Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Posting-Version: version B 2.10.1 6/24/83; site ritcv.UUCP
Path: utzoo!watmath!clyde!burl!ulysses!allegra!mit-eddie!godot!harvard!seismo!rochester!ritcv!mjl
From: mjl@ritcv.UUCP (Mike Lutz)
Newsgroups: net.lang.c
Subject: Re: C stack frame sizes
Message-ID: <1434@ritcv.UUCP>
Date: Sat, 15-Dec-84 11:26:01 EST
Article-I.D.: ritcv.1434
Posted: Sat Dec 15 11:26:01 1984
Date-Received: Mon, 17-Dec-84 03:47:35 EST
References: <18092@arizona.UUCP>, <6255@brl-tgr.ARPA> <4722@utzoo.UUCP> <264@ist.UUCP>
Organization: Rochester Institute of Technology, Rochester, NY
Lines: 15

>One point which hasn't been made about the issue of stack frame sizes,
>... [is that] if you call a function right at the start of your program which
>uses lots of stack space it stays allocated.

Actually, this was use to advantage in the original Pascal compiler
system from Amsterdam.  One of the C programs in the set (I forget
which one) used *gobs* of data space, so to make it run on a PDP-11 the
I/O buffers were declared as local variables in the main function, and
thus were on the stack.  Given the constraints imposed by the PDP-11
MMU, this actually resulted in a net gain in useful data memory as
compared to allocating buffers dynamically using sbrk.
-- 
Mike Lutz	Rochester Institute of Technology, Rochester NY
UUCP:		{allegra,seismo}!rochester!ritcv!mjl
CSNET:		mjl%rit@csnet-relay.ARPA