Message-ID: <B0000029294@stargate.astr.lu.lv>
From: "Andris Pavenis" <pavenis AT lanet DOT lv>
To: Eli Zaretskii <eliz AT is DOT elta DOT co DOT il>
Date: Tue, 5 May 1998 13:54:49 +0300
MIME-Version: 1.0
Content-type: text/plain; charset=US-ASCII
Content-transfer-encoding: 7BIT
Subject: Re: C++ debugging with GDB
CC: Nate Eldredge <nate AT cartsys DOT com>, djgpp-workers AT delorie DOT com
References: <199805050025 DOT AA015687941 AT typhoon DOT rose DOT hp DOT com>
In-reply-to: <Pine.SUN.3.91.980505112831.27724J-100000@is>
Precedence: bulk

Date sent:      	Tue, 5 May 1998 11:29:00 +0300 (IDT)
From:           	Eli Zaretskii <eliz AT is DOT elta DOT co DOT il>
Subject:        	Re: C++ debugging with GDB

> 
> On Mon, 4 May 1998, Andrew Crabtree wrote:
> 
> > Had to go to a really slow link to get it.
> 
> I usually go directly to ftp.gnu.org, and it usually works fast
> enough.  But then I mostly do this when North America is asleep.
> 
> Also, it might be that everybody and their dog were downloading that
> release together with you.
> 
> > It handled all of the cross stuff no problem.  I have 
> > everything built and will begin testing shortly.
> 
> I understand that you are building it on a Unix host.  Does this mean
> that it won't build on DOS/Windows?
> 

I didn't met any problems to build gdb-4.17 for Linux. I used sources of gdb-4.17 
also to rebuild RHIDE under Linux. All worked Ok

For DJGPP things were much worse (I tried only gdb-4.16.86 in March). After some
hacking as target i[3456]86-pc-msdosdjgpp was not supported. I took some files from 
port of 4.16 and after some more hacking and I suceeded to build it but it still was 
seriously broken:
   - viewing call stack (bt) was broken. To force it to compile I had to take
     some parameters from ceiling and of course they were wrong. I simply don't know
     what I must put there...
   - Previously I was not able to get normal reaction to SIGINT in DJGPP port of
     gdb-4.16 when I built it from sources (The process I'm debugging does no
     more exist when I'm getting next gdb prompt after SIGINT, this is not so
     in binary archive built by Robert [gdb416b.zip] and in binaries I have built
     under Linux

I didn't try to mess with development snapshots in DOS more after March.
     
> > GDB on the other hand appears to be totally broken (as reported
> > by somebody else).  I mean totally.  Not ever an option
> > to configure djgpp in any type of way.
> 
> GDB 4.16 was configured by a series of batch files, IIRC.  I guess
> your best bet would be to fetch gdb416s.zip from any DJGPP site and
> look into the djgpp subdirectory inside there, to see how it was
> done.  I'm not suggesting that this is how you should do it for 4.17,
> but it might answer all kinds of questions.  I have 4.16 source tree
> on my machine, so I could try to help you out with some questions, if
> Robert is unavailable or doesn't remember what he did.
> 

One suggestion. I think we should get rid of these batch files. I think
that the way that is used in ports of gcc-2.8.0 and gcc-2.8.1 is better:
   additional bash scripts that sets needed environment parameters and
   invokes configure and make.

Perhaps it is not so terrible to ask poeple who want to build some packages
from sources to install bash


Andris