Typesafety versus Worse is Better - was Re: Fwd: is there any word processing software for the pdp11?

David Brownlee abs at absd.org
Thu Dec 4 03:25:03 CST 2014

On 4 Dec 2014 08:47, "Mark Wickens" <mark at wickensonline.co.uk> wrote:
> Maybe the *real* answer to the question is that the smart programmer would
> pick two languages to get the job done... systems level routines in C or
> assember, high-level routines in a more domain-specific language. I'm
> thinking here of VAXTPU for example. VMS, with the common-calling
> makes this a trivially viable option. Personally I can't say what language
> I would use for applications level programming without more investigation.
> I understand the benefits of knowing a language. It would be easier for me
> to write in C because I don't have the learning curve. For the purposes of
> discussions it would better to have the same view of the programming doing
> the task as a reader of a scientific paper. Highly intelligent but with no
> domain specific knowledge.
> Has anyone considered whether it would be possible to port lua to the PDP?
> It has a compact run time system. There is a microprocessor port
> http://www.eluaproject.net/

As an aside on embedded lua into an OS to provide a safer programming
environment for certain tasks including in kernel, I think that has been
done with a couple of Linux distribution and NetBSD will ship with it as

I think this is akin to how kde4 switched to python for a whole set of
scripting functionality in the toolkits and widgets. For the specific tasks
performance was effectively unchanged from the previous C++ code, and the
maintainability and (I think) footprint went down - quite a lot of complex
c++ code went away.

Over time the boundary between the higher and lower level languages should
be pushed down,  keeping C in that particular subset of tasks where its
characteristics give it a worthwhile edge (I really like C, but I would be
among the first to admit that for many tasks making C faster than a more
suited high level language would effectively require writing massively more
code, in effect implementing significant chunks of that other language's
data handling)

I'm not suggesting that lua or python are really the ideal language for
large systems, but I think it's a step in the right direction


More information about the cctalk mailing list