Re: linux-next: Fixes tag needs some work in the nfs-anna tree

From: Takashi Iwai
Date: Wed Jan 16 2019 - 01:32:23 EST


On Wed, 16 Jan 2019 00:38:11 +0100,
Paul Gortmaker wrote:
>
> [Re: linux-next: Fixes tag needs some work in the nfs-anna tree] On 15/01/2019 (Tue 23:12) Takashi Iwai wrote:
>
> > On Tue, 15 Jan 2019 22:41:21 +0100,
> > Chuck Lever wrote:
> > >
> > > Hi Stephen-
> > >
> > > On Jan 15, 2019, at 4:38 PM, Stephen Rothwell <sfr@xxxxxxxxxxxxxxxx> wrote:
> > >
> > > > [I am experimenting with checking the Fixes tags in commits in linux-next.
> > > > Please let me know if you think I am being too strict.]
> > > >
> > > > Hi all,
> > > >
> > > > Commit
> > > >
> > > > deaa5c96c2f7 ("SUNRPC: Address Kerberos performance/behavior regression")
> > > >
> > > > has problem with this Fixes tag:
> > > >
> > > > Fixes: 918f3c1fe83c ("SUNRPC: Improve latency for interactive ... ")
> > > >
> > > > The subject should match the subject of the fixed commit.
> > > >
> > > > --
> > > > Cheers,
> > > > Stephen Rothwell
> > >
> > > I shortened the commit title so that the Fixes: line is shorter than 68
> > > characters. I can leave these titles alone if that's preferred.
> >
> > I've sometimes shorted the subject like the above, too, as I find a
> > too long text annoying. Maybe the partial string matching should
> > suffice, especially when it ends with "..." ?
>
> The problem is consistency. Perhaps you shorten at four words. A
> person searches with five words or 70 chars - they never see your commit.

What's the reason to search for words instead of commit ID?

> The idea of consistency across the "Fixes:" tags is to allow a level of
> automated processing so that the creators of the stable releases can do
> a lot less manual hands-on processing. They have enough work to do.

Yes, I know, but the important point for stable pick-up is the
correctness of the commit ID, no?

I can understand the need for validity check of the Fixes tag,
especially to check whether the given commit ID is really correct, in
linux-next stage. But this can be verified even with a partial string
match.


thanks,

Takashi