Re: 2.9.15 --> 3.0

[John W. Eaton]

On  7-Oct-2007, John Swensen wrote:

| John W. Eaton wrote:
| > OK, I have the following items for 2.9.15, then bug fixes only until
| > 3.0 (really, I mean it this time :-):
| >
| >   * GPLv3
| >   * diag should preserve class/type
| >   * filename completion (if possible)
| >   * verify that the new residue works (could use some tests)
| >   * improve residue doc string
| >   * interp2 bug
| >   * num2str bug
| >   * any other pending bug reports?
| >
| > Is there anything else?
| >
| > jwe
| >
| >   
| Here are 2 patches to fix the num2str bug.  One is the num2str fixes and 
| the other adds the strtrim() function.  The strtrim() function is almost 
| a complete copy of the deblank() function.  Let me know if you want any 
| other changes. 

I applied the patch, which seems to fix the problem of extra space in
the output.  We still have

| 2) num2str doesn't account for signed zeros while mat2str does:
| 
| octave-2.9.13.exe:50> num2str(-0,10)
| ans =          0
| octave-2.9.13.exe:51> mat2str(-0,10)
| ans = -0

I haven't looked closely, but mat2str is also a .m file, so I think
num2str shoudl also be able to ahndle -0 without any changes to the
C++ bits of Octave.

Thanks,

jwe