Path: utzoo!attcan!uunet!mcsun!ukc!stc!stl!argon.siesoft.co.uk!neon!dcb@oxygen.siesoft.co.uk
From: dcb@oxygen.siesoft.co.uk (David Binderman)
Newsgroups: comp.lang.c++
Subject: NIH 2.204 & ATT 2.1 problem
Message-ID: <1871@neon.siesoft.co.uk>
Date: 31 Jul 90 15:24:33 GMT
Sender: usenet@neon.siesoft.co.uk
Organization: Siemens System Development Group, Reading, UK.
Lines: 30

Hello there,
I have found a problem in building NIH 2.204 with ATT 2.1, instead
of the usual ATT 2.0.

I find that the test suite programs for stackproc and process
execute to completion, but the message strings written out are wrong
because the msg variable is overwritten with junk, when compiling
with 2.1.

The same NIH source works fine thru 2.0, so the underlying 
C compiler must be ok.

I realise that NIH 2.204 was written a long time before ATT 2.1, but can
anyone give me any clues on why ATT 2.1 causes so much trouble ?

I have this same problem on two different ports of ATT 2.1 onto two different
boxes (nearly ANSI C on Intel 386 & K&R C on NSC 32000), so I don't
think my ports of ATT 2.1 are wrong.

Is this problem solved in the NIH 3.0 corresponding to the Gorlen, Orlow
& Plexico book ? Is this NIH 3.0 easy for a UK site without Internet
access to get ?

Enquiring minds with newly broken code want to know,
Regards

--
David C Binderman, Siemens SDG, 65 - 73 Crockhamwell Road, Reading, Berkshire
RG5 3JP, UK  dcb@siesoft.co.uk  ...uunet!mcsun!ukc!siesoft!dcb  +44 734 443025
to the sound of Fat Lucy doing Nessun Dorma: "This isn't Rawhide !"