Re: [boost] Re: compile time member detection

From: Gabriel Dos Reis (gdr_at_[hidden])
Date: 2003-11-23 00:43:49

Mat Marcus <mat-boost_at_[hidden]> writes:

[...]

| errors. The compiler vendors indicated that this was tantamount to
| providing a will_this_compile metafunction and is too
| difficult.

More or less.  From what I understood, that kind of speculative
instantiation will require more work because existing compilers have
not antitipated that possibility.  But, it is not unimplementable as
it has been said in the first morning.  It would just need work. 

Anyway, Bjarne and I are tryng to collect all feedbacks we got and
to write up further elaborations.  Many feedbacks are actual ideas we
considered, discussed, but left out in the final papers we submitted --
we wanted to provide and a "light" seed :-)

| Dietmar proposed a workaround requiring manual declarations
| of conformance which IMO bears strong resemblance to Haskell's type
| class mechanism.

Yes.  The suggestion Dietmar In Kona made was actually part of a
draft we had, but it is missing from the final papers. 
It seems like we should have been less cautious :-)

We do welcome feedbacks, especially those that don't tell us to do 
something because that is what language X does ;-p

-- Gaby