From: David Abrahams (dave_at_[hidden])
Date: 2004-06-10 14:43:46
"Vesa Karvonen" <vesa_karvonen_at_[hidden]> writes:
> David B. Held:
>>It's hard enough getting one's head around the PP lib without having to
>>be an expert in it already.
> [...]
>>It's a shame, because there's a lot of code that would benefit from the
>>PP lib that almost certainly isn't because of the difficult learning
>>curve.
>
> Learning Boost.Preprocessor is like learning a rather verbose and
> intricate programming language. It is not really surprising that it
> can be difficult to learn, although the documentation, at least the
> parts that I wrote, could certainly be improved.
When did he write that? Not in this thread.
I'm writing an appendix on the PP lib for the MPL book today. I think
that'll demonstrate that it's possible to provide a reasonable gentle
and high-level introduction without exposing too many ugly details.
The basic outline I'm following is below. If anyone would like to
suggest changes, I'd be grateful.
- Motivation
- Example
- Alternatives
- Handcoded: error-prone
- External Program Generated: not configurable
- Fundamental PP Abstractions
- Macros
- Non function
- Function macros
- Tokens
- Sequences of tokens not including un-parenthesized commas
Note issues with commas and templates
- How to read PP docs
- BOOST_PP_ prefix
- Header file convention
- Z, R, D suffixes
[We're not going to cover anything but automatic recursion in
detail, but we need to explain these things so that people know
how to ignore them in the PP docs ;->, and because their presence
leaks into the macros they write for passing to higher-order PP
lib macros]
- PP library abstractions
- Iteration
- Horizontal:
- Examples:
BOOST_PP_ENUM_PARAMS...
- Problems: Debuggability, Readability
- Vertical
- Local Iteration
- File Iteration
- Control structures
- BOOST_PP_IF / BOOST_PP_IDENTITY / BOOST_PP_EMPTY
- BOOST_PP_EXPR_IF
- BOOST_PP_COMMA_IF
- Data structures
- sequences
- tuples
- arrays
- lists
> It is possible to
> implement a language using the C99 (or even the less capable C++97)
> preprocessor that would be much less verbose and intricate and would
> be significantly easier to use (see
> http://groups.yahoo.com/group/boost/files/Order/ for details), but
> it would come at some cost in efficiency (on some compilers the cost
> would probably be prohibitive and on some compilers the cost is
> insignificant).
Well, if you really have a conforming C99 preprocessor, you could use
Paul M's Chaos, which I think is both easier to use and *way* more
efficient than the current Boost PP.
-- Dave Abrahams Boost Consulting http://www.boost-consulting.com