On Mon, 20 Aug 2001, Richard B. Kreckel wrote:
On Mon, 20 Aug 2001, Pearu Peterson wrote:
Meanwhile I implemented these hooks, and it works perfectly. Please find the patch attached to this message.
Thanks a lot for your patch, I am currently investigating it. This won't be finished today, hopefully tomorrow. Just one question: You know that this issue has been discussed already with no final consensus? The last rant is archived at <http://www.ginac.de/lists/ginac-devel/msg00259.html>. (I don't see clear yet. This email is just to make sure we don't miss any opportunity.)
Okay, I've looked at Peru's patch now. It is indeed quite trivial. Let me restate it: where we have foofunction_eval(const ex &x, const ex &y) with a fixed number of arguments he proposes to add an alternative where there is foofunction_eval(exvector) directly, i.e. not to do the translating effort from seq[0] to x and seq[1] to y. His patch simply adds additional support for this into function.pl along the old named arguments. Interstingly, this seems to solve a bunch of problems of scripting-interface maintainers. Alex (and Christian): May I ask you for a braindump of yours as to why we haven't done this all the time? Is it just a matter of being able to access `x' and `y' inside the implementations of foofunction_eval() or were there any other less trivial considerations? (This way all the function.{h,cpp} being generated from function.pl would be unnecessary. Not that I want to throw it out, though...) Hasta luego -richy. -- Richard Kreckel <Richard.Kreckel@Uni-Mainz.DE> <http://wwwthep.physik.uni-mainz.de/~kreckel/>