[Bf-python] Const dict arguments
Willian Padovani Germano
wgermano at ig.com.br
Wed Jul 20 06:30:32 CEST 2005
Ken Hughes wrote:
> Ok, so should the tp_getset exactly duplicate x.getStuff() methods
> which do not use const dicts, or should they only use const dicts? In
> other words, we're implementing new code but should that wrap the
> older code which is using strings?
Hi, since we're planning the missed cleaning / standardization of the
API, we can't avoid breaking things here and there. Thankfully:
1) these changes will be easy to fix in scripts and we can document them
well to help users;
2) we can take the chance to also start collecting all working scripts
(with proper licences, of course) in a single repository, making sure
these work with the changed API and are cross-platform, well documented
and tested;
3) the bulk of changes can be added "at once", meaning that script
writers shouldn't have to update scripts more than once because of this
overhaul, if we plan and implement it well.
So if we all agree that what we are discussing is an overhaul of the
API, I suggest that compatibility, though always very important, should
not be the main focus *this time*. What do you all think? Script
authors are also welcome to express their opinions, of course.
BTW, strings vs. const dicts is not the only issue you will find.
Angles, lists vs. multiple args, etc. should also come up. As I wrote
in my previous email, after you work on getsetters and we invite users
to express their opinions, we'll be in a good position to decide about
these matters. It's probably better to keep the current code there for
now and use ifdef's to separate old get/setattr functions from your
additions, btw. Trivial to remove once all matters have been decided.
--
Willian
More information about the Bf-python
mailing list