Hi Alexei, Alexei Sheplyakov schrieb:
It would be nice to avoid merge commits in the official repository.
ups. Next time ...
One of the reasons was exactly to get the code in inifcns_nstdsums.cpp into a clean, oo, and of course reentrant form.
I don't quite understand how converting G{2,3}, Li and friends into full-blown GiNaC classes can help. Could you elaborate, please?
You removed global variables and made them local or passed them as function parameters. But if every function were a class, these variables could just as well become class members. It would be thread-safe, too (and no need for additional function parameters).
AFAIR the new RTTI is a prerequisite for converting functions into classes. Is this correct?
If by "new RTTI" you mean the current mechanism using the static void pointer, then yes. Otherwise, in the previous implementation, people would have to choose unique t_info numbers for their own function definitions (not pratical ...). Regards, Jens