Re: filesystem oops?

cmpilato@collab.net writes:
> Sorry, I should have explained that a little better. There is an
> svn_fs_predecessor_id() function, for operations that need to fly
> backward in node-Id ancestry, it's ridiculous to be allocating each
> predecessor from the pool when the simple calculation can be done
> in-place (using once-alloced memory).

Sure, but that can still be factored.

> So...rather than have our current repository have one wrong node, we
> should have hundreds? The 0 vs. 1 is a total bikeshed -- don't waste
> my time on it. The point is that all the current code (from
> predecessor calculation to id distance calculation to successor
> generation and beyond) expects .1. Why rewrite the whole friggin'
> system for such a ridiculous bikeshed?

How much code are we talking about?

We have to do a repository cleanup anyway (probably export and
re-import) eventually.

---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Sat Oct 21 14:37:04 2006