Re: [Issue 533] New - implement reserved checkouts

Branko Čibej <brane@xbc.nu> writes:
> 1. A lock must not create a new revision.

Is this really a requirement? I used to think so, but now I'm not so
sure.

The two reasons I've heard so far are:

   1. It causes noise in the logs and in the commit emails.

      But I don't buy that, because we can easily make the commit
      email script filter out revisions which only change lock states.
      The same goes for 'svn log'.

   2. It "uses up" revision numbers, or makes them less meaningful.

      I *really* don't buy that one, heh. I mean, we have the whole
      RapidSVN project in our repository, for crying out loud --
      commits in that subtree are entirely unrelated to Subversion.
      It doesn't get much less meaningful than that :-). And similar
      arguments apply to branches and tags anyway.

      In general, we've already determined that one can't simply
      assume that every revision in a repository is meaningful for a
      particular line of development. One has to apply filters, and
      now we're just talking about one more filter.

Revision numbers are cheap. We decided that long ago, and we've been
behaving that way in practice ever since.

I'm not saying that we definitely want locks to be versioned; just
that some of the most common objections to versioning locks are really
non-problems.

-K

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Tue Aug 13 17:58:43 2002