Re: make check failure for octave 2.9.6 under Mac OS X 10.4

[Marius Schamschula]

David,

I uninstalled octave/octave-forge 2.9.6 and reinstalled octave 2.9.6 only. Here is the tail of the output of test_sparse.m

 assert(x,as);

  ***** assert(issparse(horzcat(as,as)));

  ***** assert(issparse(vertcat(as,as)));

  ***** assert(issparse(cat(1,as,as)));

  ***** assert(issparse(cat(2,as,as)));

  ***** assert(issparse([as,as]));

  ***** assert(issparse([as;as]));

  ***** assert(horzcat(as,as), sparse([af,af]));

  ***** assert(vertcat(as,as), sparse([af;af]));

  ***** assert(horzcat(as,as,as), sparse([af,af,af]));

  ***** assert(vertcat(as,as,as), sparse([af;af;af]));

  ***** assert([as,as], sparse([af,af]));

  ***** assert([as;as], sparse([af;af]));

  ***** assert([as,as,as], sparse([af,af,af]));

  ***** assert([as;as;as], sparse([af;af;af]));

  ***** assert(cat(2,as,as), sparse([af,af]));

  ***** assert(cat(1,as,as), sparse([af;af]));

  ***** assert(cat(2,as,as,as), sparse([af,af,af]));

  ***** assert(cat(1,as,as,as), sparse([af;af;af]));

  ***** assert(issparse([as,af]));

  ***** assert(issparse([af,as]));

  ***** assert([as,af], sparse([af,af]));

  ***** assert([as;af], sparse([af;af]));

  ***** assert(as==bs,sparse(af==bf,true))

  ***** assert(bf==as,sparse(bf==af,true))

  ***** assert(as!=bf,sparse(af!=bf,true))

  ***** assert(bf!=as,sparse(bf!=af,true))

  ***** assert(as+bf,af+bf)

  ***** assert(bf+as,bf+af)

  ***** assert(as-bf,af-bf)

  ***** assert(bf-as,bf-af)

  ***** assert(as.*bf,sparse(af.*bf,true))

  ***** assert(bf.*as,sparse(bf.*af,true))

  ***** assert(as./bf,sparse(af./bf,true),100*eps)

  ***** assert(bf.\as,sparse(bf.\af,true),100*eps)

  ***** test

sv = as.^bf;

fv = af.^bf;

idx = find(af~=0);

assert(sv(:)(idx),sparse(fv(:)(idx),true),100*eps)

octave(1247) malloc: *** error for object 0x102fca40: incorrect checksum for freed object - object was probably modified after being freed, break at szone_error to debug

octave(1247) malloc: *** set a breakpoint in szone_error to debug

panic: Segmentation fault -- stopping myself...

Segmentation fault

On Jun 28, 2006, at 1:14 AM, David Bateman wrote:

Marius Schamschula wrote:

David,

octave-forge had nothing to do with the seg fault. The tests were run

prior to installing octave itself. The issues with the octave-forge

build were largely fixed by using CVS. The tests were run prior to

installing octave itself.

Ok, then the seg-fault really does exist and needs to be treated.

I just ran your test fusing the installed version of octave/octave-forge

and got the following:

octave:6> test('test_sparse.m','verbose',stdout)

processing test_sparse.m

/private/tmp/octave-2.9.6/test/test_sparse.m

  ***** test # segfault test from address@hidden <mailto:address@hidden>

n = 510;

sparse(kron((1:n)', ones(n,1)), kron(ones(n,1), (1:n)'), ones(n));

  ***** assert(spinv(sparse([1,1;1,1+i])),sparse([1-1i,1i;1i,-1i]),10*eps);

!!!!! test failed

error: `spinv' undefined near line 1 column 8

  ***** error sparse(1,[2,3],[1,2,3]);

spinv is definitely in 2.9.6 and so this must be some sort of

installation issue. Sorry the seg-fault is more important to fix, so can

you run the test again without octave-forge installed so that the first

error is the seg-fault.

D.

Marius

--

Marius Schamschula                               Webmaster

        The Huntsville Macintosh Users Group

                    www.hmug.org

webmaster at hmug dot org    marius at schamschula dot com