Changing a commit message

[Raúl Núñez de Arenas Coronado]

Saluton John :)

On Tue, 13 May 2008 17:43:28 -0500, John Arbash Meinel dixit:
> Raúl Núñez de Arenas Coronado wrote:
> | OK, an annotation would not be history, but that doesn't solve the
> | distribution of such annotations unless they are local. And then you
> | have to be careful about which branch are you working on (different
> | branches may have different anotations).
> 
> So the basic idea is that you have 2 DAGs. One is the branch history,
> the other is the annotated history. When you modify an old commit,
> you update the annotated history with a new node. The annotation
> graph would probably be repository wide, but there are various
> arguments there.
[...]
> 
> The downside is that it is another DAG to maintain, and another
> possible source of conflicts. eg. What happens when we both "fix" a
> commit message, but we disagree about the exact text. Now we need a
> way to communicate to the user that there is a conflict in the change
> to the revision message from XXX.

Exactly, that's what I meant in one of my last messages: Bazaar would be
adding yet another source of conflicts for the users, and we are not
talking about conflicts in the data, which is more or less inevitable
and the users must live with that, but with the metadata. And even more:
the conflicts may happen in revisions that the users may not be
interested in, if they branched to do some particular job with new code.

> Anyway, the technical solution is actually pretty simple. The
> downside is how much it makes the UI convoluted, and requires users
> to do things that they don't really want to do.

I agree. Myself, as a possible user faced with this problem, would be
very annoyed if I'm working in revisions > 100 and somebody messed up
the messages for revision 25, since merging other branches would be a
hard job without this additional problem.

Raúl "DervishD" Núñez de Arenas Coronado
-- 
Linux Registered User 88736 | http://www.dervishd.net
It's my PC and I'll cry if I want to... RAmen!
We are waiting for 13 Feb 2009 23:31:30 +0000 ...