Date:	Sat, 12 Sep 1998 12:01:25 +0200 (WET)
From:	Andris Pavenis <pavenis AT lanet DOT lv>
To:	"Mark E." <snowball3 AT usa DOT net>
cc:	DJ Delorie <dj AT delorie DOT com>, djgpp-workers AT delorie DOT com
Subject:	Re: Wide string functions available
In-Reply-To:	<199809111938.TAA144020@out4.ibm.net>
Message-ID:	<Pine.A32.3.91.980912115418.35996C-100000@ieva05.lanet.lv>
MIME-Version:	1.0

On Fri, 11 Sep 1998, Mark E. wrote:

> > 
> > until it stabilizes.  Plus, 2.02 is in (almost) beta, which means no
> > new functionality until it releases.
> > 
> 
> In that case, then include include/wchar.h and include/wctype.h in 
> the distrubution, but stripped of the function prototypes but leaving 
> everything else. That way, when stdlibc++ is upgraded to the 
> current version of SGI's STL, compiling their string class won't 
> generate a complaint about missing headers.
> 

I better suggest to use some define (e.g. -D__EXPERIMENTAL__ or something
like) to enable experimental stuff. Such way should allow us to hide
functionality that is not tested enough from most users and and still 
have this all in libc.a to allow using it by those who understands what 
they are doing. So these features will be tested more widely and and 
perhaps will sooner ready for general user.

One more suggestion. Such experimental header can issue warning that
they contain experimental stuff.

For example even stable Linux kernels (2.0.XX) contains some features
that are marked as experimental.

Andris

- Raw text -