Path: utzoo!utgpu!news-server.csri.toronto.edu!mailrus!cs.utexas.edu!tut.cis.ohio-state.edu!pt.cs.cmu.edu!wb1.cs.cmu.edu!ram
From: ram@wb1.cs.cmu.edu (Rob MacLachlan)
Newsgroups: comp.lang.lisp
Subject: Re: Floating point in CL?
Message-ID: <10621@pt.cs.cmu.edu>
Date: 1 Oct 90 03:11:36 GMT
References: <TIM.90Sep27140541@watt.cstr.ed.ac.uk> <10607@pt.cs.cmu.edu> <JWZ.90Sep29124432@kolyma.lucid.com>
Organization: Carnegie-Mellon University, CS/RI
Lines: 20

>From: jwz@lucid.com (Jamie Zawinski)
>Rob writes:
>> You should also make sure your non-float arithmetic is getting open-coded,
>> i.e. all your DOTIMES variables should be declared FIXNUM.  
>
>Since the result of changing the value of a dotimes counter is undefined,
>isn't it reasonable for the expansion of dotimes to declare it fixnum if
>it can tell that it will never reach the bignum range?
>

Yes, if it can tell.  It should really be adequate for the repeat count to
be known to be a fixnum, but in practice not all compilers can hack that.
And unless the repeat count is obviously a fixnum (such as a constant), it
may in any case be necessary to add some sort of declaration.

Although in practice I am sure it is rare (unheard of) for a dotimes count
to not be a fixnum, with today's machine speeds, it is not a totally
nonsensical idea.

  Rob (ram@cs.cmu.edu)