Dave Brown wrote:
> Attached is a script to do setup/teardown of httpd for testing svnsync
Hi Dave. That looks really useful. I haven't tried it , but
especially if you could update it a bit to live under
subversion/tests/cmdline and figure out its arguments automatically, I
think we want this.
There are all sorts of things in it (addresses 127.0.0.x springs to
mind) that other people can refine or parameterize once they start to
> mirroring. I started manually making sure that issue 2939 didn't come
> back with 1.7, but davautocheck.sh inspired me to automate it. The
> happy news is that 2939 has not come back in trunk or 1.6.6, and it
> reproduces the bug on 1.5.1, where originally reported.
> I think this would be a good addition to the tests somehow, as AFAIK
> mirroring through SVNMasterURI in httpd has no automated testing. The
> script currently tests only 2939, but can be the basis for testing all
> kinds of changes over mirroring.
Cool. It would be good to separate the test code from the
infrastructure, of course, but that's not essential at first.
> If people agree that it would be a good test to add, I'll propose a
> proper patch. (It needs a bit more polish to fit the test cmdline
> interface as defined in subversion/tests/README). Also, is there a
> problem with relying on svnmucc?
No problem relying on svnmucc for this: that can be a requirement for
running these particular tests.
> The script currently takes two arguments: your working copy of the code
> to test, and a test-work directory. (It would figure these out if it
> lived below tests/...) It ought to work everywhere davautocheck.sh
> works, but I've tested on unbuntu and centos. And it will complain if
> svnmucc isn't built in your wc. See the comments at the top of the
> script for more details.
 I have trouble setting up Apache - seems my whole package management
is hosed. :-(
Received on 2009-11-18 16:35:58 CET