Re: Understanding MPL

From: Alisdair Meredith (yg-boost-users_at_[hidden])
Date: 2003-04-03 18:09:06

David Abrahams wrote:

> If you are coming to the ACCU conference in Oxford, I'll be giving a
> talk which describes how the MPL works in detail.

... and as a quick bit of feedback from some-one who was there <g>

This was an excellent (if fast paced) introduction.

Just like when you first learn STL you need to get your head around
several concepts at once before you can make headway, I found there were
similar concept in the MPL that you really need to understand close
together, get the big picture, then the details start to become clear.

In particular, discriminating the interface from the implementation is
important (which is easy when somone is there to explain it to you,
maybe less so when looking at a page of angle-brackets!) and seeing how
a large part of MPL is simply repackaging complex types as something
clearer.  Yes, a lot of 'syntactic sugar' seems to be sprinkled in
there, but that is the difference between being able to read your
metaprograms, and writing something for the sake of doing something cool
but distinctly write-only.

While I do not think I am in much of a better position to explain MPL to
anyone (sorry Edward!) I am quite confident I could go away and write
interesting metaprograms with the aid of the docs now, as I can shoot
for the concepts I need.

Oh, and one massively cool thing about MPL is its ability to make
meta-programs portable.  Most of the hacks you will need are already
implemented for you in the library, so just forget about them and write
the meta-code <g>

-- 
AlisdairM