Path: utzoo!attcan!uunet!samsung!zaphod.mps.ohio-state.edu!mips!smsc.sony.com!dce
From: dce@smsc.sony.com (David Elliott)
Newsgroups: comp.sys.mac.programmer
Subject: Re: just to let you know...
Message-ID: <1990Jul8.224915.8737@smsc.sony.com>
Date: 8 Jul 90 22:49:15 GMT
References: <1990Jul7.025556.4968@smsc.sony.com> <102102@tiger.oxy.edu> <1990Jul5.150741.12535@smsc.sony.com> <9008@goofy.Apple.COM> <1990Jul6.183526.6460@eng.umd.edu> <1990Jul8.194321.9582@Neon.Stanford.EDU>
Sender: news@smsc.sony.com (Usenet News System)
Reply-To: dce@smsc.sony.com (David Elliott)
Organization: Sony Microsystems, San Jose, CA
Lines: 42


In article <1990Jul8.194321.9582@Neon.Stanford.EDU>,
philip@pescadero.Stanford.EDU (Philip Machanick) writes:
|> And getting the wrong answers quickly? Are C compilers any faster than
|> Pascal compilers of equivalent quality (whatever that means)? If
|> speed is really such a factor, I would think integrating the various
|> phases which are separately implemented in traditional C compliers
|> into one program would probably give a bigger speedup than the time
|> lost to better type checking. For that matter, is proper type checking
|> more expensive than implementing arcane rules for default type
|> conversions?

I'm not arguing that adding the checking would be wrong.  In fact, I
welcome it.  Then again, my Mac II is the slowest machine I have to
work on, the others being 7-10 times as fast, so it doesn't bother
me to have a program take a few seconds longer to compile.  On the
other hand, and this was my point, there are people who will argue that
the checking already exists in the form of lint, which is there when you
need it and not there slowing down the compiler when you don't.

You argue that one could speed up the compiler to offset the additional
cycles used to check for problems, and I agree.  Then again, if you
add the same speedups to the compiler and don't add the checking, you
get a faster compiler.  Again, when you need the extra checking, lint
is there.  Why add it?

I do see your point, and I really don't have a problem with adding this
kind of checking to the compiler.  Then again, the slowest machine I
use is a Mac II (everything else is 5-10 times as fast), and I don't do
enough development on it to care if it is a little slower.  But, there
are people out there who want that extra speed.  Isn't that one of the
reasons that the old LightSpeed C was popular?

(Of course, none of this takes into account that the problem that
started this whole discussion isn't caught by lint.  There are a number
of areas in C where lack of strictness gets people into trouble, but
that's a language issue.)

David Elliott
dce@smsc.sony.com | ...!{uunet,mips}!sonyusa!dce
(408)944-4073
"You know my motto: Forgive and uh... the other thing."