Re: What's cooking in git.git (Jan 2019, #02; Tue, 15)

From: Junio C Hamano <gitster@pobox.com>
Date: Jan 17, 2019, 18:35 UTC
Message-ID: <xmqqlg3jnncx.fsf@gitster-ct.c.googlers.com>
In-Reply-To: <20190116212631.GK25639@hank.intra.tgummerer.com>

Thomas Gummerer <t.gummerer@gmail.com> writes:

Show 20 quoted lines

> On 01/15, Junio C Hamano wrote:
>> * ps/stash-in-c (2019-01-04) 27 commits
>>  - tests: add a special setup where stash.useBuiltin is off
>>  - stash: optionally use the scripted version again
>>  - stash: add back the original, scripted `git stash`
>>  - stash: convert `stash--helper.c` into `stash.c`
>>  ...
>>  - sha1-name.c: add `get_oidf()` which acts like `get_oid()`
>>  - Merge branch 'sd/stash-wo-user-name'
>> 
>>  "git stash" rewritten in C.
>> 
>>  Comments?
>
> I read over the series last weekend.  I had some small comments on
> 21/26 and 25/26.  I forgot to reply with my overall assessment on the
> thread though.  I'd be happy to give my Reviewed-by for the whole
> series with or without the changes I suggested.  I double-checked the
> re-introduced "legacy" stash script to make sure it is actually the
> same as it is on master.

Thanks for helping. And thanks for reminding us about the late addition of the legacy stuff, which makes the progression of the series less than ideal, but the benefit that would come from a possible reroll to start the series from the last three patches would be fairly limited anyway. Such a reorganized series would have allowed investigation of regressions and bugs during the development comparing the original and rewritten implementations slightly easier, but experience from seeing the evolution of these "reimplement in C" topics tells us that we see major part of the regression fallouts after the series is declared "feature complete", so in the long run, the less-than-ideal organization of the topic does not matter much in practice.

> So I'd be happy to get this merged into 'next', and if anything comes
> up to fix that on top.

OK.