Fwd: Defect in XRAT Utility Syntax Guidelines

[Bruce Korb]

Bruce Korb wrote:
> 
> Geoff Clare wrote:
> 
> > The problem with options being recognised after operands, as I see it,
> > lies mainly in accidental or unexpected occurrences.  For example on
> > a UNIX system this is safe:
> >
> >     rm -f a.out *.o
> >
> > but on a GNU system it isn't, because if a file called "-X.o" exists
> > then rm would report the -X as an invalid option.
> 
> Geoff is completely correct here.  I've never liked the option reordering
> stuff myself, but I'm not quite Quixotic enough to advocate changing that
> behavior.  Instead, I would like to suggest that in the reordering process,
> any argument found past the "normal" end of options be tested for file
> existence.  I.e., if ``stat(arg_string, &stat_buf)'' returns zero, then
> the argument gets left with the operands and not moved in with options.
> Sorting arguments is icky anyway.  This is only a little bit ickier. :)
> 
> Cheers - Bruce