LWG Feedback: Rational number library

From: Beman Dawes (bdawes_at_[hidden])
Date: 2001-11-14 09:48:14

Various Boost libraries were proposed to the C++ standards committee's 
Library Working Group (LWG) at the Redmond meeting for inclusion in the 
upcoming Library Technical Report (TR). What follows is my recollection of 
the reaction by the LWG to the Rational Number proposal.

The only real technical issue I can remember being mentioned was a question 
as to why the design had chosen eager rather than lazy evaluation.  None of 
the Boost people present could remember.  Assuming rational numbers are 
decided to be "in scope" for the standard (see below), the proposal will 
need to add more rationale for design decisions.

Most of the discussion was on the question of whether or not rational 
numbers are within the scope of the TR and next standard.  A number of the 
LWG members seemed to be in the position of Bill Plauger, who said "I'm on 
the fence on this one."

The problem has nothing to do with the Boost rational number proposal 
itself, but rather that while the committee has some specific library needs 
they would like to see met, and some disqualifiers to weed out unsuitable 
libraries, the primary general rule for determining if a proposal is "in 
scope" seems to be to ask "How useful is this?" (and possibly "how hard is 
it for a user to do it themselves?").  "How useful" is a pretty subjective 
question.

Those present could identify several real-world uses rational numbers, but 
not an overwhelming number.  It is possible that by polling users enough 
could could be found to push rational over the top.  But that just 
sidesteps the bigger problem: the committee does not have an overall vision 
for the scope of the library.

[I'll post some of the above on the LWG reflector to see if the committee 
can firm up their notion of scope.]

--Beman