Bug report for GMP-4.3.0

[Torbjorn Granlund]

Paul Zimmermann <Paul.Zimmermann at loria.fr> writes:

  >      Did your build of gmp 4.3.0 complete all of its tests
  > without errors? I have been seeing four errors here in both
  > the gmp 4.3.0 release and in the gmp patched 4.3.0 mercurial
  > on Intel Darwin9. See my previous messages on the gmp-bugs
  > mailing list.
  
  I have just compiled NTL 5.5 with GMP 4.3.0 on a Core 2 under Fedora 10 and
  I have a similar failure:
  
  lasagne% make LLLTest
  lasagne% ./LLLTest < LLLTestIn > LLLTestOut.me
  LLL_FP...0.004
  LLL_QP...0.009998
  LLL_XD...0.009999
  LLL_RR...0.079988
  G_LLL_FP...0.004
  G_LLL_QP...0.019997
  G_LLL_XD...0.022996
  G_LLL_RR...0.219967
  LLL...0.006999
  lasagne% diff LLLTestOut LLLTestOut.me | wc -l
  14
  lasagne% diff LLLTestOut LLLTestOut.me | md5sum 
  c83001aec60b3d215817041671f5b2ef  -
  
While this might be a GMP bug, it is difficult for the GMP developers to
debug it.  If somebody can isolate the NTL part, and send a GMP test
case, then I will look into it.

-- 
Torbjörn