Date: Mon, 17 Nov 1997 21:44:01 -0800 (PST)
Message-Id: <199711180544.VAA11183@adit.ap.net>
Mime-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
To: muller AT janus DOT u-strasbg DOT fr (Pierre Muller), djgpp AT delorie DOT com
From: Nate Eldredge <eldredge AT ap DOT net>
Subject: Re: GDB and extended type
Precedence: bulk

At 05:30  11/17/1997 GMT, Pierre Muller wrote:
>>On 17 Nov 1997, Pierre Muller wrote:
>>
>>>     GDB v4.16 does not know how to handle the Intel internal real format 
>>> coded in 10 bytes !!
>>This is a known problem.  GDB was compiled with long double support 
>>disabled (by a sad omission).
>>>     Are there any plans to change that ?
>>You can do it yourself, if you need it badly enough.  Download the sources
>>(v2gnu/gdb416s.zip), edit the file gdb/config.h to change this line: 
>>        /* #undef HAVE_LONG_DOUBLE */
>>into this:
>>        #define HAVE_LONG_DOUBLE 1
>>then build gdb, and live happily ever after!
>  I did that but GDB still says 
>something like Don't know how to deal with float of 10 bytes
>
>  I took a look inside the sources and the long double type
>is expected to be twice the size of the double type (i.e. 2*8=16 not 10 !!)
>
>   So I still don't have the value of my extended !!
It isn't that simple, sorry. Eli neglected to mention that `long double' is
only supported with stabs debugging. Doing this requires getting the GCC
source, patching and recompiling it. I think Robert's web page has the patch
(http://www.tu-chemnitz.de/~sho/rho) or I can email it, if necessary. It's
fairly small.

Nate Eldredge
eldredge AT ap DOT net