wrapper functionality by hand.
* Removed the swig files (*.i and *_wrap.c)
* Created initial datablock functionality. Most of it has been copied from
the old implementation.
* Created some general functions in gen_utils.[ch]
* Blender.Get(), Blender.Set() and Blender.Redraw functions should work in a
script now.
* Started implementation on an Event function call (BPY_do_pyscript)
Michel
Blender.
* Moved api2_2x/interface.c to ./BPY_interface.c
This will be the general api layer from which all variants of the Blender
api. Currently only the 2.2x variant is initialised.
* Used swig (www.swig.org) to create Python wrappers for a couple of dummy
functions.
* Started implementation of the Blender and Blender.Object modules.
Michel
If I had to guess REMAKEIPO is kind of miss named.
since what it does it free up memory it almost looks like it should be a part of
REMAKEALLIPO we both felt a little overwhelmed by the code though and opted for
just calling both of them in delete_nlachannel_keys
I also added in allqueue (REDRAWVIEW3D, 0);
since things could probably change around since your recalcing the ipo's...
Kent
The third is for actually loading the plugins in blender.
For some reason its not identifying the plugins correctly,
but it compiles and runs so I'm including it hoping someone else
might see where I've messed things up...
I have a simple example at http://www.cs.umn.edu/~mein/test.tgz that
works, if anyone is interested in playing with it.
Kent
Using this flag the following component is build:
source/blender/python
and the following components are not build:
intern/python
source/blender/bpython
Without specifying the flag, things act as they used to do.
Also created a dummy interface.c file and created the directories
which will contain the replacement code for:
- intern/python
- source/blender/bpython
Michel
leaves one object center pink" bug was actually designed by somebody
to mark the active object (which need not necessarily be selected).
I've added a couple of source comments to explain this so the next
person trying to fix this doesn't bother.
a short, so we cast to an int first, then to a short when a button that
modifies a short value is pressed. (Allieviates the bug where the
Unified Renderer button modifies the values of a bunch of other buttons).
action channel from the cooresponding
action when the bone's IPO is deleted
from the IPO window (the old behavior
was to set the channel's ipo to NULL
without removing it from the action).
This is to alleviate the problem that
there is a lot of action channel code
that tries to access the channel's ipo
(chan->ipo) without testing whether
it is NULL or not (resulting in a
segfault).
This seems to be the best way to
solve this problem ... unless there
is a reason why it might be useful
to have an action channel with no
ipo attached to it?
Thanks to zel365 for uncovering this
problem.
