Newsgroups: comp.os.msdos.programmer
Path: utzoo!utgpu!watserv1!watmath!att!cbnewsd!cbnewsc!tjr
From: tjr@cbnewsc.att.com (thomas.j.roberts)
Subject: Re: Long wait times with Borland's integrated debugger
Organization: AT&T Bell Laboratories
Date: Fri, 2 Nov 90 13:41:16 GMT
Message-ID: <1990Nov2.134116.26086@cbnewsc.att.com>
References: <1990Oct30.184443.2766@athena.mit.edu>
Lines: 21

Yes, indeed, TC++ is *A LOT* slower than TC 2.0.

You really want MORE than 640K to use it effectively.
Adding 1 Meg of Expanded memory increases throughput so much
that my 16 MHz 8086 is significantly faster in compiling
than a 16 MHz 80286 without any EMS. The EMS improves both
compilation times and debugger response times.

On an 80386SX with fast IDE drive and 1.25 Meg disk cache, the
lack of EMS is not noticeable, and the faster processor makes
TC++ a joy.

BEWARE: TC++ has a bug when using EMS - it DOES NOT RELEASE THE EMS
WHEN EXITING TC++. This is a disaster if you run TC++ twice without
rebooting. There is a kludgy work-around - use an EMS debugger to
delete the first EMS handle (-511 on my machine) before running
TC++ - this assumes there are no other "permanent" EMS users on
the system, such as RAMDISKs.

Tom Roberts
att!ihlpl!tjrob