Rainer Jung wrote:
> 
> Also seeing Ruediger's observation concerning the empty shell loop in
> test Makefile for apr-util on Solaris; r780412 fixes it.

Yup, that's going to be picked up.  Very nice.

> On Windows testpass fails, because there's no crypt but the test still
> tries to validate the Unix crypted hashes against the password. So the
> test is broken.

Not exactly.  In the test suite;

#ifdef WIN32

or similar is absolutely verboten.  It indicates an APR design flaw, because
only APR_HAVE_... or APR_HAS_... or a result code of APR_ENOTIMPL can be
used to determine if APR on a given architecture supports something.

There's no test for "we have crypt" so the failure is valid.


Reply via email to