Re: [boost] Re: In the Spirit of Challenge

From: Paul Mensonides (pmenso57_at_[hidden])
Date: 2002-11-08 16:33:48

----- Original Message -----
From: "Andrei Alexandrescu" <andrewalex_at_[hidden]>

> [about a C++ preprocessor]
> > This one I'd like to see, but it might be too easy.  There isn't much to
> > Cpp.
>
> Explain this to the MSVC and MWCW folks. They both got it wrong. That's
the
> main reason for which I'd like an open-source, good quality C++
> preprocessor.

I asked, but got no response.  The main reason that they haven't paid much
attention to it (I'd guess) is that the PP lib's use of the preprocessor is
pathological.  It is anything but typical.  The biggest reason is that they
use shortcut methods based on assumptions (which in many cases aren't
correct).  These "shortcuts" break encapsulation and cause all kinds of
problems.  I have yet to see a preprocessor that is completely correct in
every way--even EDG has its bugs (as does Borland's, IBM's, Sun's, GCC's,
etc., etc.).  Luckily, many of these bugs are benign--except on VC and MW,
where they are a *major* nuisance.

> And by the way, I understand some preprocessors take a pathologically long
> time with the Preprocessor Library (something I'm too scared to look at)
so
> if Boost had a preprocessor offering, that would come in handily. Then,
> users could combine the Boost preprocessor with their compiler.

Yeah, that would be nice.  By the way, the current library is implemented
for EDG in an EDG-friendly way, which can offer significant speed increases
(actually bringing the library into usability on EDG).

Paul Mensonides