I agree that ASCII files are the most general interchange mechanisms and
so deserve special support. But it needs to be the absolutely simplest
and minimal form possible. Perhaps something that will Open, Close,
(re)Position and read and write strings and numbers. Anything fancier
defeats the goal of simplicity. Perhaps the same interface that In and
Out provide, but directed to ASCII files, will be sufficient. When it
is demanded that format control, extreme efficiency, blocking, etc be
added, it should be refused. If you need such fine grained control then
you are no longer a "civil engineer" but a computer programmer. There
are plenty of mechanisms within Oberon/F to allow you to control the
layout precisely. It is time to learn Oberon in more detail. This is
not so different of Mathematics. To properly apply
mathematics in anything but a rote way, one has to master the relevant
concepts and develop sufficient skills. To complain that mathematics is
"too difficult" for the physicist will get you nowhere. It is the same
with using a programming language as a tool.
Design of an standard I/O library is tantamount to language extension.
To do it properly requires great skill, experience and good taste. To
come up with a "standard" library is simple. To come up with one that
is still perceived as a gem after a decade of use is very, very
difficult. I have not bheen impressed with any of the standard C
libraries. Perhaps the concept of a framework is more relevant these
days than I/O libraries, with the exception of "math" libraries.
Venkat Rao
GTE Telephone Operations