Topic for #milkymist is now Radical Tech Coalition :: Milkymist One, Migen, Milkymist SoC & Flickernoise :: Logs: http://en.qi-hardware.com/mmlogs
elldekaa joined #milkymist
cjdavis joined #milkymist
wolfspraul joined #milkymist
aw_ joined #milkymist
aw joined #milkymist
xiangfu joined #milkymist
kristianpaul joined #milkymist
kristianpaul joined #milkymist
cjdavis left #milkymist
cjdavis joined #milkymist
aw joined #milkymist
aw_ joined #milkymist
<xiangfu> wolfspraul, about those DMX device I bought.
<xiangfu> wolfspraul, for now what we can improve is add more patches that support dmx. for now there is only one.
<xiangfu> next will be support more dmx output variables. now only 8. maybe all the 512 dmx out that will be better.
<xiangfu> dmx input variables. 8 is enough I think. I don't know why we needs more dmx input. since I don't have much experience in DMX.
<xiangfu> more output definitely good for m1
<xiangfu> the next thing is when the dmx cables arrive. I take 1 or 2 videos. and send email to Naihanli.
<xiangfu> the DMX patch is different because it highly depends on what DMX device you connect to m1.
<wolfspraul> ok
<wolfspraul> we should make this work well in the media wall
<wolfspraul> that is, when you think you are ready, prepare a few ideas/demos, and go to jj and demo her, see what she likes or doesn't like
<wolfspraul> then you can optimize right there
<xiangfu> yes. I can create one or two DMX patches for the media wall
<wolfspraul> it needs to work in a robust way too, so you don't always have to travel along with the media wall
<wolfspraul> so it must be possible to wire it all up including the existing controller without disrupting anything
<wolfspraul> now that m1 always boots at least the dmx passthrough must always work, I guess
<wolfspraul> little details that make it work in the actual media wall...
<xiangfu> in fact, by default. m1 block all dmx message. it act like 'take controller form any DMX controller'
<xiangfu> that a little thing I can improve. :)
aw_ joined #milkymist
<wolfspraul> sounds good
<wolfspraul> it should always be fun to hookup m1 to something, so the first impression should not be "ooops, now it broke something"
<wolfspraul> so the default for both dmx and midi, in my opinion, would be to just route from in to out, if that is possible
<wolfspraul> that's just my stupid 'first thought' of course :-)
aw_ joined #milkymist
wolfspraul joined #milkymist
cladamw joined #milkymist
cladamw joined #milkymist
<wpwrak> hmm, rejon doesn't seem to be very curious about his answers ...
<wolfspraul> what answers?
<wpwrak> to his mail
kristianpaul joined #milkymist
kristianpaul joined #milkymist
lekernel_ joined #milkymist
kristianpaul joined #milkymist
rejon joined #milkymist
wolfspraul joined #milkymist
<xiangfu> wpwrak, Hi
<xiangfu> how to add text to video ? what tools you use?
<xiangfu> finished one video about M1 and dmx fixtures. :)
xiangfu joined #milkymist
<wolfspraul> xiangfu: check the dirtpan video here http://en.qi-hardware.com/wiki/Copyleft_Hardware_News_2011-06-01#NanoNote
<wolfspraul> he used melt http://www.mltframework.org/
<xiangfu> I found openshot is easy for me. GUI. :)
cladamw joined #milkymist
<cladamw> xiangfu, about the lastest release and snapshot. No rush now. just let me know after you built them. tks !
<xiangfu> cladamw, oh.
<xiangfu> if no rush. I will finish it tonight. try to make it simple.
<xiangfu> cladamw, for now. it have to reflash snapshot first. then flash the release image. even after those two reflash. we still missing the data partitions.
<xiangfu> I want one command. then done. :D
<cladamw> xiangfu, yes. i found that data partitions missed. so take your time. just let me know once you have that newest one. :-)
<xiangfu> cladamw, maybe I will update --rc3 option. :)
<cladamw> xiangfu, alright. I think that I can modify my own 'reflash' for log records. But please keep using two '00 XX' mac address arguments, so I can still have them to reflash.
<xiangfu> cladamw, ok.
Martoni joined #milkymist
<kristianpaul> x
wolfspraul joined #milkymist
wolfspraul joined #milkymist
azonenberg joined #milkymist
<GitHub92> [flickernoise] xiangfu pushed 1 new commit to master: http://git.io/1Uljsg
<GitHub92> [flickernoise/master] DMX: enable chain mode by default - Xiangfu Liu
<xiangfu> wpwrak, the 2itfs usb code not working with my Rapoo 1800, only keyboard working.
<lekernel_> xiangfu: chain mode by default is when you have a controller _and_ lights, which is more rare imo
<xiangfu> lekernel_, yes. but if there is a controller. that will break user feel m1 break something.
<xiangfu> I also agree with Wolfgang. :)
<xiangfu> lekernel_, how to test bios? flash it to m1?
<xiangfu> the auto-on make debug easy. no needs to press middle button now. :D
<cladamw> ( Web Update ) Will feature of "Web Update" include standby stream in the future ? If yes, it will help production a lot when every m1 has been packed already then I don't need to open upper case to reflash.
<xiangfu> cladamw, if the 'standby' broken. you have to open the case. :) because you can't boot to flickernoise or rescue any more
<cladamw> xiangfu, yes i know. from http://milkymist.org/wiki/index.php?title=Flashing_the_Milkymist_One#Web_update_.28recommended.29 i can see its purpose.
