Re: [boost] Report from Berlin C++ Standards Committee meeting

From: David Abrahams (dave_at_[hidden])
Date: 2006-04-11 00:27:35

"Robert Ramey" <ramey_at_[hidden]> writes:

> On the other hand - I don't see the motivation for including concepts
> in the core language.  Haven't we been able to implement concept
> checking with the current facilities of the library?  

Concepts in the language go way beond checking.  There's a great deal
of mess that we deal with today because we don't have true concept
support: everything from the perils of ADL, to the need for the
"typename" and "template" keywords, to tag dispatching, which is
really just a hack we invented to simulate concept-based overloading,
to many uses of enable_if, will largely disappear from code written
with in-language concept support.

> It seems to me that the main problem these days with the C++
> language these days is that its too hard to write a correct compiler
> for it.  Making the core language fancier will only make this
> problem worse.

Actually, implementing concept support presents no great difficulty,
as Doug has demonstrated by implementing ConceptGCC.

> While, the commitee is at it - how about re-visiting two-phase lookup.
> Apparently there are enough implementers skeptical about it that
> they've declined to implement it.

AFAICT you just mean that Microsoft hasn't implemented it yet.  All
the other major compiler vendors who actually do work on their C++
front ends at a reasonable pace (I'm not counting Borland in that
group) have got two-phase lookup implemented.

> (for good reason in my opinion).  I believe this is related to the
> "export" feature - which has also failed to gain traction.

No, it's not related to export; it was implemented so that outright
errors in templates could be detected immediately where the template
was defined, rather than later when it is instantiated.  Two-phase
lookup makes sense even for non-exported templates.  An expression in
a template that doesn't depend on the template parameters shouldn't
change its meaning arbitrarily based on code that happens to come
*after* the template definition.

If you don't like that idea, you're probably not going to like
in-language concept support, since it is capable of moving *all* the
checking to phase 1.  That's a major benefit.

> So its seems to me that there are a number of issues more
> important/urgent than the ones currently being focused on.

I'm afraid it's the same old song: if you want the committee to focus
on other issues, write a proposal that gets those issues on the table.

-- 
Dave Abrahams
Boost Consulting
www.boost-consulting.com