[asterisk-dev] Subsystems in commit message summaries: +1

[Walter Doekes]

Reviving this, with a separate thread id.

>> On Mon, Oct 21, 2013 at 8:17 PM, Olle E. Johansson <oej at edvina.net> wrote:
>>> I would kindly like to suggest that we add the module name to
>>> these reports...

> On Tue, Oct 22, 2013 at 07:53:53AM -0500, Matthew Jordan wrote:
>> The other option is we follow the convention that commit messages should,
>> in their summary, list the affected modules. So the cookie summary should
>> be:
>>
>> http: Tolerate presence of RFC2965 Cookie2 header by ignoring it
>>
>> Some folks do this all the time; some don't. It isn't part of the
>> recommended commit message guidelines [2]. The downside of making it part
>> of the commit message guidelines is that it eats into the 80 character
>> limit for summaries and is (yet another) manual step.
>>
>> [2] https://wiki.asterisk.org/wiki/display/AST/Commit+Messages

Shaun wrote:
> My 2¢...
>
> +1 from me for adding the subsystem at the beginning of the commit
> message. Especially if the commit does affect some limited number of
> subsystems and is not a general change across the entire project.
>
> I am also of the opinion that prepending the subsystem should save
> space since a well written commit summary normally indicates the
> affected subsystem. For example, "http: Tolerate presence..." is
> more concise than saying "Tolerate presence...in the res_http
> module"
>
> It also quickly allows a reader to determine if they even care about
> the patch. For example:
>
>   "Fix major leak of sensitive customer data in chan_defrob"
>
>   vs
>
>   "chan_defrob: Fix major leak of sensitive customer data"

[snip]

> Hopefully in the not-too-far-off future the commit messages can be
> part of the commit under review. This way the person committing a
> patch doesn't have to draft the summary. The task of drafting the
> summary can be assigned to the patch authors so the extra manual
> step shouldn't be too onerous.
>
> I also fully accept that I'm probably in the minority.

I see Matt has adopted this convention for a few commits at least:

 
http://lists.digium.com/pipermail/asterisk-commits/2013-October/065841.html

That probably means my +1 won't be necessary.


Cheers,
Walter