Re: [PATCH v4 00/27] Use ref transactions for all ref updates
- From
Ronnie Sahlberg <sahlberg@google.com>
- Date
- Apr 29, 2014, 22:24 UTC
- Message-ID
- <CAL=YDWkv_XEXCpuAE8inD6Lj5NNHOEf9womimE70AxinO1Ufng@mail.gmail.com>
- In-Reply-To
- <xmqq4n1bke4b.fsf@gitster.dls.corp.google.com>
On Tue, Apr 29, 2014 at 2:41 PM, Junio C Hamano <gitster@pobox.com> wrote:
Show 20 quoted lines
> Ronnie Sahlberg <sahlberg@google.com> writes: > >> This patch series is based on mhagger/ref-transactions and expands on the >> transaction API. It converts all external (outside of refs.c) callers to >> use the transaction API for any writes. >> This makes most of the ref updates to become atomic when there are failures >> locking or writing to a ref. > > I think I saw some more comments to be addressed on this round, but > I'll try to push it through into 'pu' by attempting to resolve > conflicts so that I can ask you to eyeball the result. This series > seem to conflict with your own rs/ref-update-check-errors-early > topic that is already in 'next'. > > If I screwed up the conflict resolution (which is very possible), I > may have to ask you to rebase this series on top of acc62aa (which > is the merge of rs/ref-update-check-errors-early to 'next'), but I > am hoping that conflict resolution I'll push out today will be > correct. We'll see. >
Thanks.
I just sent v5 of the patch series where I have addressed the comments from Eric and Michael so far.
Otherwise, how hard would it be to drop rs/ref-update-check-errors-early ?
I can just redo those patches later, and that might be the easiest solution.
Regards ronnie sahlberg
> Thanks.