Re: [PATCH 2/2] gitweb: New improved formatting of chunk header in diff

Jakub Narebski [off-list ref] writes:

O.K. This one introduced new feature, and wasn't needed for earlier 
accepted patch (output empty patches) to have sense. And it is rc1 
phase...

I personally do not mind minor gitweb 'feature' updates in -rc
cycle, just like I am planning to see if there are gitk updates
I haven't pulled from Paulus (I think there is at least one
patch from last month or so), unless the changes do not break it
so badly.

By the way, where I can find proper specifiction of unified diff format? 
Do I understand correctly that bot from and to ranges can be without 
number of lines part if it simplifies to 0?

When Linus did apply.c and I did diff.c, we primarily worked off
of sources to GNU patch.

There is a POSIX draft proposal now.

http://www.opengroup.org/austin/mailarchives/ag-review/msg02077.html

See also updates about the proposal.

http://thread.gmane.org/gmane.comp.version-control.git/29331/focus=29389

Junio C Hamano wrote:

Jakub Narebski [off-list ref] writes:

quoted

O.K. This one introduced new feature, and wasn't needed for earlier 
accepted patch (output empty patches) to have sense. And it is rc1 
phase...

I personally do not mind minor gitweb 'feature' updates in -rc
cycle, just like I am planning to see if there are gitk updates
I haven't pulled from Paulus (I think there is at least one
patch from last month or so), unless the changes do not break it
so badly.

I'll resend corrected patch then...
 

quoted

By the way, where I can find proper specifiction of unified diff format? 
Do I understand correctly that bot from and to ranges can be without 
number of lines part if it simplifies to 0?

When Linus did apply.c and I did diff.c, we primarily worked off
of sources to GNU patch.

There is a POSIX draft proposal now.

http://www.opengroup.org/austin/mailarchives/ag-review/msg02077.html

See also updates about the proposal.

http://thread.gmane.org/gmane.comp.version-control.git/29331/focus=29389

Thanks a lot.

Personally I think that we should push for git extended headers to be
(as an option) in POSIX for unifed diff... well, with the exception
of index and similarity lines. ;-)

-- 
Jakub Narebski