Re: Running 1.4.x tests against trunk

On 6/29/07, David Glasser <glasser@mit.edu> wrote:
> Most of these are issues with the attempt to make the new depth
> support emulate the old -N support. They can indicate a two different
> things:
>
> (a) They should be passing, and the backward-compatability code on
> trunk needs to be fixed in order to make them pass.
>
> (b) The trunk behavior is different, but -N was never well defined and
> it's OK if we change its behavior a little bit.
>
> In my opinion, even if we do accept a little bit of changed behavior
> for -N, we should change any tests on the 1.4.x that fail for type (b)
> reasons to make them pass against both 1.4.x and trunk servers. That
> way, they won't mask type (a) errors, which are serious.

Nope, they're just bugs in svnserve; Karl found the erroneous lines,
and I'll commit tomorrow once both test suites pass.

> (The one non-depth failure above, commit #34, is definitely of type
> (b): an error message changed. I think we should change the expected
> error to be a regexp that matches both versions. Should I just do
> this on trunk and propose it for backport? Or should I do it on a
> branch from 1.4.x?)

Karl agreed that making the trunk version pass against both 1.4 and
1.5 and proposing that for backport seems reasonable.

(The question of testing client back-compatibility is still tricky though...)

--dave

-- 
David Glasser | glasser_at_mit.edu | http://www.davidglasser.net/
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org