn mirror infrastructure?
Ben
We could also disable I/O for the SVN URLs belonging to the new Phabricator projects I created. This would mean they would no longer watch for commits, which might cause some features in Differential and Herald to not work.
I’ll look into what the effects of disabling I/O are.
I couldn’t find clear documentation of what the effects of choosing “No I/O” would be for the Diffusion repositories, so I posted a question in the Phabricator Discourse community.
In the mean time, I’ll trawl through the source to see if I can find anything illuminating.
Any news on getting the clang git repo unstuck?
It’d be great if we could find someone who knows how the git-svn import works.
Otherwise, assuming the Phabricator import is causing the git-svn import to serialize and wait, the clang repo import is at 99.67% complete:
https://reviews.llvm.org/source/clang/
so for now I’ll just wait until it completes. Hopefully the other repositories will finish quickly (they only care about a small percentage of the commits). If they don’t, I’ll disable I/O for them.
Ben
I believe Anton is the one who knows how it works.
Should be fixed now.
Just to close the loop, I confirmed with Anton that this issue was not related to the Phabricator configuration change I made.
Ben
Sorry for the wrong suspicion, it was the first thing a came across when looking for coincidences.
Thanks for fixing the problem!
Jonas
I can confirm (as starter of this thread) that it works for me as well now. Thanks for the fix!
/Björn