[COMMIT] Clean up tests/automated/mule-test.el, fix a problem revealed by it

[Stephen J. Turnbull]

Aidan Kehoe writes:

 >  >  >  > (3) The fact that the summary does not report the unexpected errors.
 >  >  > 
 >  >  > This may be hard to do well.  I'll take a look at it.

Remind me again later ... I forgot I promised to do this ....

 >  >  >  > Unexpected error (error "Can't activate input method
 >  >  >  > `german-postfix'") while executing interpreted code.
 >  >  > 
 >  >  > This is due to the fact that the Makefile suppresses adding packages
 >  >  > to the load-path.
 > 
 > No, it was due to Mike Sperber changing the load path for packages and not
 > telling anyone.

OK, but in any case automated tests should test that packages are
available before invoking functions defined in them.  If this is
annoying or hard to implement, we should improve test-harness.  What
do you think?

 >  > os-tests.el calls #’Skip-Test-Unless in a similar context (when components
 >  > required for a test are not available) so something similar might be the
 >  > least intrusive approach; I’m not certain what the right check would be,
 >  > though-- (assq 'leim packages-package-list) ? But leim doesn’t provide all
 >  > the input methods--egg and skk are exceptions, at least.
 > 
 > I've decided to use Skip-Test-Unless, with the not-particularly-robust
 > assumption that if input-method-alist is greater than one in length (the
 > first entry coming from canna, the #’register-input-method call for which is
 > dumped), then all the package input methods are available.

I think that any input method that is available should be registered
via LEIM, and autoloaded.  I went through a fair amount of trouble to
get something that sorta works for Canna, Wnn and SKK.

 > This reveals that the following input methods used in language
 > environments are not available in packages:
 > 
 >   latin-8-prefix latin-10-prefix latin-5-prefix latin-4-prefix latin-4-prefix
 >   latin-4-prefix latin-4-prefix latin-4-prefix belarusian bulgarian-bds 
 > 
 > Using Known-Bug-Expect-Failure calls to say that we know this already is
 > awkward,

Is that worth fixing for similar cases?  K-B-E-F was a quick hack that
did what was needed at the time.  (I actually did think about it, but
concluded that experience was necessary -- for me at least -- to
figure out what more, if anything, it should do.)

Sorry, I don't have time (yet) to look at the actual patch.