[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Encodings.

Paul Schlie wrote:
> Record based text common on some pda's cell phones, etc. aren't files
> they're simple data base fields which are access through distinct
> api's which have no relationship to conventional C file functions for
> example.
> You'll like this (therefore), it's likely ideally necessary to define
> a common convention by which scheme may call C and/or Java foreign
> procedures ala a c-lambda function and implied related facilities for
> example, through which formatted text, numerical, and/or binary
> objects which may represent encoded images and/or icon values (in
> whatever format they require) may be passed back and forth. (back to
> srfi-50 I guess)

Huh? I can't parse that. I think you're saying that you'd use some text
layer on top of a foreign database function. If so: No, that's not the
best way to do it. Indeed, you can use standard I/O functions like READ
and DISPLAY on that kind of system. You just need a Scheme system that
understand the native text format, and a program that understands the
limits of the format. Cobol systems do this all the time.

There are *much* easier ways, language-wise, than what you seem to be
suggesting here. But you can't do it with a simple text layer over
binary stream I/O. Despite the popularity of Unix, everything is *not* a
byte stream! In some ways, the binary stream model is inferior; Unix
programmers who work with block-oriented devices have known this for a
long time.

In short, the "text over binary" abstraction is not always appropriate,
and your zeal for that model can't change that fact.
Bradd W. Szonye