Re: [boost] Technical session at the October C++ meeting?

From: John Max Skaller (skaller_at_[hidden])
Date: 2001-08-15 17:42:00

David Abrahams wrote:
> 
> I agree that type_traits is extremely useful and should be considered for
> standardization.
> 
> That said, it is only used as an /implementation detail/ of other boost
> libraries. There is no reason a vendor couldn't implement any of the boost
> library interfaces without using boost::type_traits (except that it would be
> a lot of work). From a standardization POV, I don't see how library
> implementation dependencies are relevant to which libraries are considered
> first.

        I do, and I think you nailed it: implementors will want to
see HOW to implement something, not just what the interface looks 
like. After all, some people will want to build a test example 
of the library to try it out, and they may want to fit it into
their own library architectures.

        Now, I HATE traits libraries, (template equivalent
of global data .. ) but if the type traits library is actually
used by other boost libraries, then that's a very strong argument
that its a useful component worth standardising.

        Furthermore, its worth proposing becase there
might be a better way to provide the functionality with
a core language modification. So I'd have to recommend
putting it forward.

        Also, while I think of it, I suggest that the proposals
go for 'tentative acceptance' by the committee. This is a committment
to examination and refinement, while not necessarily accepting the
proposal as it stands: its a warning to clients NOT to depend on
the interface as it stands, and allows the committee to fiddle
without accepting 'backwards compatibility' as any kind of
argument against doing so.

-- 
John (Max) Skaller, mailto:skaller_at_[hidden] 
10/1 Toxteth Rd Glebe NSW 2037 Australia voice: 61-2-9660-0850
New generation programming language Felix  http://felix.sourceforge.net
Literate Programming tool Interscript     
http://Interscript.sourceforge.net