Closed ConcurrentCrab closed 3 months ago
Also please tag, the last one is from ~2 years ago.
Attention: Patch coverage is 65.30612%
with 17 lines
in your changes missing coverage. Please review.
Project coverage is 57.36%. Comparing base (
be41c53
) to head (575c161
). Report is 19 commits behind head on master.
:umbrella: View full report in Codecov by Sentry.
:loudspeaker: Have feedback on the report? Share it here.
@aymanbagabas not the exact change you suggested but I think this looks good as well.
@aymanbagabas not the exact change you suggested but I think this looks good as well.
Even better, thank you!
@ConcurrentCrab could you please rebase the commits to use conventional commits?
@aymanbagabas sure but... I'm not sure what else that would entail? FAICT all commits are in the format
<scope>: <title>
<body>
@ConcurrentCrab basically this
[feat|fix|refactor](<scope>): <title>
<body>
@aymanbagabas reworded the commit messages.
Hi charm.sh,
I was implementing the lfs-transfer protocol for gitea (https://github.com/go-gitea/gitea/issues/17554), and in the process I vendored your excellent transfer library from this repo. I made some changes for it to work better with Gitea's internal API, but I think those are improvements in the general sense too, hence I'm opening this PR.
All tests pass with the internal local backend in this repo.
Most of these changes are pretty self explanatory I hope, only the last commit is sort of contentious. In its defence, the old 2-stage Upload API was frankly horrific with the UploadState and whatnot. This works around that by using the same clever trick as
HashingReader
: aVerifyingReader
. This makes it much simpler for backends, as handling corrupt input is just the same as handling errors from the reader. Also makes it very simple for backends that verify input themselves to just omit wrapping the reader.