Date: Mon, 14 Jun 1999 11:15:25 +0300 (WET)
From: Andris Pavenis <pavenis AT ieva01 DOT lanet DOT lv>
To: Eli Zaretskii <eliz AT is DOT elta DOT co DOT il>
cc: djgpp-workers AT delorie DOT com
Subject: Re: Small changes to DJGPP.ENV
In-Reply-To: <Pine.SUN.3.91.990614103651.21962P-100000@is>
Message-ID: <Pine.A41.4.05.9906141101410.71812-100000@ieva01.lanet.lv>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Reply-To: djgpp-workers AT delorie DOT com
X-Mailing-List: djgpp-workers AT delorie DOT com
X-Unsubscribes-To: listserv AT delorie DOT com
Precedence: bulk



On Mon, 14 Jun 1999, Eli Zaretskii wrote:

> 
> On Mon, 14 Jun 1999, Andris Pavenis wrote:
> 
> > Adding bin/cpp.exe to gccXXXb.zip will break gcc unless we'll remove
> > COMPILER_PATH there.
> 
> >From your original message I understood that this is only true for
> the EGCS distribution, gcc281b.zip is unaffected by this.  Did I 
> misunderstand you?

Specifying 

#define MD_EXEC_PREFIX "$DJDIR/bin"

does almost the same as specifying COMPILER_PATH except of search order:
	when specifying COMPILER_PATH=$DJDIR/bin gcc searches $DJDIR/bin
	for executables before gcc version specific directory which breaks
	things as I noted before if $DJDIR/bin contains cpp.exe.
	Specifying $DJDIR/bin in MD_EXEC_PREFIX the search order is
	different (at first version specific directory, then $DJDIR/bin
	and only after that DOS PATH). I think second way is correct

As I said I have added MD_EXEC_PREFIX in gcc configuration not later as
last update of port of gcc-2.8.1 in June 6, 1998 (tested as I had source
archive on one my CD). So we have it in almost all updates of port of 
gcc-2.8.1  

> > So if we leave things as they are it will be more hard to add bin/cpp.exe
> > if we'll need it for RSXNTDJ or something like in future. Things that can
> > break something should be fixed if possible. 
> 
> How can we be sure that removing it won't break more things?  This is my 
> primary concern for v2.03: I want a stable bug-fix release that would not 
> break anything.

I think we should not expect any problems from that with gcc-2.8.1 and
egcs-1.1.2 (and also with upcoming gcc-2.95). Removal of [gcc] and [cpp]
sections were recommended by readme files in gcc-2.8.1 

> If COMPILER_PATH is bad today, wouldn't it affect people who use GCC 
> 2.8.1, or install RSXNTDJ?  If so, I wonder why don't we hear about such 
> problems on c.o.m.d.  Perhaps some other factors are involved here?
> 

Earlier gcc versions didn't install $prefix/bin/cpp.exe which is DIFFERENT
from $prefix/lib/gcc-lib/$target/$version/cpp.exe. Therefore we didn't saw
these problems before. 


Andris