Re: [RFC PATCH 1/1] mv: integrate with sparse-index

From: Shaoxuan Yuan <shaoxuan.yuan02@gmail.com>
Date: Mar 17, 2022, 08:37 UTC
Message-ID: <CAJyCBOTfaaeqhiRS6xFzQHpf-H35ATygKJqWYDijfPDJOGcShQ@mail.gmail.com>
In-Reply-To: <20ffd93d-e3dd-4df6-5ec7-d3577cac910d@github.com>

Hi Derrick and Victoria.

On Wed, Mar 16, 2022 at 1:14 AM Derrick Stolee <derrickstolee@github.com> wrote:

Show 13 quoted lines

> You mention in your cover letter that the ensure_not_expanded tests
> are not added yet (same with performance tests). Now that you've
> gotten feedback on this version of the patch, I might recommend the
> organization you might want for a full series:
>
> 1. Add these 'mv' tests to t1092 _without_ the code change. These
>    tests should work when the index is expanded, and making the
>    code change to not expand the index shouldn't change the
>    behavior.
>
> 2. Add the performance test so we have a baseline to measure how
>    well 'mv' does in the normal case (and how it is slower when
>    expanding the index).

I'm a bit caught up here.

Do I just do a before-code-change test and after-code-change test, and benchmark the after against the before?

Or do you mean I should also perf test out-of-cone arguments with 'mv' so that the index could be expanded? According to my understanding, the sparse-index could be required to expand when out-of-cone actions happen and the 'ensure_full_index()' is called. And do a 3-way comparison among before-code-change, after-code-change, and after-code-change- index-expanded, no?

-- 
Thanks & Regards,
Shaoxuan