Re: Tuple library accepted

From: jaakko.jarvi_at_[hidden]
Date: 2001-06-29 02:50:39

--- In boost_at_y..., John Max Skaller <skaller_at_o...> wrote:
> Jeremy Siek wrote:
> 
> > I am concerned with the compile-time and complexity issues of
bringing in
> > the whole tuple library when all I need is a two-argument bind(). 
> 
> That's a pragmatic concern at the moment, but I'd like to suggest 
> that the tuples library, when Standardised, should strongly
> recommend support for tuples of up to 10,000 elements.
> In other words, there'd be no header file at all,
> the compiler would have to generate instances on demand.
> 
> This gets rid of the compile time overhead completely,
> allows a reasonable limit on the size of tuples,
> allows important optimisations, and may be mandatory
> for generalising certain other library components
> efficiently.
Some more daydreaming:

If tuples would be a language feature, instead of just a library, I'd
really like to see that parameter lists of functions would be tuples.
Much of the code repetition in template libraries originate from the
need to provide different versions of the same code for similar
functions that have different number of parameters (see. binders,
lambda, function, ...)
Would the argument list be a tuple, a single template 
could cover all function types.

Tuples are a proposed addition to Eiffel, and paramter lists
are tuples are in Eiffel.

Jaakko