Brock Peabody
Brock Peabody
How slow? I see about a 400-500ms difference (in a repo with about 8,000 submodules) between `git meta close .` and `git meta close a-submodule-path`. It definitely doesn't crawl the...
@jhedwardyang not sure I follow. What code is pushing what refs twice? Also, where is the error coming from? The closest thing I see to this error message in `libgit2`...
Got it. The phrase "synthetic hook" made me think it was a problem with the server-side hook.
This issue is actually a little less straightforward than I thought. We can't just "not push all submodules" as different ref-specs may or may not indicate that different meta-refs need...
The way it works right now is... expedient. I'm not sure that it's tractable to figure out what the exact right set of submodule commits to push is. To do...
Why would we assume that users push only one commit at a time?
I think it's OK to use that point as an optimization, but I don't think the tool should be broken if someone has more than one commit.
How about we push (a) open submodules and (b) any submodules changed in the commit being pushed (opening them if necessary)?
I need to overhaul the system to derive submodule existence from the tree -- a submodule exists at path X iff there is a COMMIT type entry at X.
1. I think there is either a problem in the way libgit2 renders conflicts (not our flaky test related to this) or I am not using their API correctly. In...