Win98, nmake and ActivePerl are down to just the problems with the
split_command test which I have to think through some.  See the snapshot.

Some notes:

MM_Win32.t had lots of tests which were trying to test the Makefile
fragments themselves.  This won't work in the long run since formatting
changes will break the tests, and that's exactly what just happened.  I've
eliminated most of them.

In the past, the same method which generated a Makefile fragment also
initialized the values for that fragment.  This caused big duplication of
code between platform modules since you couldn't just alter a value without
overriding the whole method.  More and more, the methods have been seperated
into an initialization method (init_*) and a Makefile generation method.
Now that the seperation is more distinct, tests can call an init_* method
and test the values directly.


Twisty maze of escaping, all different:

Testing split_command and oneliner is a problem.  There's two levels of
escaping going on, first for the Makefile, then for the shell.  On Unix,
they're similar enough that I got away with running Makefile/shell escaped
commands straight through the shell.  Apparently this isn't the case on
Windows.  Not sure what to do about that.


-- 

Michael G. Schwern   <[EMAIL PROTECTED]>    http://www.pobox.com/~schwern/
Perl Quality Assurance      <[EMAIL PROTECTED]>         Kwalitee Is Job One
And it's made from all-natural baby skin, so you know it's good for the
environment.
        http://www.goats.com/archive/000606.html

Reply via email to