Re: Parsing "commands"

[Søren Hauberg]

son, 15 02 2009 kl. 23:50 -0500, skrev John W. Eaton:
> On 15-Feb-2009, Soren Hauberg wrote:
> 
> | lor, 14 02 2009 kl. 20:12 -0500, skrev John W. Eaton:
> | > Although I see all tests passing with this change, it is still quite
> | > experimental and probably not quite right yet, so I have not checked
> | > it in.  But it would help for it and the previous lexer/parser patch
> | > to get some additional testing.
> | > 
> | > Comments?
> | 
> | My intuition tells me that it's fairly hard to test such a change. So, I
> | think the best way to test it, is to use it in "real life" situations.
> | So, I'd say, commit the changes, to get more wide testing.
> 
> OK, you're probably right.  I checked these changes in.

Okay, so I'm playing a bit around with a recent checkout. I've created
the following function

  function retval = cmd (varargin)
    str = sprintf ('%s ', varargin {:});
    retval = length (str);
  endfunction

I see the following non-compatible situations

** Octave **
octave:4> cmd - 1
parse error:

  syntax error

>>> cmd - 1
          ^

** Matlab ** 
>> cmd - 1

ans =

    -1

** Octave **
octave:11> cmd + - 1
ans =  1

** Matlab **
>> cmd + - 1

ans =

    -1

** Octave **
octave:13> cmd += 2
parse error:

  syntax error

>>> cmd += 2

** Matlab **
>> cmd += 2

ans =

     5

But these are fairly odd situations. In general things seems to work
very well.

Soren