X-Authentication-Warning: delorie.com: mail set sender to djgpp-workers-bounces using -f
Date: Thu, 30 Sep 2004 14:26:57 -0600
From: Brian Inglis <Brian DOT Inglis AT SystematicSw DOT ab DOT ca>
Subject: Re: ANNOUNCE: DJGPP port of GCC-3.4.2
In-reply-to: <200409301704.i8UH4nUu020036@speedy.ludd.ltu.se>
To: djgpp-workers AT delorie DOT com
Message-id: <57nol0hi6o2nat28e5vn8i7muv1km25oqd@4ax.com>
Organization: Systematic Software
MIME-version: 1.0
X-Mailer: Forte Agent 1.93/32.576 English (American)
Content-type: text/plain; charset=us-ascii
References: <200409102121 DOT 19881 DOT pavenis AT latnet DOT lv>
 <200409301704 DOT i8UH4nUu020036 AT speedy DOT ludd DOT ltu DOT se>
Content-Transfer-Encoding: 8bit
X-MIME-Autoconverted: from quoted-printable to 8bit by delorie.com id i8ULRX5f005388
Reply-To: djgpp-workers AT delorie DOT com
Errors-To: nobody AT delorie DOT com
X-Mailing-List: djgpp-workers AT delorie DOT com
X-Unsubscribes-To: listserv AT delorie DOT com
Precedence: bulk

On Thu, 30 Sep 2004 19:04:49 +0200 (CEST), ams AT ludd DOT ltu DOT se wrote:

>I'm CCing djgpp-workers too.
>
>According to Andris Pavenis:
>> On Friday 10 September 2004 20:53, you wrote:
>> > In article <200409101743 DOT i8AHhRmY028387 AT delorie DOT com> Andris Pavenis 
>> <pavenis AT latnet DOT lv> wrote:
>> > > Note for users of C++ IO classes fstream, ifstream, ofstream
>> > > ============================================================
>> > >
>> > > There is a regression against GCC versions 2.95.3 and
>> > > earlier: Member functions tellp(), tellg(), seekp() and seekg()
>> > > are broken when stream is opened not in binary mode. If You are going
>> > > to use any similar functions You must open stream in binary mode.
>> >
>> > Perhaps you should add that g++ generates opcodes for 486 or later
>> > (i. e. doesn't support 386s) unless that has been corrected?
>> 
>> I'm not sure, but as far as I understand, it could possibly work for 386, if 
>> rebuilt from sources for i386-pc-msdosdjgpp. 
>
>Do I understand you correctly that you know that your gpp package
>doesn't work on 386s? "Work" here means that the compiler _and_ the
>executables it builds do run on a 386. (I know that some older version
>didn't make workable 386 executables.)
>
>No, I misunderstood something:
>	Then I guess we don't know if it does work or not. But 1 below
>	should be considered.
>
>Yes, that's what you're saying:
>	Well, IMO:
>	1. You really really really should point out that things built
>	   with gpp only is for 486 (or Pentium or whatever) or later.
>
>	2. Shouldn't it be compiled for i386-pc-msdosdjgpp then so
>	   that it potentially does work 386s?
>
>If the bug still is in gpp even when built for i386-pc-msdosdjgpp,
>then we have to wait to get it fixed, but 1 still applies. Plus we
>might need to change some texts which say "DJGPP requires at least a
>386" adding ", 486 (or whatever) for C++ programs".

'info gcc invoking submodel i386' says under -mtune:

    "While picking a specific CPU-TYPE will schedule things
     appropriately for that particular chip, the compiler will not
     generate any code that does not run on the i386 without the
     `-march=CPU-TYPE' option being used."

There was a problem noted last year with 486 lock codes being
generated in parts of libstdc++ causing SIGILL illegal opcode errors,
that I think was worked around with a rebuild. 
If the problem still exists, that's a g++/libstdc++ bug and not a
limitation of DJGPP, and should be noted as such, with a Bugzilla
link, if one exists; was it reported?  

-- 
Thanks. Take care, Brian Inglis

Business:  +1(403)547-8816 	Brian DOT Inglis AT SystematicSW DOT ab DOT ca
Residence: +1(403)239-6520 	BWInglis AT Shaw DOT ca
Cellular:  +1(403)708-7006 	Brian_Inglis AT CompuServe DOT com
Facsimile: +1(403)547-8816 	Brian_Inglis AT CSi DOT com