Re: Third input argument to __makeinfo__

[c.]

Rik,

On 10 Sep 2011, at 18:43, Rik wrote:

> On 09/09/2011 04:10 PM, Carlo de Falco wrote:
>> After a bit more testing, the attached is the best I managed to come up with.
>> I'll wait longer before pushing this time ;)
>> c
> 9/10/11
> <...>
> 3) The for loop over patterns won't work as intended
> 
> +  args = regexp (text, see_also_pat, 'tokens');
> +  for ii = 1:numel (args)
> +    if (! isempty (args{ii}))
> +      expanded = feval (see_also, strtrim (strsplit (args{ii}{:}, ',',
> true)));
> +      text = regexprep (text, see_also_pat, expanded);
> +    endif
> +  endfor
> 
> The first time through the loop, args{1}{:} will be expanded into a
> replacement string.  But the regexprep will replace all @seealso references
> with the expanded string from the first instance.  The second time through
> the loop the regexprep will match nothing.

oops, you are right!
I used this instead:

see_also_pat = '@seealso *\{([^}]*)\}';
args = regexp (text, see_also_pat, 'tokens');
for ii = 1:numel (args)
expanded = fsee_also (strtrim (strsplit (args{ii}{:}, ',', true)));
text = regexprep (text, see_also_pat, expanded, 'once');
endfor

> On the other hand, it occurred to me that we never have more than 1
> @seealso reference per function.  Thus, I took out the for loop and just
> put in a check to make sure that we never exceed 1 @seealso reference per
> function.  I think this is a fair tradeoff because this function is not a
> general purpose one and does not need to be written to handle any generic
> input.

from a quick grep it seems this happens in some octave-forge functions

> Feel free to use this patch or merge it with your own.  I'm off on vacation
> for a week so go ahead and check in a solution to Mercurial whenever you
> have one.  I won't be around for review.

I merged in all your suggestions except removing the handling of multiple 
@seealsos
and pushed a changeset.

c.