X-Authentication-Warning: delorie.com: mail set sender to djgpp-workers-bounces using -f
From: Andris Pavenis <pavenis AT latnet DOT lv>
To: djgpp-workers AT delorie DOT com
Subject: Re: Fix for environ pollution
Date: Thu, 28 Oct 2004 16:54:51 +0300
User-Agent: KMail/1.7.1
References: <10408121516 DOT AA22369 AT clio DOT rice DOT edu> <411C4D77 DOT A09F108B AT yahoo DOT com> <200408130519 DOT i7D5J9sU013440 AT envy DOT delorie DOT com>
In-Reply-To: <200408130519.i7D5J9sU013440@envy.delorie.com>
Cc: DJ Delorie <dj AT delorie DOT com>
MIME-Version: 1.0
Content-Type: text/plain;
  charset="iso-8859-1"
Content-Transfer-Encoding: 7bit
Content-Disposition: inline
Message-Id: <200410281654.51361.pavenis@latnet.lv>
Reply-To: djgpp-workers AT delorie DOT com

On Friday 13 August 2004 08:19, DJ Delorie wrote:
> > The point is that 'environ' is not in static storage, it is only
> > in the parameter list of main, under any name you like.  Or not.
>
> I think you're missing the point that "environ" *is* defined to be a
> global symbol in the runtime, not just a parameter to main().  It
> happens to be passed to main (at which point it is effectively
> nameless as far as the runtime is concerned), but for historic reasons
> we must provide a global symbol "environ" as well, even without any
> #includes.  The only way to do this cleanly is with PROVIDE.

It causes problems and will cause many complaints due to presence
of GCC own djgpp-x.djl. As result _environ stays unresolved if linker
script provided with GCC is being used, which is default for any recent ports
of GCC.

I think, that a better way is to use a separate source file in libc which 
contains definition of _environ only. Perhaps it would be best to revert 
the change of lib/djgpp.djl and put _environ in a separate fiel.

Andris