Re: [PATCH 0/2] fix weird git --paginate behavior

From: Nguyen Thai Ngoc Duy <pclouds@gmail.com>
Date: Apr 6, 2010, 07:00 UTC
Message-ID: <x2zfcaeb9bf1004060000s7672df7dw5804ad54cb272a18@mail.gmail.com>
In-Reply-To: <20100406040145.GA30403@progeny.tock>

2010/4/6 Jonathan Nieder <jrnieder@gmail.com>:

Show 17 quoted lines

> Hi,
>
> Nguyễn Thái Ngọc Duy wrote:
>
>> Signed-off-by: Nguyễn Thái Ngọc Duy <pclouds@gmail.com>
>> Signed-off-by: Junio C Hamano <gitster@pobox.com>
>
> Thanks for posting this series.  I had not read it in full before.
>
> The series is very long and it is hard to make certain it does not
> introduce regressions.  Here is my idea for working around this: the
> idea is to grab patches from the series to form short, independently
> justifiable mini-series.  Once most of the functional changes are
> merged, it should be a lot easier to evaluate the core code change
> that makes setup more brittle (which I like a lot).
>
> What do you think?  Is it worth the trouble?

I don't know. There are few independent patches like this. Most of them will need startup_info struct, introduced in 01/43. I splitted the series in two parts, with hope that the first (simpler) part could graduate early. But as you pointed out, even the first part introduced a regression.

Anyway if you want to pick them out, candidates are:
 - 19-20/43: calculate prefix anyway and remove the hack in index-pack
 - 22-23/43: moving enter_repo(). This makes enter_repo() more
consistent with setup_git_dir*, no real gain.
 - 24/43: the bug only exposes when set_git_dir() is used extensively
in setup_git_dir*, so this patch alone seems useless.

So that's 7 patches shorter if all of them are sent independently.

-- 
Duy