New failures related to recent developments

[Torbjorn Granlund]

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

  Torbjorn Granlund <tg at gmplib.org> writes:
  
  > Please take care of any fixing.  It's your baby after all.  :-)
  
  Checked in now. Hope it's sufficient.
  
We'll see tomorrow morning.

Note that the failures are sticky, a passing "make check" does not
supplant a previous failing one, since that could lead us to miss a
failure.  Once the new builds are done, I will check that the previously
failing seeds now preduce no error, then clear the failure to make the
reporting square become green (gmplib.org/devel/tm-date.html).

  I've added some make rules to make it possible to run make
  check-mini-gmp from the gmp tree. It's should work both when building in
  the srcdir and when using a separate build dir. Limitations: It uses
  mini-gmp/tests/Makefile which requires GNU make. And built files are not
  removed by make clean (I'd need to check the automake manual for where
  to put that).

Please then add both a clean and a distclean target (perhaps working in
the exact same way).
  
  No changes to the actual tests, yet. There's lots of room for
  improvements there.
  
Do you test reuse at all?

  The bug in mpz_abs_add actually triggered failures in mini-gmp's t-gcd
  and t-powm. But they didn't earlier when I built mini-gmp separately, I
  don't understand why.
  
Different optimisation?

-- 
Torbjörn