Re: 0.33 == Beta

On Tue, Nov 11, 2003 at 11:36:02PM -0600, kfogel@collab.net wrote:
>...
> For #1429, part of merging these changes into trunk is testing old
> client against new server, new client against old server, and 'make
> check' with the new code of course. If it passes all that, then the
> wider community is the next logical step

Agreed. The question is whether the "wider community" is the dev community
or the user community.

It may be that #1429 is okay because the failure mode is relatively easy
to recover from. (e.g. some problem in doing a checkout/update)

#1578 has a bit more problematic failure mode. Obviously, it is working
for the typical case (based on Mike's testing). But that edge condition.

> (we do have patch-level
> releases in case something is still wrong, after all).

Good point. Problems related to #1429 will probably be easy to spot and to
fix. And they'd probably turn up sooner rather than later. I wouldn't be
so quick to say the same about #1578.

> And if it
> doesn't pass all that, then of course it won't go into 0.33 or
> anything else until it's fixed :-).

Oh, sure. I certainly didn't doubt that the branches were passing the
regression test, plus other kinds of hand testing.

> The #1578 have been extensively scrutinized and run through a ringer
> of automated tests by Mike and others. I'm not saying there's zero
> risk, but I don't think it's as high as might be expected.

How many people actually and truly understand the Mystic Voodoo that Mike
just went through? Name two :-)

"I ran 'make check'" doesn't pass muster :-)

> Still, there's certainly something worrying about putting both a
> client/server protocol change *and* an fs schema change into one
> release... But maybe we should merge one now and wait on the other.

Yup. I'd be -0 on merging #1429 (still prefer to wait post-0.33). And I'd
be -0.9 for merging #1578 for 0.33.

Cheers,
-g

-- 
Greg Stein, http://www.lyra.org/
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org