From: pavenis AT lanet DOT lv
To: Eli Zaretskii <eliz AT is DOT elta DOT co DOT il>, djgpp-workers AT delorie DOT com
Date: Thu, 9 Mar 2000 18:34:06 +0200
MIME-Version: 1.0
Content-type: text/plain; charset=US-ASCII
Content-transfer-encoding: 7BIT
Subject: Re: bug in gcc2952
Message-ID: <38C7EE9E.17188.F15CDE@localhost>
References: <38C7E560 DOT 8538 DOT CD4348 AT localhost>
In-reply-to: <Pine.SUN.3.91.1000309180337.3167E-100000@is>
X-mailer: Pegasus Mail for Win32 (v3.12c)
Reply-To: djgpp-workers AT delorie DOT com

On 9 Mar 2000, at 18:06, Eli Zaretskii wrote:

> > > 
> > > For that matter, is it safe to look at tooldir_prefix[1] and
> > > tooldir_prefix[2] without making sure that tooldir_prefix[] has enough
> > > characters in it?  What if tooldir_prefix[] is simply "/"?
> > > 
> > 
> > Perhaps it's Ok as gcc is looking whether tooldir_prefix is not an 
> > absolute path there (relative and absolute tooldir_prefixes are handled 
> > slightly differently). We should not treat d:foo or d: as absolute.
> 
> In my experience, it is safer to treat "d:foo" as absolute, not as 
> relative.  In particular, if relative directories get prepended ./ or 
> ../, then it will fail with d:foo.

tooldir_prefix is set at gcc build time. So treating c:/foo as absolute
path is required, but using c:foo there should be treated as error.

Andris