Message-ID: <B0000054516@stargate.astr.lu.lv>
From: "Andris Pavenis" <pavenis AT lanet DOT lv>
To: djgpp AT delorie DOT com, alextxm AT tin DOT it
Date: Wed, 9 Dec 1998 15:17:35 +0200
MIME-Version: 1.0
Content-type: text/plain; charset=US-ASCII
Content-transfer-encoding: 7BIT
Subject: Re: DJGPP 2.02 and GCC 2.8.1 PROBLEM [urgent!] [SOLVED!]
References: <Version DOT 32 DOT 19981208202949 DOT 00e1a2a0 AT box1 DOT tin DOT it>
In-reply-to: <Pine.SUN.3.91.981209123242.7705M-100000@is>
X-mailer: Pegasus Mail for Win32 (v3.01d)
Reply-To: djgpp AT delorie DOT com

On 9 Dec 98, at 12:33, Eli Zaretskii wrote:

> 
> On Tue, 8 Dec 1998, Alessandro Pisani wrote:
> 
> > >If that doesn't help, set LFN=n in the environment and see what happens 
> > >then.
> > 
> > WOA !!! THIS WORKS !!!!!!! TA-DAAA !!! 
> > Keeping djgpp 2.02 djgpp.env and setting LFN=n *in the environment* gcc
> > 2.8.1 works fine.
> 
> This is a severe bug, and it's probably triggered by something
> specific to your system setup.  Please report the information I asked
> for: the size and time stamp of gcc.exe you are using.  It is
> impossible to debug this problem further without knowing exactly which
> version of the binary you use, since otherwise there is no way to
> deduce the locus of the bug from the crash traceback you posted.

Unfortunatelly I don't have unstripped binaries of gcc-2.8.1 any more.
Also reproducing them would be difficult as I have changed not only
gcc version (now I'm using egcs-1.1.1) but also upgraded to djdev202.

Therefore I suggest to repeat tests with binary I made available at:
	http://www.lanet.lv/~pavenis/gcctest.zip

Archive contains freshly compiled gcc.exe with debug information
so it will be possible also to use symify.exe or even run it under 
gdb.

> 
> > gxx.exe that was working fine was the djgpp 2.02 one (it works fine both
> > with LFN=y and LFN=n)
> 
> Once again, I asked whether this is gxx.exe from djdev202.zip, or the
> one from gcc281b.zip.  Please report the size and time stamp of
> gxx.exe as well.
> 
> > *predefines:
> > -Dunix -Di386 -DGO32 -DMSDOS -DDJGPP=2 -DDJGPP_MINOR=1
> > 
> > should I change it in -DDJGPP_MINOR=2 for DJGPP 2.02 ?
> 
> Yes, you should.  However, if you have specs in your lib subdirectory,
> GCC will use that one, and it already sets DJGPP_MINOR to 2.

There is more things with specs from gcc281b.zip:
     file tells to link in crtf.o which is redundant (and harmless) for 
     DJGPP 2.02.
> 
> > Also, the file djgpp.djl in the same dir is longer than the djgpp.djl
> > in the %djdir%\lib, that is the DJGPP 2.02 one. Should I substitute the
> > GCC one with the DJGPP 2.02 one ?
> 
> You should be able to use either one.
> 
> > PS: I hope that a fix to the buggy LFN-gcc 2.8.1/djgpp 2.02 behaviour
> >     will be soon available ;)
> 
> Only if you help us debug this problem.  It obviously never happened
> to anybody else, although there are quite a few people who work with
> beta releases of v2.02 and GCC 2.8.1 with LFN enabled for quite some
> time.  We need your help to debug this.
> 

I also was not able to reproduce the problem. So I can only suggest 
to try debugging it using gcc.exe executable I mentioned above.

Andris