[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: New language syntax
Points on getting it accepted (based on my experience)
(1) If it's something new to people they want their hand
held. A packaged solution that gets you to a desired result
is needed. Inmos came close with B008, hardware, code: what
I think was their shortcoming was they didn't "reach for
the world" and offer occam tools for the host system, so
the "occam world" was automatically (and unnecessarily)
(2) More recent efforts have attacked that problem, but
the beautiful hardware flexibility offered by B008 and
TRAMs is a thing of the past.
(3) The world needs us. System and driver code (the stuff
at asynchronous hardware/software interface) is a
nightmare and continually getting worse. TRAMs used to
make that easy.
(4) We can do modules that are hardware-equivalent. Nobody
else can. Go with our strengths.
(5) A C-appearing preprocessor syntax is a convenience.
A little occam-friendly OS that writes what the rest of
the world calls system or kernel code is a necessity. If
we ride other people's horses we can't solve (3).
(6) The only way to sell it is to develop it and do
things with it - easily - because otherwise nobody will
believe we can do things with it easily.
>From IanPage99@xxxxxxx Wed Dec 20 02:36:35 2000
Date: Wed, 20 Dec 2000 05:36:00 EST
Subject: Re: New language syntax
To: mdp2@xxxxxxxxxxxxxxxxxxxxxxxx, tjoccam@xxxxxxxxxxxxx
Content-Type: text/plain; charset="US-ASCII"
X-Mailer: AOL 5.0 for Windows sub 105
In a message dated 12/19/00 12:48:53 GMT Standard Time,
> >Tony Hoare's assertion that "occam really should have *been* C"
> I don't understand what this is supposed to mean, but do remember life
> at INMOS before there was an occam and one of things Tony Hoare did for
> us was to propose a language that looked very much like BCPL.
I'm sure that Tony meant that 'occam should have been C' in the same
way that he tried to make it BCPL in the early days before C was
the force it later became. If C had been the dominant language,
I have little doubt the language he tried to push Inmos towards would
have looked like C.
I always believed, and now because of the Handel-C experience I
am certain of it, that the biggest problem that occam had in gaining
major acceptance was that it was simply a new language. The effort
and resources needed to get *any* language accepted on a global
basis are immense.
So, if you can embed your favourite programming system inside
an existing language it will be the easiest to have accepted (even
though your principles are most compromised with this solution).
The next next best strategy is to keep it as close as possible
to C (or C++, or java, or whatever) and a hand-waving "well, its
really just C isn't it?" actually does cut some ice with both
programmers and managers.
Going for a new language is the hardest of all, and you need
the same sort of resources behind it as were needed to get C
or java accepted (a massive pull - like a free operating system
too good to ignore written in it - and massive support in the field).
And don't expect an overnight success even with those things
behind you; its bound to be a long hard struggle.
Founder, Celoxica Plc (formerly ESL),
8 Milton Park, Abingdon, Oxfordshire OX14 4RT
+44 1235 863 656, fax 863 648