[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <e09057e0eefb221549ef9686826e2d190ef36a9c.camel@perches.com>
Date: Tue, 15 Oct 2019 08:50:00 -0700
From: Joe Perches <joe@...ches.com>
To: Geert Uytterhoeven <geert@...ux-m68k.org>
Cc: Geert Uytterhoeven <geert+renesas@...der.be>,
Andy Whitcroft <apw@...onical.com>,
Andrew Morton <akpm@...ux-foundation.org>,
Linux Kernel Mailing List <linux-kernel@...r.kernel.org>
Subject: Re: [PATCH] checkpatch: use patch subject when reading from stdin
On Tue, 2019-10-15 at 08:49 +0200, Geert Uytterhoeven wrote:
> Hi Joe,
Rehi Geert.
> On Mon, Oct 14, 2019 at 4:48 PM Joe Perches <joe@...ches.com> wrote:
> > On Mon, 2019-10-14 at 11:20 +0200, Geert Uytterhoeven wrote:
[]
> > > I gave your solution a try.
> > > It only enables the reset-on-next-patch feature when using stdin.
> > > Thanks to the printed subject, it's now obvious to which patch a
> > > message applies to.
> > > However, the output is significantly different than when passing
> > > a split patch series. Can this be improved upon?
> > >
> > > Note that the only reason I'm using stdin is that I use formail to split
> > > a bundle in individual patches. Once checkpatch supports bundles (or
> > > mboxes) containing multiple patches, there's no longer a need for
> > > using formail, and the reset-on-next-patch feature should be
> > > enabled unconditionally.
> >
> > Using your collection of little tools idea,
> > why not write a trivial script like:
> >
> > grep "^Subject:" $1
> > checkpatch.pl $1
> >
> > and use that as the command line for formail
> > instead of adding unnecessary complexity to
> > checkpatch?
>
> That would be another possibility.
>
> But given more maintainers are starting to apply patchwork bundles (cfr.
> the workflows discussions), it makes sense to make their lives easier.
But given this particular change only works for stdin, then this
patch splitting idea wouldn't generically work.
> This is also useful for maintainers who save all patches to apply into a
> single mbox, and run checkpatch+git-am on that.
Which also wouldn't generally work for checkpatch <mbox>
> Summarized: git-am handles multiple patches, checkpatch requires
> splitting.
I still think it's better to introduce YA script that disaggregates
aggregated patches/emails and feeds each individual patch to
checkpatch rather than making checkpatch learn how to disaggregate.
Using "git mailsplit" as part of some additional script could work.
Powered by blists - more mailing lists