This is the mail archive of the gcc@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]

Re: The future C++ template model in gcc


[Gabriel Dos Reis]
> Florian Schintke <schintke@zib.de> writes:
> 
> | But what are the real reasons to generate all instantiations?
> 
> A template is instantiated for given template-arguments.  The set of
> template-arguments is unbounded.  So what do you mean by "to generate all
> instantiations"?
OK, with all instantiations I meant all necessary instantiations.
Why is it not possible to generate for a template implementation
one very generic code, that works for all instantiations.
Like a function that works for all parameter values, a
template implementation that works for all instantiations.

> | So my question is:
> | Why isn't it possible to generate a generic 'template meta
> | instantiation' that works for all possible instantiations?
> 
> How does that mean?
My idea is to not instantiate the templates,
but to provide a generic template implementation that works 
for all template arguments. 

Florian Schintke
-- 
Florian Schintke <schintke@zib.de>, http://www.zib.de/schintke/


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]