🚀 go-pugleaf

Chuck Noren asked about the current state of the standardization of class
libraries for C++, and suggested some candidates for inclusion.  I have
been working with the folks on the X3J16 committee who are addressing this
issue.

At the first technical meeting in March, the committee set two initial
concerns as priorities for the library efforts:
 1) compatibility with the ANSI C library, and
 2) stream I/O.

We need to make specific, detailed proposals for anything that is to be
considered for inclusion into the standard.  It helps if the proposals are
based on actual implementations.

The discussion between the March and July meetings did cover some suggested
classes for inclusion in the standard, including several of those Chuck
mentioned in his posting.  The consensus at the July meeting was that
attempting to develop a comprehensive class library (on the scale of NIHCL,
as Chuck suggested) was far too optimistic for the immediate (2 year)
future.

Something else to consider is that standardizing a large library, while
helpful to those interested in porting their applications, creates a lot of
work for the compiler vendors trying to conform to the standard.

The main problem we face is that there is precious little ``existing
practice'' to standardize in the area of libraries for C++.  Paul Chisolm's
comment about the string class is an indication of the problem -- there are
many different opinions of what a ``standard'' class should look like.

Another factor is that the availablity of exceptions and templates will
have a significant impact on the design of C++ class libraries.

Those of you who have built C++ class libraries probably have some
suggestions to make regarding the contents of the library.  This seems to
be as good a forum as any for hashing out the issues.


--
Mike Vilot,  ObjectWare Inc, Nashua NH
mjv@objects.mv.com  (UUCP:  ...!decvax!zinn!objects!mjv)