[boost] Re: Formal Review: Template Metaprogramming Library (MPL)

From: David B. Held (dheld_at_[hidden])
Date: 2002-07-25 13:46:59

"Paul A. Bristow" <boost_at_[hidden]> wrote in message
news:AHEJIHEOOOBMJPAGPLIPKEFKDCAA.boost_at_hetp.u-net.com...
> [...]
> Can you please explain how the word 'fold' applies.  I am still
> mystified by this!

It's a term from pure functional languages.  I'm not really a functional
programmer, so I can only guess that the name implies that it "folds"
a sequence into a single return value (the fact that it operates forwards
and backwards can be thought of as "folding" the sequence back on
itself?).

> [...]
> Is shortness really that much more important than clarity?  I think
> that MPL has a major problem with users minds boggling, and _c
> is an unnecessary extra thing you need to know that isn't obvious.

I agree that MPL is not for the faint of heart (I don't even have an
application for much of it).  But it is a very verbose library to begin
with, since it cannot take advantage of things like the intrinsic C++
operators (+, -, =, etc.).  Very often, names are extended by adding
::type and stuff to them.  Taking that into account, making long names
for every element would make the code difficult to follow, IMO.  I'd
just as soon see the shortest names possible.

> [...]
> > "implementation". Remember Sutter's "Pimpl"? :)
>
> It vaguely rings bells, but it doesn't gives many clues -
> [...]

"Pimpl" == "Pointer to Implementation".  Throughout Boost, "impl"
is commonly used to refer to an implementation class.

Dave