Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!watmath!clyde!burl!ulysses!mhuxl!ihnp4!zehntel!hplabs!sri-unix!rbbb@rice
From: rbbb%rice@sri-unix.UUCP
Newsgroups: net.unix-wizards
Subject: Re: abnjh.490 Tapes on Unix
Message-ID: <17425@sri-arpa.UUCP>
Date: Sun, 11-Mar-84 03:57:27 EST
Article-I.D.: sri-arpa.17425
Posted: Sun Mar 11 03:57:27 1984
Date-Received: Tue, 13-Mar-84 08:31:48 EST
Lines: 26

From:  David Chase 

FLUSH THE OLD!!!  The SleazeNix paradigm (flaky kernel, no design) SUCKS!!
Rewrite the flaky /dev/ralph-and-fido dependent programs, along with the
malloc-free-realloc dependent programs.  If any one really cared about
software development and progress, these things would have been done long
ago.  I'm tired of having to build and collect my own stable of "real"
software (stuff that fails gracefully, stuff that is flexible and easy to
use, stuff that knows that is running on a VAX, not a PDP-7).

(Flame off --)

I would very very very much like to see an extended standard library to do
things like Block Moves, Substring generation, extraction, etc.  This
standard library would come with a REAL definition and portable C code,
but Unix implementors would be free (encouraged) to code up efficient
versions in assembler.  This was done for BCPL, with good (tasteful and
useful) results.  I wrote some of these myself, resulting in cleaner,
faster code (For instance, the VAX movc3 instruction runs 7 times faster
than anything the C compiler would generate for me).  I realize that
programming in assembly language gives most Unix jocks the trembles, but
you only have to do it once correctly, and then it's all over.

(Not an assembler hacker, just a fan of speed and abstraction)

drc