Re: [OctDev] setfields.m seems to tests for nargin incorrectly

[Andy A. <ad...@nc...>]

I've made a change to the CVS to support this. Note,
however, that s= struct() only works in newer versions
of octave (works in 2.9.2 and not in 2.1.42 - I don't
have others lying around to test)

setfields now reads:

    function s = setfields(s,varargin)
    if nargin == 0
      s= struct; % doesn't work on older versions of octave
    elseif rem(nargin,2) != 1,
      error('setfields: expected struct, key1, val1, key2, val2, ...\n') ; 
    endif
            
    for i=1:2:nargin-1
      if ! isstr(varargin{i}) ,
        error('setfields: called with non-string key') ; 
      else
        eval( ['s.',varargin{i},'=varargin{i+1};'] ) ;
        end
      end
    end




>> Ben Barrowes wrote:

OK.
I encountered this because optimset.m has the line 45:

opt = setfields ();

which caused the original setfields.m:

function s = setfields(s,varargin)
if rem(nargin,2) != 1,
      error('setfields: called with odd number of arguments\n') ;
endif

to complain.
nargin==0 should be a valid call to setfields, so maybe some easy fix
would be:

function s = setfields(s,varargin)
if nargin>0
  if rem(nargin,2) != 1,
        error('setfields: called with even number of arguments\n') ;
   endif
else
    s=struct;
end