Re: [Gnu-arch-users] request for comments on partial commits

[Rene Schallner]

On Feb 22, 2006, at 07:57 , Andy Tai wrote:

> Hi, for the partial commit problem I am thinking of a simple method:
> the user will have to commit the whole tree after a file renaming or
> file deletion operation, which means the change to the directory
> content.  Are there any objection to this approach as a short-term
> measure?   Thanks for comments on this.

This is rather an opinion than an objection: That makes moving/ 
renaming/deletions quite "dangerous" operations, especially when they  
happen in directories completely unrelated to other "normal" file  
changes.

It would seem irrational to me if I had to commit the whole tree  
(example: consisting of ./foo/*.c and ./bar/*.c) when I mv/rm in one  
directory.

If you hack foo/main.c you can always make a quick bugfix to bar/ 
module.c and commit the bugfix without having tagged beforehand.     
But if you renamed/deleted foo/utils.h, you'd be unable to commit  
just bar/module.c.  I'd think "Why does tla bother about ./foo/ when  
i want to commit a file in ./bar/ ????"

Would it be possible to make a special case for directories  
unaffected by mv/rm operations?