each of the enum values is a bit-mask that can be combined into a single
integer that represents a collection of settings. these three options were
active in the bit-mask that was returned in adam's example:
> kOtherPlugSet (0x4000)
> kIncomingDirection(0x800)
> kConnectionMade(0x01)
here's a little more python code to show what's going on:
we'll continue from adam's example, where the integer contained in messageID is
18433.
get the string representation of the hexidecimal bitmask
>>> hex(18433)
'0x4801'
python, in it's all-knowingness, provides an easy shorthand for working with
hexidecimal values. enter in the hex representation, and it gives you an int
>>> 0x4801
18433
use the "bitwise or" to combine values:
>>> 0x4000 | 0x0800 | 0x0001
18433
so, it follows that:
>>> (0x4000 | 0x0800 | 0x0001) == 0x4801
True
to determine if a given option is in play, use "bitwise and". in our case,
we'll test if kIncomingDirection is enabled:
>>> bool(0x4801 & 0x800)
True
now test if kConnectionMade is set:
>>> bool(0x4801 & 0x01)
True
finally, test for a bitmask that has not been set (kAttributeAdded = 0x40):
>>> bool(0x4801 & 0x40)
False
bitmasks are often used in C/C++. more "descriptive" data types are usually
favored in python, but since the maya python API is derived from C++, you'll
have to leg up on the concepts. i'm no expert myself (i've basically just told
you all i know about bitmasks) so if i'm missing something, hopefully someone
else will jump in with a more thorough explanation.
-chad
On Jan 19, 2010, at 3:01 PM, Adam Mechtley wrote:
> If you look at the C++ devkit examples (or for that matter the explanation up
> in the Public Types section for MNodeMessage) you can see the problem.
> Namely, the enum values correspond to hexadecimal numbers. If you run hex()
> on the number, you will see what's going on:
>
> hex(18433) is 0x4801, which means:
> kOtherPlugSet (0x4000)
> kIncomingDirection(0x800)
> kConnectionMade(0x01)
> I'm guessing the reason for this paradigm is so you are not sending a
> separate message for each of these events—otherwise you would have two or
> three messages sent for each attribute. As you can see in the C++ examples
> (e.g. nodeMessageCmd.cpp), a simple if-statement does not suffice—you must
> instead use the bitwise operator:
>
> if ( msg & MNodeMessage::kConnectionMade ) {
> cout << "Connection made ";
> } else if ( msg & MNodeMessage::kConnectionBroken ) {
> cout << "Connection broken ";
> } else {
> return;
> }
>
> Thus, the same goes for Python:
>
> if messageId & OM.MNodeMessage.kConnectionMade:
> # Do cool stuff
>
> If a particular message code is not included, the if-statement will return 0
> (false). If it is included, it will return decimal representation of the hex
> number for that enum, and thus pass the test.
>
> On Tue, Jan 19, 2010 at 2:59 PM, Judah Baron <[email protected]> wrote:
> We're using some of the MMessage classes to register callbacks. For
> MNodeMessage callbacks, the first argument passed in by Maya is the
> messageID. If we check to see if the messageID ==
> OpenMaya.MNodeMessage.kConnectionMade it's always false because the value of
> OpenMaya.MNodeMessage.kConnectionMade is 1. The actual messageID coming in to
> the callback is 18433. The other MMessage related IDs are also not what I
> expected.
>
> Prior to 2008 I didn't have this problem. But prior to 2008 there were other
> issues. As I recall, the MMessage family of classes were not fully
> implemented back then. They seem to be fully implemented now, in that they
> can actually be used, but associating the IDs is a little strange, especially
> when working with multiple versions of Maya.
>
>
>
> On Tue, Jan 19, 2010 at 10:42 AM, Adam Mechtley <[email protected]>
> wrote:
> Could you elaborate a little or provide a specific example? In general, I
> have not had problems using enumerated types—certainly not where the values
> are incorrect. An enum in the C++ class simply corresponds to an integer
> value as far as Python is concerned, so corresponding zero-based integers can
> be used interchangeably where an enum is required. There are, as always, some
> exceptions, including a couple of bugs I am aware of:
> MTransformationMatrix.getRotation Error (when using a rotation order enum)
> api enums passed as reference
> Python cannot accept pointer to integer as reference to enum type (need to
> use short instead)
>
> On Tue, Jan 19, 2010 at 12:21 PM, Judah Baron <[email protected]> wrote:
> Has anyone experienced problems using Maya's enum values?
>
> We have had to use integer values directly because the enum values we get
> through the class is not correct. It appears to me that the enums coming
> through the python class are all zero based and 'localized' to the assocaited
> class, whereas maya's enums are generally a lot larger and seem to be coming
> from a larger 'global' list of unique values.
>
> thanks,
> -Judah
>
>
>
> --
> http://groups.google.com/group/python_inside_maya
>
>
> --
> http://groups.google.com/group/python_inside_maya
>
>
> --
> http://groups.google.com/group/python_inside_maya
>
> --
> http://groups.google.com/group/python_inside_maya
--
http://groups.google.com/group/python_inside_maya
