Jed Brown

@felipec You sent a patch for something else (in the test suite) and said you were not interested in submitting a patch for this. Gitifyhg doesn't have a convention of...

@felipec Maybe it would help to look at the commit itself rather than the PR summary: https://github.com/jedbrown/gitifyhg/commit/aa189f4a31b22a94eabc1ea9c6f0e2296122ee0c

I don't know, man. Named branches are bad, bad, bad. There are plenty of other sequences that can result in the refs for A and B being in the same...

Attaching a simple call graph from a repository with sizes: ``` 83 MB is hg revlog 195 MB as git loose objects 60 MB after `git gc --aggressive` 2.2 GB...

On the same repository, just compressing the 2.2 GB stream in one shot with zlib takes 82 seconds and produces 600 MB of output. I'm a novice with `perf`, but...

Just running zlib on the stream takes half as long as the entire fast-import, so it's not _that_ fast.

It's not pipe bandwidth per se, since we can run those 2GB through multiple pipes in 6 seconds (30x faster than our fast-import). I think the sequentialized nature of compression...

More data: `git fast-export --all` produces a 950 MB output stream (less than half the size of the input stream) in 9 seconds and it takes 54 seconds to `fast-import`....

The hg notes ref doesn't clone automatically, but if you add the ref manually (or have gitifyhg do it, but then it's also talking to the git mirror which it...

Thanks, @felipec. Is there an API with coarser granularity? Each file extract has to walk down the delta chain to a version where the full file is stored. That is...