Re: set-current-module broken in current Guile CVS version

[Dirk Herrmann]

On Wed, 27 Dec 2000, thi wrote:

>    From: Matthias Koeppe <address@hidden>
>    Date: 27 Dec 2000 13:06:41 +0100
> 
>    Some of the Guile changes since 2000-12-13 broke
>    `set-current-module'.
> 
> oops, previous mail sent too fast... anyway, could someone w/ write
> privs comment on the status of the patch referenced in the previous mail
> (originally sent 24-Mar-2000)?  if the patch is insufficient, by all
> means let me know so i can rework it and resubmit...  i think it's high
> time guile had a non-vestigial "make check".

Believe me, I _always_ run the test-suite before committing changes.  The
problem is not the lack of a 'make check' command (although this would be
nice to have, agreed), but rather of the large areas of untested stuff
within the test-suite itself.

But:  The problem that we address here did not show up with the current
test-suite, and would not with the type of tests that the test-suite
already contains.  The reason is, that the problem only occurs during
interactive use (namely when using the top-repl), but not when executing
a script.  Try the following script (and, for comparison, enter the
commands interactively):

  (display (current-module)) (newline)
  (define-module (guile-ilisp))
  (display (current-module)) (newline)

This behaviour _can_ also be tested, but requires a different testing
strategy, namely to spawn a guile process from a test-case and 'emulate'
interactive behaviour for that process.  The output of the process then
has to be checked against the desired output.

Best regards,
Dirk Herrmann