Testing static and shared builds

[Torbjorn Granlund]

nisse at lysator.liu.se (Niels Möller) writes:

  Torbjorn Granlund <tg at gmplib.org> writes:
  
  > We should fix this.  Any autoconf/automake hacker listening that
  > volunteer?
  
  I don't think it makes much sense to have make check handle this.

Why not?

  I think it would be more natural to have the scripts run on the test
  machines (which I'm not familiar with, are they stored somewhere on
  shell?) configure two build trees with and without --disable-shared.
  Didn't you do something similar to test multiple ABI:s recently?
  
Sure, we could solve this for the automated builds, along the lines you
suggest.

The darwin problem would not have been caught through, since the GMP
project's core2-apple-darwin machine died after 13 month, just as the
warranty had expired.

Our testing is vital, but users' testing for compatibility problems and
compiler bugs is extremely important to avoid problems our testing can
never catch, however many machines we add to our test environment.

-- 
Torbjörn