Joseph S. Myers wrote:
> Doing it right at the end of branch-to-trunk merges for a release (which
> is where we are right now, just after merges from Graphite) is probably
> the optimal timing in terms of minimising the amount of branches that will
> need fixing up. The problem is doing it without warning or consensus.
Yep. Given that, maybe we should make a global whitespace cleanup an
official part of the release branching process? It would be nice to keep it
from all building up again, and we could avoid the pain if it was regular and
predictable.
> But I think hastily removing SVN revisions (so breaking all subsequent
> checkouts of trunk) is much worse than hastily checking in such a bulk
> change.
I think the risks of manually editing a vcs database outweigh the pain of
having a few irritating merges to do.
cheers,
DaveK
