Quoting Greg Hudson <ghudson@MIT.EDU>:
>
> On Mon, 2007-12-17 at 09:49 -0500, C. Michael Pilato wrote:
> > svnlgo@mobsol.be wrote:
> > > These options were added to stop the test suite caching credentials.
> >
> > Wait a second. In testing, we point to a custom configuration directory.
> > Why didn't we just set store-passwords=no in that configuration directory?
I think we do that right now.
> That's only half of the puzzle. You then have to provide the
> credentials when modifying the repository; thus passing --username and
> --password to every command.
>
> I'm not sure why caching credentials in the test suite is a bad thing.
Because:
a) a test should not have to rely on the environment in which its executed
b) a test should not behave differently when the environment changes
We encountered reason (b) when running the tests in parallel. Most of the tests
use the same username and as such don't interfere with each other. However,
some of the lock and authz tests use a second user account, which resulted in
random errors when a test was run while this other username was cached.
Lieven
----------------------------------------------------------------
This message was sent using IMP, the Internet Messaging Program.
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Mon Dec 17 17:07:50 2007