Re: nlinfit in octave

[Olaf Till]

Hi Asma,

I have made some changes to statget/statset/__all_stat_opts__, mainly
in the copyright. But before I change or look at anything else there
is still something to do for you.

In the testblocks e.g. of statset, you can't test with options which
are never contained in __all_stat_opts__ persistant store because no
function (nlinfit is the only one) ever advertises them for
__all_stat_opts__.

If nlinfit advertises options due do a call from __all_stat_opts__, it
must use statset, not optimset, to create the structure (because
__all...opts__ must recognize a recursive call to return an empty
cell-array and so prevent ...set from verifying set options against
existing options).

The issues in the last two paragraphs should actually have been
apparent from running and checking the respective code ...

Please, do check your files for spurious whitespace.

When making the patch I saw that your git repository is not up-to-date
with the version now in the optim package (lsqcurvefit.m). This should
be corrected.

Regards,

Olaf

-- 
public key id EAFE0591, e.g. on x-hkp://pool.sks-keyservers.net

statistics-options.patch
Description: Text Data

signature.asc
Description: Digital signature