Re: Argument parser

[Marco Gerards]

"Yoshinori K. Okuji" <address@hidden> writes:

> On Tuesday 16 December 2003 19:10, Marco Gerards wrote:
> > One thing that will remain  is the structure that describes the
> > arguments, this will still look like this example:
> [snipped]
> > I believe the parser will be easy to use this way.  It still can be
> > full featured and GNU compatible (which is, as you might have
> > noticed, important to me). Do you like this approach better?  If you
> > can find any other problems I can fix it.  When we agree about the
> > parser I will implement it.
> 
> Yes, it's very nice. One possible problem would be how to deal with 
> multiple instances of the same option, like:
> 
> diff -x CVS -x '*~' foo bar
> 
> Normally, most commands assume that only the last specified option is 
> significant, so it might not be necessary to worry about this problem.

Oops.  I knew about this problem but I forgot to tell you about this.
The parser I proposed before didn't have this problem.  By not adding
support for multiple instances of the same option the parser will be a
lot simpler to use.

However, I can add support for the parser function I proposed before
that can optionally be used to parse the arguments in this situation.
I see no use for supporting this so I won't do this unless you ask for
it.

[...]

> Basically, POSIX mode stops parsing options as soon as encountering a 
> non-option argument, while GNU does not stop. If you want more info, I 
> can consult my copy of the POSIX document, or you can look at susv2 on 
> the web.

Ok.  I see no reason in supporting the POSIX way only in PUPA.

I assume what I proposed is ok and I will start implementing this
soon.

Thanks,
Marco