In private conversation, we discussed some objectives for M1 a while
ago, and I sent my "to do" list. Jon just asked me for an update.
Here it is:
>> - M1: summary of NOR corruption test results so far
Done.
>> - M1: improve NOR corruption test (seems that I need to keep the
>> environmental temperature low ... either by cooling down a room or
>> by sending the M1 into the fridge)
Did that too, though the mystery variable that gave me such a
headache wasn't temperature but the lock bit iself getting
corrupted.
>> - M1: implement universal IR receiver and decoder (Verilog and C)
Pending but not forgotten.
>> - M1: implement the improved reset circuit and test if it prevents the
>> NOR corruption
I did that, and yesterday I got the M1rc4 board with what will
hopefully be the "official" implementation. Now it needs to get
torture-tested.
>> - M1: make the USB stack use all interfaces a HID device advertizes,
>> to support keyboard and mouse combi devices
Done. And the USB work got a little larger than anticipated,
with full-speed and various protocol fixes added to the fray :-)
I'm still not entirely done with it. Pending items:
- find out why there are lots of timeouts and see if this can be
fixed (may be the same as the next item)
- low-speed regression: some people's low-speed mice worked with the
orginal system but not after my changes,
- add CRC checking,
- parse HID report descriptor to enable stack to understand more
devices (e.g., Xiangfu has one that currently only works with a
firmware hack, and one of mine has a hard-coded exception to
the same effect)
>> - M1: see if we can have effects that react to video in changes in
>> specific areas
I bring that feature up from time to time. Don't know who will
attack it.
My M1 priorities are at the moment, top to bottom:
- set up NOR test with M1rc4 and start hammering.
- solve patch compiler regression: when Sebastien builds FN, he gets
a severely dysfunctional compiler. When I build it, it works great.
Not sure yet what it is.
- proceed with compiler and language improvements. What I've done so
far is just the basis for more interesting things. These include:
- making the language a little friendlier,
- add image list support (so that one patch can access more than
two images, although not at the same time),
- add input event directives that will allow a more flexible
mapping of MIDI controls and other things. The two main
objectives there are: 1) to eliminate the need for midiN, idmxN,
etc., variables, but have user-assigned names. 2) allow for more
than just eight controls of each kind.
I also have a bit more cleanup pending and 'll want the unified
compiler to be at least as fast as the old solution was. In the
future, faster.
- clean up the "Tornado Rain Dance" patch and make a tutorial video
showing how to use it. My goal is to set a baseline for demoing
M1, so that the audience of a demo goes home with the message that
this is a cool instrument, and not with the impression of having
just seen the queerest screen saver ever :)
- address the USB issues, above.
- generalize IR protocol support.
There's also the idea of adding a GUI for assinging controls, and
also to add MIDI device auto-identification, along with profiles.
This needs a bit more thought.
- Werner
_______________________________________________
http://lists.milkymist.org/listinfo.cgi/devel-milkymist.org
IRC: #milkymist@Freenode
