On Mon, 2004-07-05 at 19:49, Garrett Rooney wrote:
> > Why not change 'num_back' to a signed number 'distance', and reverse
> > its default sense? That way the API can always go some distance
> > behind *or* ahead of a given revision.
>
> The issue with that (which is also why I'm not convinced on the new
> syntax options people have proposed) is that I'm not entirely sure how
> implementing 'FOO + 10' should work in the face of copies.
It doesn't make sense now in the face of copies (and even in the absence
of copies, we can't walk forward in history efficiently right now). But
with true rename support and a schema change to support forward history
walking (which will be particularly exciting in FSFS), it might make
sense.
On the other hand, I agree with people that -rHEAD-10 sounds like we
should just subtract ten from HEAD. Branko's proposal is consistent,
but very complicated. I sort of like --limit 10, although we could try
to dream up use cases for the more complicated syntax. ("svn up -r
BASE-2c filename"?)
Branko wrote:
> In your interpretation, HEAD-3 is currently r9966 for COMMITTERS,
> but r9793 for HACKING. I think both of you are expecting a can
> opener -- for cans of worms. :-)
Do we have the same problem for some of the other -r options? (BASE,
COMMITTED, PREV).
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Tue Jul 6 20:56:42 2004