>>>>> "Brian" == Brian D Carlstrom <firstname.lastname@example.org> writes:
Brian> Eric Marsden writes:
>> >>>>> "mg" == Martin Gasbichler <email@example.com>
>> >>>>> writes:
mg> + make check tests only the S48 part (with it's borken socket
mg> implementation). The tests relevant to scsh itself can be invoked by
mg> scsh -lm scsh/test/test-packages.scm -o test-all -c "(test-all)"
>> ok, these tests work on MacOS X. This was scsh 0.6.4 (Beta 4).
Brian> I updated to the latest (as of this afternoon) cygwin. Things basically
Brian> work. I hadn't run test-all before. I found that there are failures
Brian> because Win32 with cygwin doesn't have fifos. Is there a methodology for
Brian> skipping tests based on platform? Should I not add the test with
Brian> add-test! or should I check within the test itself? I think the former
Brian> would make more sense.
There is no methodology for skipping tests yet. However, your fifo
example is there only case I'm aware of where a feature of scsh
doesn't work on all systems. I would rather declare this a bug in scsh
(which of course we cannot fix) than changing the test suite to skip
the test. We could add some output to test-all saying which tests are
known to fail due to bug in scsh or add an optional argument to
add-test! which indicates whether the test is known to fail.