Re: [PATCH] bisect: mention "view" as an alternative to "visualize"

From: Eric Sunshine <sunshine@sunshineco.com>
Date: Nov 10, 2017, 22:08 UTC
Message-ID: <CAPig+cRYnwG4aTFh_5pgARbjoKPfzMBE12w4Boc6tzyKg54T1w@mail.gmail.com>
In-Reply-To: <alpine.LFD.2.21.1711101609240.6187@localhost.localdomain>

On Fri, Nov 10, 2017 at 4:13 PM, Robert P. J. Day <rpjday@crashcourse.ca> wrote:

Show 13 quoted lines

> On Fri, 10 Nov 2017, Eric Sunshine wrote:
>
>> Thanks for the patch. Some comments below...
>>
>> On Fri, Nov 10, 2017 at 11:32 AM, Robert P. J. Day
>> <rpjday@crashcourse.ca> wrote:
>> > Tweak a number of files to mention "view" as an alternative to
>> > "visualize":
>>
>> You probably want to end this sentence with a period, not a colon.
>
>   not sure about that, what's the standard when you're introducing a
> code snippet?

It wasn't clear that you were including a snippet since it's not common practice to include the diffstat in the commit message on this project...

Show 15 quoted lines

>> >  Documentation/git-bisect.txt           | 9 ++++-----
>> >  Documentation/user-manual.txt          | 3 ++-
>> >  builtin/bisect--helper.c               | 2 +-
>> >  contrib/completion/git-completion.bash | 2 +-
>> >  git-bisect.sh                          | 4 ++--
>> >  5 files changed, 10 insertions(+), 10 deletions(-)
>>
>> The diffstat belongs below the "---" separator, otherwise this text
>> will (undesirably) become part of the commit message proper.
>
>   no, i actually want that as part of the commit message. my standard
> is, if there are 5 or more files that get changed, i like to include a
> diff --stat in the commit message so people viewing the log can get a
> quick idea of how much has changed. if that's not desired here, i can
> remove it.

The same information is available to anyone interested in it simply by asking for it:

    git log --stat ...

More generally, commit messages should contain the really important information you want to convey to the reader which _isn't_ available some other way (by, for instance, taking advantage of the tool itself -- such as the above --stat example).

On this project, the diffstat is never included as part of the commit message, and it's likely that the patch won't be accepted by Junio like that (or he'll just edit it out if he does accept it).

Thanks.