End Formal Review: Boost Cast Library

From: Thomas Matelich (sosedada_at_[hidden])
Date: 2000-12-11 12:12:07

Interesting discussion on this one.  A lot of discussion and very little
voting.  By my count there were two votes for acceptance although many
more people chimed in on discussion.

There was a detour into the discussion of the creation of the Boost
Conversion Library which would contain the Boost Cast Library along with
other functions aimed at converting or providing information about
conversions.

polymorphic_*cast:  Duplication of functionality with dynamic_cast has
concerns.
numeric_cast:  Viewed as very useful, though some don't like exceptions.

lexical_cast:  Specialization is a big issue.

One concern I have is that while popular among the real world, the only
place I know of in the standard where there is a debug version of
something is in assert.  Is this a path that we should be going down?

My official recommendation for Boost Cast Library is that it be
accepted.  Here is my list of changes to make:

Version 1:

1) Rename to Boost Conversion Library for compatibility with future
functionality that sounds quite interesting (unless alot of people
really hate this idea).

2) Implement BOOST_NO_STRINGSTREAM for lexical_cast.

3) Add documentation as to reasons why lexical_cast uses copy rather
than const reference.

For version 2, I'd like to see work on conversion functions to fill out
the functionality of this library.  I have a dream of throwing and
non-throwing versions playing together in the streets.  Conversion
querys spreading knowledge about consequences everywhere.

End Review.

wishing-I-got-one-of-them-unanimous-libraries-ly yours,
tom

P.S.  For decisions like names, etc. should we start making use of the
Poll feature of eGroups?

--
Thomas O Matelich
Senior Software Designer
Zetec, Inc.
sosedada_at_[hidden]
tmatelich_at_[hidden]