Re: OpenSolaris - update - audio check

[John W. Eaton]

On 22-Jan-2009, Jim Langston wrote:

| Thanks John, that cleared up alot of the failures

  >>>>> processing /export/home/langston/OCTAVE/octave-3.0.3/src/data.cc
    ***** assert(norm(x,1), 20);
  !!!!! test failed
  error: norm: only implemented for double and complex values
  shared variables {
    x =

       1  -3   4   5  -7

  }

This failure and the others like it are not expected.  The constant

  x = [1, -3, 4, 5, -7];

should have class "double".

    ***** test
   A = [1:10; 1:10]/10;
   wavwrite("a.wav", A);
   [B, samples_per_sec, bits_per_sample] = wavread("a.wav");
   assert(A,B, 10^(-4));
   assert(samples_per_sec, 8000);
   assert(bits_per_sample, 16);
  !!!!! test failed
  error: mx_el_ne: nonconformant arguments (op1 is 1x0, op2 is 1x4)
    ***** test
   A=[1:10; 1:10] / 10;
   wavwrite("a.wav", A, 4000);
   [B, samples_per_sec, bits_per_sample] = wavread("a.wav");
   assert(A,B, 10^(-4));
   assert(samples_per_sec, 4000);
   assert(bits_per_sample, 16);
  !!!!! test failed
  error: mx_el_ne: nonconformant arguments (op1 is 1x0, op2 is 1x4)
    ***** test
   A=[1:10; 1:10] / 10;
   wavwrite("a.wav", A, 4000, 8);
   [B, samples_per_sec, bits_per_sample] = wavread("a.wav");
   assert(A,B, 10^(-2));
   assert(samples_per_sec, 4000);
   assert(bits_per_sample, 8);
  !!!!! test failed
  error: mx_el_ne: nonconformant arguments (op1 is 1x0, op2 is 1x4)

These are not expected failures.

    ***** test
   x = -3:0.5:3;
   y = abs (sin (x));
   assert (convhull (x, y, {"s","Qci","Tcv","Pp"}), [1;7;13;12;11;10;4;3;2;1])
  !!!!! test failed
  error: convhulln: not available in this version of Octave
  >>>>> processing 
/export/home/langston/OCTAVE/octave-3.0.3/scripts/geometry/delaunay.m

This test and the others that rely on qhull should be protected with
testif.  I think that is already done in the current development sources.

There are a number of other tests that fail to meet expected
tolerances by a small amount.  I don't think those are cause for
concern.  Maybe we should relax the tolerances slightly.

You should try to find out why norm is failing as that is likely to
cause some trouble.

jwe