Re: [boost] Re: boost::bind and the preprocessor

From: Peter Dimov (pdimov_at_[hidden])
Date: 2002-03-08 07:50:29

From: "vesa_karvonen" <vesa_karvonen_at_[hidden]>
> --- In boost_at_y..., "Peter Dimov" <pdimov_at_m...> wrote:
> > From: "vesa_karvonen" <vesa_karvonen_at_h...>
> > > Aside from EDG and other troublesome preprocessors what do you
> > > suspect would make the result hard to maintain?
> >
> > Bind is already hard to maintain. The reasons are complex but the
> > simplest way to put it is that maintaining Bind requires too much
> > knowledge in several areas.
> >
> > Using the preprocessor library would increase the required
> > knowledge.
>
> Fair enough. Programming is understanding.
>
> On the other hand, using the preprocessor might also help by reducing
> the amount of repetion and thus making it easier to understand the
> differences between various components. I've often noticed that being
> able to express something in shorter form has not only made the code
> easier to maintain, but has also spawned completely new ideas about
> the code.
>
> Looking at the Boost.Bind code, I see that it is extremely repetitive.

You are right, of course. The code is repetitive and our friends at the CWG
might want to do something about it after 5 years or so, when they have
nothing better to do. :-)

I haven't found the 'perfect' solution to this problem yet. The preprocessor
library has the advantage that the preprocessor is always present; but it
adds a significant amount of obfuscation.