[GiNaC-devel] function classes

[Jens Vollinga]

Dear Richy,

Richard B. Kreckel schrieb:
>> Function classes look more C++'ish if you like. At the moment it is 
>> impossible to add look-up tables or special methods (like 
>> zeta::use_the_other_eval_mechanism() for example) to functions without 
>> returning to good ol' C programming style.
> 
> Sorry, I don't understand: like zeta::use_the_what_are_you_up_to?

I thought of functions that manipulate the cache/look-up-tables or, by 
means of static member variables, the behavior of certain methods like 
eval, evalf, ...

GiNaC functions are still defined solely through their parameters, i.e. 
a list of ex. Any other data members of a function class should only 
influence the way the value/meaning of a function is determined, but not 
the value/meaning itself. So basically these extra data members are 
restricted to caches and flags.

These caches/flags are the main motivation for me to modify the function 
system. But of course, they could be (and are) implemented with the 
current system with global variables and the like as well ...

Regards,
Jens