Message-Id: <199910312122.QAA01510@rochester.rr.com>
X-Mailer: exmh version 2.1.0 09/18/1999
To: djgpp-workers AT delorie DOT com
cc: DJ Delorie <dj AT delorie DOT com>, leisner AT rochester DOT rr DOT com
Subject: Re: -g vs -s 
In-reply-to: Your message of "Sun, 31 Oct 1999 09:43:01 +0200."
             <Pine DOT SUN DOT 3 DOT 91 DOT 991031094239 DOT 2988K-100000 AT is> 
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Date: Sun, 31 Oct 1999 16:22:28 -0500
From: "Marty Leisner" <leisner AT rochester DOT rr DOT com>
Reply-To: djgpp-workers AT delorie DOT com


Yes, stripping prevents core dumps from giving any useful information...

When you give -g, you can get line number/parameters in the core dump.

When you do nothing, you get minimal information (i.e. the stack backtrace,
no parameters).

Experience shows the difference between -s and no -s is about 10%.

On the other hand, -g often inflates binaries 1000%.  Also the
threaded -fomit-frame-pointer makes stack backtraces impossible...it
gives marginal code speedup but makes anything meaningful impossible
(My production Linux system often core's once in a while with various
distributed commands...I would love marginal information in the core dump...

I haven't used djgpp in several years, if symify is the equavilent to reading
the stack on a core dump, by all means make it the default behavior...

Marty Leisner


Eli Zaretskii <eliz AT is DOT elta DOT co DOT il> writes  on Sun, 31 Oct 1999 09:43:01 +0200
     > 
     > On Thu, 28 Oct 1999, DJ Delorie wrote:
     > 
     > > > Can you do anything useful with a core dump (a minimal stack 
backtrace?)
     > > 
     > > There are no core dumps in djgpp; only symify.
     > 
     > Suppose we did have core files: how would this change things?  I'd
     > think it won't, since symify is the functional equivalent of the core
     > dump as far as the backtrace is concerned.  I don't think stripping
     > prevents core dump from happening on Unix, does it?