Re: Re: RE: Austin Interps 211

[David Korn <dgk@xxxxxxxxxxxxxxxx>]

cc:  austin-group-l@opengroup.org
Subject: Re: Re: RE: Austin Interps 211
--------


Roger Marquis <marquis@roble.com> wrote

> No disputing that, however, my concern is in regards to existing shells
> (not existing scripts) being unable to correctly interpret scripts written
> in the future, due to the proposed change.
> 
> 

Implementations are expected to change to handle new requirements.
When some shells were written there was no such thing as multibyte
characters.

Whether the change is in the shell or in printf, implementations must
change.  There are three main advantages to putting the change in
the shell rather than in printf.

1.      Other commands would benefit from being able to specify
        unicode arguments. 

2.      The mechnasm is uniform and would follow the existing and
        future C standards.

3.      $'...' is already existing practice in the most widely used
        shells so that we would be standardizing existing practice
        rather than inventing something new.

David Korn
dgk@research.att.com