I finished more work on portmidi: 1) Integrated patches for Debian as requested by Paul Brossier. I didn't change everything: in particular some Makefiles are either not there or shouldn't be there because they are generated by cmake. I think the changes will eliminate the compiler warnings, but let me know.
2) Some fixes to the Cmake files on the Mac side broke some things on the Linux side. Cmake should work on Linux now. 3) This includes changes so "make install" works on both Linux and Mac (thanks, Albert for pointing out the problem and fixes. Install also installs portmidi_s and pmjni libraries as well as portmidi.h and porttime.h) 4) I should point out that the README_MAC.txt file Albert asked about was/is actually up-to-date. I added more comments to explain that I use pm_mac/Makefile.osx to invoke cmake in two different directories depending on the configuration type: Release or Debug. Otherwise, you have to do this manually or else cmake will just mix or overwrite files in the same directory. I did not do the same for Linux, so on Linux you're on your own if you want both Release and Debug versions. 5) Since I read some people were having problems with PyPortMidi, I included John Harrison's PyPortMidi into pm_python and also ported the API to Python 3000. There's an incompatibility in that channels are numbered 0-15 as in PortMidi. I'd like to change PyPortMidi for Python 2.6 to work the same way, but I'm afraid that might break programs and annoy users (and John :-). I see the argument for the 1-16 numbering, and that's what I did in the CMU Midi Toolkit. But after using it for quite awhile, I decided MIDI's 1-16 numbering should be used for human I/O only. The code should be 0-15. See the text in pm_python for installation instructions for PyPortMidi. -Roger _______________________________________________ media_api mailing list media_api@create.ucsb.edu http://lists.create.ucsb.edu/mailman/listinfo/media_api
