Re: [PATCH v6 00/11] Column display

From: Junio C Hamano <gitster@pobox.com>
Date: Feb 26, 2012, 23:02 UTC
Message-ID: <7vaa45dwq7.fsf@alter.siamese.dyndns.org>
In-Reply-To: <1330170078-29353-1-git-send-email-pclouds@gmail.com>

Nguyễn Thái Ngọc Duy  <pclouds@gmail.com> writes:

Show 12 quoted lines

> Nguyễn Thái Ngọc Duy (11):
>   column: add API to print items in columns
>   Add git-column and column mode parsing
>   Stop starting pager recursively
>   column: add columnar layout
>   column: support columns with different widths
>   column: add column.ui for default column output settings
>   help: reuse print_columns() for help -a
>   branch: add --column
>   status: add --column
>   column: support piping stdout to external git-column process
>   tag: add --column

The first one adds something that nobody uses, hence it cannot be blamed by bisecting, but the "API" gets updated over time by other changes. Perhaps the first two may want to become a single commit.

Other than that, the patch progression looks very logical. The column output subsystem gets enhanced using git-column as the scaffolding to help testing and developing it, and then various commands start to make use of the result when the subsystem has become usable in their context.

But I am very much reluctant to see us adding a "git column" subcommand; I cannot justify it myself because what it does is even less connected to git than the "--no-index" mode of grep/diff commands, i.e. it does not have much to do with "Git, the version control system".

Shouldn't it be "test-column" that is useful during the hacking, much like "test-date", "test-chmtime", and "test_credential"?