Would it be better for git-commits-head to have -M -C diffs?

Would it be better for git-commits-head to have -M -C diffs?

[James Cloos]

Would anyone else prefer to see the posts on the commits list use git's
copy and move detection?

Git show (used by the feed-mail-list script) does support -M and -C.

It does affect the stat section.  The recent commit 758db3f21187 would
have had its stat changed from looking like:

 arch/h8300/include/asm/Kbuild              |    1 +
 arch/h8300/include/asm/a.out.h             |   20 ++
 arch/h8300/include/asm/atomic.h            |  144 +++++++++++
                        .
                        .
                        .
 include/asm-h8300/Kbuild                   |    1 -
 include/asm-h8300/a.out.h                  |   20 --
 include/asm-h8300/atomic.h                 |  144 -----------
                        .
                        .
                        .

to looking like:

 .../asm-h8300 => arch/h8300/include/asm}/Kbuild    |    0 
 .../asm-h8300 => arch/h8300/include/asm}/a.out.h   |    0 
 .../asm-h8300 => arch/h8300/include/asm}/atomic.h  |    0 
 .../asm-h8300 => arch/h8300/include/asm}/auxvec.h  |    0 
 .../asm-h8300 => arch/h8300/include/asm}/bitops.h  |    0 
 .../h8300/include/asm}/bootinfo.h                  |    0 
                        .
                        .
                        .

but would have been 512 lines and 24271 octets instead of 10623 lines
and 291634 octets.

I think it would also be useful were the From: header of each message
set to the commit's Author.  I run each commit through a script to do
that locally, but the archives would also benefit from that.

-JimC
-- 
James Cloos <cloos@jhcloos.com>         OpenPGP: 1024D/ED7DAEA6

Re: Would it be better for git-commits-head to have -M -C diffs?

[Randy Dunlap]

On Thu, 14 Aug 2008 12:56:18 -0400 James Cloos wrote:

> Would anyone else prefer to see the posts on the commits list use git's
> copy and move detection?
> 
> Git show (used by the feed-mail-list script) does support -M and -C.
> 
> It does affect the stat section.  The recent commit 758db3f21187 would
> have had its stat changed from looking like:
> 
>  arch/h8300/include/asm/Kbuild              |    1 +
>  arch/h8300/include/asm/a.out.h             |   20 ++
>  arch/h8300/include/asm/atomic.h            |  144 +++++++++++
>                         .
>                         .
>                         .
>  include/asm-h8300/Kbuild                   |    1 -
>  include/asm-h8300/a.out.h                  |   20 --
>  include/asm-h8300/atomic.h                 |  144 -----------
>                         .
>                         .
>                         .
> 
> to looking like:
> 
>  .../asm-h8300 => arch/h8300/include/asm}/Kbuild    |    0 
>  .../asm-h8300 => arch/h8300/include/asm}/a.out.h   |    0 
>  .../asm-h8300 => arch/h8300/include/asm}/atomic.h  |    0 
>  .../asm-h8300 => arch/h8300/include/asm}/auxvec.h  |    0 
>  .../asm-h8300 => arch/h8300/include/asm}/bitops.h  |    0 
>  .../h8300/include/asm}/bootinfo.h                  |    0 
>                         .
>                         .
>                         .
> 
> but would have been 512 lines and 24271 octets instead of 10623 lines
> and 291634 octets.
> 
> I think it would also be useful were the From: header of each message
> set to the commit's Author.  I run each commit through a script to do
> that locally, but the archives would also benefit from that.


or the git-commits list could use what Linus has requested to be used
in pull requests:

Please use "git diff -M --stat --summary" to generate the diffstat:
the -M enables rename detection, and the summary enables a summary of
new/deleted or renamed files.


---
~Randy
Linux Plumbers Conference, 17-19 September 2008, Portland, Oregon USA
http://linuxplumbersconf.org/

Re: Would it be better for git-commits-head to have -M -C diffs?

[David Woodhouse]

On Thu, 2008-08-14 at 12:56 -0400, James Cloos wrote:
> Would anyone else prefer to see the posts on the commits list use git's
> copy and move detection?
> 
> Git show (used by the feed-mail-list script) does support -M and -C.

I see no particular reason not to do that. I've spent the whole day
screwing with magic git scripts on master.kernel.org, fixing the
headers-export stuff; I'll do that when I'm finished, if nobody screams.

> I think it would also be useful were the From: header of each message
> set to the commit's Author.  I run each commit through a script to do
> that locally, but the archives would also benefit from that.

It would be nice, but I'm not sure it's a good idea. They aren't always
valid email addresses. Maybe a Reply-To: header, rather than From: ? 

-- 
David Woodhouse                            Open Source Technology Centre
David.Woodhouse@intel.com                              Intel Corporation

Re: Would it be better for git-commits-head to have -M -C diffs?

[James Cloos]

>>>>> "David" == David Woodhouse <dwmw2@infradead.org> writes:

JimC> I think it would also be useful were the From: header of each
JimC> message set to the commit's Author.  I run each commit through a
JimC> script to do that locally, but the archives would also benefit
JimC> from that.

David> It would be nice, but I'm not sure it's a good idea. They aren't
David> always valid email addresses. Maybe a Reply-To: header, rather
David> than From: ?

I'd still need to pass them through the script I use; the point is to
have the summary window in my MUA display useful names in the mail
author column.  

Archive searching would also need it to be the From: header.

They would, though, probably need a Reply-To:, MFT: or similar pointing
replies to linux-kernel.  I hadn't thought of that aspect until reading
your reply....

I suppose the -M -C change would be enough for now.

-JimC (still waking up)
-- 
James Cloos <cloos@jhcloos.com>         OpenPGP: 1024D/ED7DAEA6