On Tue, Feb 13, 2007 at 10:11:46AM +0000, Olly Betts wrote:
> If you run "make check", then no -b option is
passed and all backends
> which are enabled are used, but flint is run before
remote and remotetcp
> so the .flint directory should be up to date.
Oh, I see.
> Hmm, I wonder if the issue is remotetest not apitest.
That gets run
> before apitest, but probably relies on .flint being up
to date.
Aha. That could well be it.
> Really remotetest should just be merged into apitest at
this point. It
> used to be separate mostly because apitest didn't test
the remotetcp
> backend, but it now does. I'll sort that out.
Cool.
> > > Nuking on failure is problematic, since it
prevents you inspecting the
> > > database which can be helpful in working out
why a test failed.
> > Suppress nuking on -v?
>
> Perhaps, though I'm not sure -v should do more than
give more output.
True.
> > > Perhaps ixion actually ran out of VM or
processes?
> >
> > Possibly.
> >
> > > What was the actual error message apitest
gave?
> >
> > I can't remember in detail, but I think it was
something like
> > "couldn't find 'Listening' in output".
>
> That probably means that xapian-tcpsrv failed to start
cleanly. There's
> not much more we can do if that happens AFAICS.
We could catch the error message from it and print it out?
The test
will still fail, but more helpfully (in particular, it will
be obvious
if it's a local problem, or at least it'll be obvious to me
so I won't
bother other people if it happens again  .
J
--
/-----------------------------------------------------------
---------------
James Aylett
xapian.org
james tartarus.org
uncertaintydivision.org
_______________________________________________
Xapian-devel mailing list
Xapian-devellists.xapian.org
http://lists.xapian.org/mailman/listinfo/xapian-devel
|
