On Thu, 20 Oct 2005, Madan U Sreenivasan wrote:
> On Thu, 2005-10-20 at 01:05, Peter N. Lundblad wrote:
> > do is to stip the server/client part from the patch and run the testsuite
s/stip/strip/ ;)
> > (for all RA layers), and then repeating for the other part of the patch.
> > You'll get failures for your new test, but that's expected.
> Am not sure I understand this, Peter, could you pl. elaborate on this...
> thx
I was just describing a strategy to test against an "old" client or
server. You could also test with an 1.3 client/server if you want to. The
problem with this is that you might get spurious failures ecause tests in
the new code expect a newer server. This might not be a problem since we
are so close after the branching, but after we commit your patch, it wil
be;) (Because you add a test that is intended to fail against an old
server.)
To clarify further:
When I wrote the svn_ra_reparent stuff recently, I first wrote the client
stuff complete. Then I ran an svnserve and tested against that. then I
wrote the svnserve command for reparent and tested again. (There were no
server changes for libsvn_ra_dav). I didn't need to test old client
against new server, since I only introduced a new command. You're
extending a server response, so you need to test in both directions. You
need to ensure that an old client doesn't choke on the new stuff rom the
server, and you need to ensure that a new client doesn't die when it
doesn't get that extra information from an old server. And this is
necessary for both ra_svn and ra_dav:-(
Hope this clarifies, and happy testing.
Regards,
//Peter
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Thu Oct 20 11:09:33 2005