Re: [PATCH] checkpatch.pl: Relax commit ID check to allow more than 12 chars

[Joe Perches <joe@perches.com>]

On Mon, 2023-02-06 at 09:38 +0100, Geert Uytterhoeven wrote:
> Hi Joe,
> 
> On Sat, Feb 4, 2023 at 5:59 PM Joe Perches <joe@perches.com> wrote:
> > On Sun, 2023-01-29 at 09:52 -0800, Joe Perches wrote:
> > > On Sun, 2023-01-29 at 13:34 +0100, Jonathan Neuschäfer wrote:
> > > > By now, `git log --pretty=%h` (on my copy of linux.git) prints commit
> > > > hashes with 13 digits, because of the number of objects.
> > > > 
> > > > Relax the rule in checkpatch.pl to allow a few more digits (up to 16).
> > > 
> > > NAK without updating the process docs first.
> > 
> > btw: it looks like 12 will still be sufficient for awhile yet
> > 
> > $ git count
> > total 1154908
> 
> Hmm, Ubuntu git too old?

Don't think so

$ git --version
git version 2.39.1

More likely just using Linus' tree and not a
development tree with a bunch of branches.

I've got a -next tree with history back to next-20151106
with a bunch of missing dates because I don't fetch it
every day.  It has:

$ git tag | grep next | wc -l
1134

There I get:

$ git -c core.abbrev=5 log --pretty=format:%h | \
  perl -nE 'chomp;say length' | sort | uniq -c | sort -n -k2
      6 5
 542082 6
 568573 7
  51124 8
   3249 9
    217 10
     14 11
      1 12

> I've been using core.abbrev=16 for a while, and some maintainers
> reject my patches with Fixes: tags because of that...

Perhaps because that's not the documented format?

> Is it really worthwhile to save on the number of hexits, making lookup
> of some commits more inconvenient?
> 
> Note that while "git show edb9b8" suggests edb9b8f[...],
> gitweb says bad object id:
> https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/commit/?id=edb9b8

hmm.  Not here.

$ git show edb9b8
tree edb9b8

Kconfig
Makefile
fmvj18x_cs.c