From mboxrd@z Thu Jan  1 00:00:00 1970
Message-ID: <8a398537df07d5cc97101f691a5932e2@plan9.bell-labs.com>
From: presotto@plan9.bell-labs.com
To: 9fans@cse.psu.edu
MIME-Version: 1.0
Content-Type: text/plain; charset="US-ASCII"
Content-Transfer-Encoding: 7bit
Subject: [9fans] long long warning
Date: Fri, 15 Mar 2002 09:07:51 -0500
Topicbox-Message-UUID: 678254e4-eaca-11e9-9e20-41e7f4b1d025

Danger Will Robinson! Danger, Danger!

The Plan 9 long long implementation is at best a
convenience.  Because every op results in a subroutine
call, the performance is an order of magnitude worse than
it could be.  If you have the odd long operation to do,
it's useful.  Trying to do compute intensive work with
it is not playable.  For example, the extended arithmetic
package, libmp, does everything with unsigned longs.

We really could use a compiler with long long
implemented as inline ops.  I'ld volunteer but
my stack is already too deep.  Maybe in six
months.  If anyone could do it sooner, and make
the corners like ?: work, it would be appreciated.