Re: MacOS X 10.3.9 build but load-save.m test failure

I'm cutting and pasting from the mail archives on the website to respond 
to these mails as I'm the wrong side of my corporate firewall to get my 
mails... So forgive the fact that I haven't  properly quoted the text 
I'm responding to.. Dmitri, your "whos" test on binary.mat basically 
proves that the problem with binary.mat is independent of zlib or not, 
and in fact is the fault of octave_sparse_matrix:load_binary...

It would be good if you could use gdb to break in 
octave_spare_matrix::load_binary and check the values of nc, nr, and nz.
If you don't want to use gdb just add

std::cerr << "nr,nc,nz: " << nr << " " << nc << " " << nz << "\n";

just before the definition of the SparseMatrix m.... In any, although 
this problem is closer to being understood, I still fail to see why the 
current code doesn't work on BE platforms...

Ok, I see the problem for the mat7 files, the following patch works for 
me....

D.

--
David Bateman                                David.Bateman@xxxxxxxxxxxx
Motorola Labs - Paris                        +33 1 69 35 48 04 (Ph) 
Parc Les Algorithmes, Commune de St Aubin    +33 1 69 35 77 01 (Fax) 
91193 Gif-Sur-Yvette FRANCE

The information contained in this communication has been classified as: 

[x] General Business Information 
[ ] Motorola Internal Use Only 
[ ] Motorola Confidential Proprietary

*** src/ls-mat5.cc~     2005-04-26 21:28:29.000000000 +0200
--- src/ls-mat5.cc      2005-04-28 23:05:48.232422117 +0200
***************
*** 441,446 ****
--- 441,449 ----
                      X_CAST (Bytef *, inbuf), element_length) !=  Z_MEM_ERROR)
        {
          // Why should I have to initialize outbuf as I'll just overwrite!!
+         if (swap)
+           swap_bytes<4> (tmp, 2);
+ 
          destLen = tmp[1] + 8;
          std::string outbuf (destLen, ' ');