Re: an autotest design flaw

[Ben Pfaff]

Bruno Haible <address@hidden> writes:

> What could be the fix? IMO one needs to get away from the concept of
> generating something at distribution time. I think the right implementation
> would be a library of shell functions.

We've been doing something like that in GNU PSPP.  The current
implementation is not good or anything I would recommend, because
it repeats the same library of shell functions in every test
script, instead of factoring them out, but it does seem to work
OK.

One thing we've done to make debugging, etc. easier is that each
PSPP invocation is written with $SUPERVISOR at the beginning of
the command.  Then I can debug a particular test with something like
        make check TESTS=tests/whatever.sh SUPERVISOR='gdb --args'
and I can automatically run valgrind on every test with
        make check SUPERVISOR=valgrind
Quite useful.
-- 
Ben Pfaff 
email: address@hidden
web: http://benpfaff.org