Re- X3J16

[uunet!netcom.com!segfault!rfg]

Bill Gibbons writes:

>  I find it rather unfortuate that the NCEG work was done far too late for
>  X3J16 to take into consideration.

I find that assertion rather dubious, but we'll pass that for now.

>  There are many parts of the NCEG work which, while useful in a language
>  designed *only* for numeric work, are cumbersome for a general-purpose
>  language.  It was entirely appropriate for NCEG to propose every feature
>  which would be useful.  It would also be appropriate for X3J11 to reject
>  or scale back many of the proposals.  That would not be a criticism of
>  NCEG, but rather a recognition that you can't add every feature which
>  might be useful to some group.

OK.  So let's say for the moment that X3J16 should ignore all of the outright
"extensions" proposed by NCEG.  I still believe that this leaves a useful
(and non-trivial) subset of the NCEG work (i.e. the "clarifications") as
worthy candidates for codification into the forthcoming C++ standard.

What am I talking about?  Well, just to give a couple of examples, NCEG
has clarified that certain kinds of optimizations are not permissible.
Two examples I can think of right off the top of my head are (a) that
parentheses must be obeyed as written, and (b) that (x-x) doesn't
always (and necessarily) equal zero.

>  But add all the NCEG proposals to C++ at this time?  Impossible.

I'm not sure whether to laugh or to cry.

It may indeed be too late NOW for X3J16 to give any sort of serious
consideration to ANY of the work of the NCEG, but it isn't as though
either X3J16 or the NCEG sprung into being (fully formed, as from the
brow of Zeus) just yesterday.

In his role as chair of the X3J16 "extensions" working group, I mentioned
quite clearly to Stroustrup well over a year ago that the work of the
NCEG should be considered by X3J16.  At that time, he said to me words
to the effect of "Don't worry.  I won't forget about the work of the NCEG.
They won't let me."

So what happened?


-- Ronald F. Guilmette, Sunnyvale, California -----------------------------
------ domain address: rfganetcom.com -------------------------------------
------ uucp address: ...!uunet!netcom.com!rfg -----------------------------