[boost] Re: result of compose discussion

From: Reid Sweatman (reids_at_[hidden])
Date: 1999-06-21 14:14:09

> -----Original Message-----
> From: Greg Colvin [mailto:greg_at_[hidden]]
> Sent: Monday, June 21, 1999 11:30 AM
> To: boost_at_[hidden]
> Subject: [boost] Re: result of compose discussion

> > but I think that leading underscores are discouraged by the 
> C++ standard.
> 
> Not just discouraged, but reserved for the implementor and 
> forbidden to the
> user.

Unless the spec has changed since I last looked, the symbols reserved to
implementation use were those that contained double underscores and those
that began with an underscore followed by a capital letter.

> Perhaps, but "ary" is a proper English suffix, from the 
> Latin, "un", "bin", and
> "tern" are proper English prefixes, also from the Latin, and 
> unary, binary, and
> ternary are all in my dictionary.  So why kludge up ugly 
> names when English has
> the names we need?  That leaves the 0ary case, which could be 
> nullary or nilary. 
> Neither appears in my 1985 unabridged dictionary, but nullary 
> is in fact already 
> used in the funtional programming community, and likely will 
> make it into more
> recent dictionaries.  So we have perfectly reasonable English 
> words for the
> concepts we are implementing, and I don't see a good reason 
> not to use them.

Well, for the simple reason I stated in my original post:  operator notation
has no grammatically consistent form for the null-domain case, since while
C++ functions can take zero arguments (and hence, may not be functions or
mappings in the mathematical sense), operators can't.  Why stick with
operator notation when it doesn't fit the problem domain?

------------------------------------------------------------------------

eGroups.com home: http://www.egroups.com/group/boost
http://www.egroups.com - Simplifying group communications