Problem with gcc 2.953 debug back-annotation (fwd)

[Eli Zaretskii]

Regarding the below: does anyone know whether GDB is supposed to work 
with programs where some of the object modules were compiled with 
-gcoff, others with -gstabs?  Or does this sound as a GCC or 
DJGPP-specific issue?

TIA

---------- Forwarded message ----------
Date: Tue, 3 Apr 2001 11:18:45 +0300
From: pavenis@lanet.lv
To: djgpp-workers@delorie.com
Subject: Re: Problem with gcc 2.953 debug back-annotation

On 3 Apr 2001, at 9:12, Andris Pavenis wrote:

> 
> 
> On Mon, 2 Apr 2001, Eli Zaretskii wrote:
> 
> > 
> > On Mon, 2 Apr 2001 pavenis@lanet.lv wrote:
> > 
> > > however I have examples where debugging
> > > is practically impossible when compiled with -gcoff both with gcc-
> > > 2.95.2 and 2.95.3, so it's not a regression
> > 
> > Can you post these examples?  Thanks.
> > 
> 
> It's real applications (C++), that uses large number of other stuff, so it
> would be difficult. However browsing output of 'objdump --debug' doesn't
> show anything bad. In real life after linking debugger wants to use some
> STL header file instead of real source (strange with -g)
> 

It seems to be binutils problem as presence of that problem depends 
on order of object files in command line when linking (when some of 
sources are compiled with -g and some with -gstabs+ or -gstabs, if all 
C++ sources are compiled with -g, the I have no problem)

Andris