Rename/copy detection - options/display improvement proposition

Hi,

I think that:

   1. git-status should accept -M<n> -C<n>, same as git-diff
   2. git-diff should have an option to omit diff for deleted files


1. git-status incorrectly recognizes some changes as copied, e.g.:

# Changes to be committed:
#	copied:     SimpleMatcher.h -> BSlashMatcher.h
#	copied:     SimpleMatcher.h -> StringMatcher.h
#	renamed:  SimpleMatcher.h -> TabMatcher.h

(generally SimpleMatcher.h is recognized as being copied into many
different headers)

git-diff shows similarity index 50-60%.
git-diff -M80 -C80 correctly stops recognizing this as copies/renames.

I think git-status also should learn -M and -C, as it works the same
way as git-diff with defaults.


2. git-diff does not show diff for copied files. But when a files is
removed, the diff is shown.

I think seeing what was deleted is sometimes useful, but sometimes it
is not and just clutters the
diff.



-- 
Piotrek

On Thu, Mar 3, 2011 at 4:49 PM, Piotr Krukowiecki
[off-list ref] wrote:

Hi,

I think that:

  1. git-status should accept -M<n> -C<n>, same as git-diff
  2. git-diff should have an option to omit diff for deleted files


1. git-status incorrectly recognizes some changes as copied, e.g.:

# Changes to be committed:
#       copied:     SimpleMatcher.h -> BSlashMatcher.h
#       copied:     SimpleMatcher.h -> StringMatcher.h
#       renamed:  SimpleMatcher.h -> TabMatcher.h

(generally SimpleMatcher.h is recognized as being copied into many
different headers)

git-diff shows similarity index 50-60%.
git-diff -M80 -C80 correctly stops recognizing this as copies/renames.

I think git-status also should learn -M and -C, as it works the same
way as git-diff with defaults.


2. git-diff does not show diff for copied files. But when a files is
removed, the diff is shown.

I think seeing what was deleted is sometimes useful, but sometimes it
is not and just clutters the
diff.

I have committed the changes and git-commit, git-log--stat output also
shows the renames - so I suppose this is a global issue

-- 
Piotrek