Steinar Bang wrote:
>>>>>>"Nico Kadel-Garcia" <nkadel@comcast.net>:
>
>
>>In short, no. In long, you could rebuild the repository from a dump
>>to revision 92, then add the files on top and recreate revision 93
>>with everything in place, then do revision 94. But it's not trivial
>>to merge changes.
>
>
> Hm... that's one option that would do it.
>
> But it would be a problem with a large repository, and would require
> all others to refrain from committing during the process, wouldn't it?
>
> I guess what I will have to do in this, and future cases when I mess
> up, is to do what I did when I messed up in CVS: live with it, and try
> not to mess up in the future...:-)
If you are concerned about doing this a lot, what you might do is create
a personal branch for working on the feature you're implementing. You
can play more fast and loose with the stuff in your personal branch
because it's yours. You can make as many commits as necessary to the
branch in order to make all the physical changes required by the single
logical change. Once you're happy everything on your branch is working
correctly, you can merge it into trunk (or wherever) with a commit
message describing the logical change. That gives you a single
changeset semantically described by a single commit message: Zen
perfection.
Whether this will work depends on how soon you realize you omitted
something. If you check in a revision and immediately slap your
forehead thinking, "D'oh!", this will work. If you wait for another
user or developer to figure it out and notify you that there's a
problem, this won't help a bit.
--
Danny MacMillan
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscribe@subversion.tigris.org
For additional commands, e-mail: users-help@subversion.tigris.org
Received on Sat Apr 29 01:06:04 2006