gup almost works in the default native Windows cmd space as-is (i.e. not depending on sh at all), and I think it should, too, since sh is a pretty arbitrary constraint that gup probably doesn't even make use of. (Is that true, actually?)
But, since Windows is against it with full force, and the canonical assoc+ftype+pathext solution fails for me (cannot execute gupfiles, for no apparent reason, despite everything seems to be set), a hopefully simple addition could push it through the obstacle for good.
For an experiment, I changed
self.gupfile = GUPFILE if indirect else target + '.gup'
to
self.gupfile = GUPFILE if indirect else target + '.gup.cmd'
and it seemed to work OK with the renamed test gupfile (the simplest example from your repo).
(Note: currently, this is the ONLY way I can run gup on this Win7-32 setup.)
Could it be officially supported, so that
- the extension can be set on the command-line
- and invocations from within a gupfile would use the current gupfile's extension as the default? (I'm not sure if this is already the case, or how non-trivial it can get, if not.)
Not having the ".gup" naming fixed might be useful on Unix, too.
gup almost works in the default native Windows cmd space as-is (i.e. not depending on sh at all), and I think it should, too, since sh is a pretty arbitrary constraint that gup probably doesn't even make use of. (Is that true, actually?)
But, since Windows is against it with full force, and the canonical assoc+ftype+pathext solution fails for me (cannot execute gupfiles, for no apparent reason, despite everything seems to be set), a hopefully simple addition could push it through the obstacle for good.
For an experiment, I changed
to
and it seemed to work OK with the renamed test gupfile (the simplest example from your repo).
(Note: currently, this is the ONLY way I can run gup on this Win7-32 setup.)
Could it be officially supported, so that
Not having the ".gup" naming fixed might be useful on Unix, too.