Re: Changelog quality

[Dmitry Torokhov <dmitry.torokhov@gmail.com>]

On Wednesday 13 January 2010 09:24:03 pm Németh Márton wrote:
> Greg KH wrote:
> > On Wed, Jan 13, 2010 at 01:04:52PM -0500, Alan Stern wrote:
> >> On Wed, 13 Jan 2010, Greg KH wrote:
> >>>> It would be nice if there was a way to link automatically a git commit
> >>>> to an archived copy of the email message in which it was originally
> >>>> submitted.
> >>>
> >>> Ingo has been doing this on some patches by putting a message-id field
> >>> in the signed-off-by area showing what lkml message a patch came from.
> >>>
> >>> See git commit id 6432e734c99ed685e3cad72f7dcae4c65008fcab in Linus's
> >>> tree as an example of this.
> >>>
> >>> I have no objection if people want to do this for any patches going
> >>> through my tree as well.
> >>
> >> If it has to be included manually by the patch submitter then it isn't
> >> automatic.  Not to mention that the submitter would need to know the
> >> email's message-id before the email message was sent!
> >>
> >> Something like this should belong in a maintainer's script.
> >
> > Yeah, that would get messy, as usually a patch happens in a different
> > thread than the original problem occurs, so it would take a lot more
> > work on my part to try to match things up.  If I notice it, I will try
> > to in the future.
> 
> What about introducing a new tag like 'SmPL-Used:'? The used SmPL could
> be placed somewhere in the web. This would require only one line in
> the changelog: e.g.:
> 
> SmPL-Used: http://coccinelle.lip6.fr/sp.php#usb_submit_urb
> 
> Usually when an SmPL is created in a general manner it can easily
> match 100+ places. These can be combined in groups which for example
> belong to one maintainter. That was what I did with adding const to
> the different ID tables.
> 

Hm, I usually am only marginally interested in how patch was generated,
you may have used a script to 'ed' to generate it for all I care.
I do care however about rationale behind the change and that is what
should go into changelogs.

So if these scripts could go below the "---" separator I'm all for it.

Thanks.

-- 
Dmitry