Re: [PATCH v3 0/2] use unique out file in t5411
- From
Jiang Xin <worldhello.net@gmail.com>
- Date
- Feb 13, 2021, 15:13 UTC
- Message-ID
- <CANYiYbHZsvGpVPUbtpm3fPWgnNaL32yUMk5B-rrm_NUkm9fXmg@mail.gmail.com>
- In-Reply-To
- <xmqqft22w9v2.fsf@gitster.c.googlers.com>
Junio C Hamano <gitster@pobox.com> 于2021年2月12日周五 上午5:52写道：
Show 16 quoted lines
> > Jiang Xin <worldhello.net@gmail.com> writes: > > > From: Jiang Xin <zhiyou.jx@alibaba-inc.com> > > > > t5411 has some test cases for the behavior of failed 'git push' > > commands. Even the 'git push' command ended, the output file > > may still be updated with messages from 'git-receive-pack' command > > running in background. This breaks test cases which reuse the > > same 'out' file. > > I do not recall hearing from SZEDER who raised concerns about "a lot > of churn that can be avoided by a change in the code to die more > gracefully" in his previous round's review. This has been in 'seen' > for a long time, so let's move it down to 'next'. >
SZEDER's concerns about "a lot of churn that can be avoided ..." is in this post:
https://public-inbox.org/git/20210120124906.GA8396@szeder.dev/
I think current implementation of the client side of git-push is OK. It's no necessary to add more code and more roundtrips for git-push client side to complete the protocol in order to wait for end of the server.
-- Jiang Xin