Greg Stein <gstein@lyra.org> writes:
> This is precisely why I'm more comfortable with forward motion than reverse.
> It sounds like there are some ripple effects that you aren't going to be
> able to back out effectively, so we shouldn't even try.
Nah, I think that fs-test thing is a red herring (I mean, there is
some bug that happens to get stimulated by the access baton patches,
but that could happen with any sort of change).
The access baton patches did not have a huge effect on the rest of the
code. Callers of libsvn_wc had to be adjusted, of course, and the way
some things were stored in a hash at commit time changed, but
basically no code since the access baton commits has been really
dependent on those changes (except subsequent commits for the same
issue, which are part of the reversion, of course). I checked this
assessment with Ben, who agrees.
I can't *guarantee* no ripple effect, but judging from the code,
backing these out should be pretty safe.
(Again, not arguing wholeheartedly for reverting here -- just saying
we can do it if we want to. I'd also rather move forward, if we give
ourselves enough time.)
> Regardless, Philip should check in any regression tests that he has,
> independent of other patches and fixes. That regression test can stay even
> if we roll back. Tests can't hurt :-)
+1. Assuming it passes :-).
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Mon Jul 15 22:13:15 2002