On Mon, 05 Jul 2004 00:46:00 +0200, Branko Čibej <brane@xbc.nu> wrote:
> rooneg@tigris.org wrote:
>
> >Author: rooneg
> >Date: Sun Jul 4 12:16:25 2004
> >New Revision: 10137
> >
> >Modified:
> > trunk/subversion/tests/fs-helpers.c
> >Log:
> >Make the libsvn_repos tests respect the FS_TYPE environment variable so
> >that 'make fsfscheck' will run the repository tests over the intended fs
> >backend.
> >
> >* subversion/tests/fs-helpers.c
> > (svn_test__create_repos): set the fs type in the config if FS_TYPE is
> > present in the environment. don't set the berkeley db errcall if we're
> > not creating a bdb filesystem.
> >
> >
> -1 Why use the environment here? Please just pass in a new fs-type
> parameter to the helpers, and get fs-test and friends to interpret the
> --fs-type option. All other tests do that. We shouldn't start using the
> environment for passing in such info. This is just hackish laziness.
Umm, excuse me, but the environment variable was already there, it's
how the makefiles pass the fs type on to the test suite now. It might
not be the most 'elegant' solution, but I don't think it's the end of
the world considering that this is the test suite, and it's
essentially a bit of global state anyway. Of course, it now occurs to
me that this probably isn't the way such things happen on windows, so
yes, a command line flag would probably be best in the long run. I'll
put it on my TODO list, but I don't know when I'll get to it.
> By the way, fs-test.c should be split into a BDB-specific part and an
> FS-agnostic part, too.
Sure, but that's a more extensive job, and I just happened to notice
this while messing around with the libsvn_repos tests, so I figure I'd
make an incremental improvement to the situation while I was in the
neighborhood.
-garrett
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Mon Jul 5 01:16:08 2004