<cladamw> i would like to know if there's good way even if later our m1 rc3 stocks stay too longer time then one day we still need to ship them out. Then suddenly let me do every piece to 'upgrade' or even 'reflash' all images. I still don't like it. even the work "opening it -> reflash it -> closing it -> packing it again" doesn't spend too much time. :-)
<cladamw> ha ~ but seems in the short term it won't be happened. :)
<xiangfu> Milkymist One & DMX fixtures.
<xiangfu> Scopeuk-AFK, ^^
<cladamw> xiangfu, bravo ! the video !
<wpwrak> xiangfu: looks great !
<wpwrak> the M1 could use a bit of illumination, though. it almost disappears in the middle of all the lovely lights. my videos have the same problem
<lekernel_> no, web update doesn't reflash the standby
<xiangfu> wpwrak, (illumination) oh. yes.
<wpwrak> xiangfu: (combo device) maybe it would help if you could do an lsusb -v -d xxxx:xxxx on that device. and maybe dump the report descriptors: http://lists.en.qi-hardware.com/pipermail/discussion/2011-November/009206.html
<xiangfu> wpwrak, http://pastebin.com/AkXnxzU4 I compile the latest bios. flash it by m1nor. when it boot. it give those message. but keyboard works fine.
<xiangfu> wpwrak, ok. wait. I missed this email.
<wpwrak> (messages) hmm, strange. that doesn't look as if anything should work. is the device on port A and there's something else on port B ?
<xiangfu> the device is on port B. nothing on port A.
<xiangfu> now I switch to port A. rebooting.
<wpwrak> lsusb -v looks good
<xiangfu> same error.
<wpwrak> strange. if it says it's disabling the device, nothing should work
<wpwrak> can you still type in the BIOS after is says it's disabling ?
<xiangfu> in the BIOS not working. but works in flickernoise.
<wpwrak> aha ! then your BIOS and RTEMS are mismatched
<wpwrak> and maybe the bitstream too
<xiangfu> reflashing soc
<wpwrak> hmm, seems that your mouse sends an extra byte
<xiangfu> I flashed your images. mouse not working.
<xiangfu> wpwrak, how can you find there are extra byte in usbhid-dump.txt?
<wpwrak> the Usage (AC Pan), section
<wpwrak> ah, and the report ID. 6 bytes.
<wpwrak> maybe yuo can try this, as the beginning of softusb-input/main.c:process_mouse (before the first length check): buf++; len--;
<wpwrak> then ... make -C softusb-input
<wpwrak> make -C software/libhal
<wpwrak> make -C software/bios
<wpwrak> then flash the bios. next, in rtems,
<wpwrak> cd build; cp ../../milkymist/software/libhal/softusb-input.h ../c/src/lib/libbsp/lm32/shared/milkymist_usbinput/softusb-input.h && make -j 5 && make install
<xiangfu> the makefile have commit by lekernel_ :)
<wpwrak> and finally cd flickernoise/src; make bin/flickernoise.fbi && m1nor bin/flickernoise.fbi
<wpwrak> ah yes. maybe it's a little easier now. haven't tested them yet :)
<xiangfu> I tested today. 'make bios' works fine. :D
<wpwrak> but the (messy) rtems and flickernoise build should still be the same
<xiangfu> compiling rtems
<wpwrak> maybe we should move the softusb firmware loading from rtems into flickernoise at some point in time. maybe just have a "loader" function in rtems. that would avoid these strange detours with the copying of headers
<wpwrak> rejon: don't feel like forwarding ? not interested in what us technical folk think about those problems, and potential solutions ? :)
<xiangfu> wpwrak, I dump another mouse. which is working in m1. there is no AC Pan in that mouse.
<xiangfu> what is Usage (AC Pan)?
<xiangfu> where I can find the report ID is 6 bytes. thanks werner. (learning ...)
<lekernel_> wpwrak: what's that rejon mail?
<wpwrak> the report id is just one byte. it's at the beginning of the report. no idea what the AC pan is - first time i see that :)
<wpwrak> lekernel_: he had some usability issues with his slideshow
<wpwrak> not entirely unexpected of course, given that image support isn't really designed for that :)
<wpwrak> xiangfu: this is where the HID protocol is defined: http://www.usb.org/developers/devclass_docs/HID1_11.pdf
<xiangfu> wpwrak, finished compile bios. rtems/flickernoise.
<xiangfu> wpwrak, with (buf++; len--;) keyboard and mouse works fine now.
<wpwrak> victory !! :)
<wpwrak> a clean solution will be a bit harder to make, though. at the moment, we don't even try to parse those report descriptors
<xiangfu> why we do 'buf++; len--; ' ?
<wpwrak> to skip the report ID
<wpwrak> we don't need it because we get all the information we want already from the endpoint number
<xiangfu> wpwrak, thanks.
<xiangfu> reading the PDF.
<xiangfu> wpwrak, my full-speed mouse working now. :)
<wpwrak> oh, also with the report hack ? interesting
<xiangfu> wait. I reflash my m1 back to 'current' release and test again.
DJTachyon joined #milkymist
<xiangfu> wpwrak, yes. ''buf++; len--; '' make my full speed mouse working.
elldekaa joined #milkymist
Thihi joined #milkymist
errordeveloper joined #milkymist
<GitHub59> [migen] sbourdeauducq pushed 2 new commits to master: https://github.com/milkymist/migen/compare/5c7131d...4d1a960
<GitHub59> [migen/master] fhdl: replication support - Sebastien Bourdeauducq
<GitHub59> [migen/master] wishbone: decoder + shared bus interconnect - Sebastien Bourdeauducq
elldekaa joined #milkymist
antgreen joined #milkymist
elldekaa joined #milkymist
sh4rm4 joined #milkymist
elldekaa joined #milkymist
<wpwrak> let's see how many mailboxes this overflows :)
elldekaa joined #milkymist
elldekaa joined #milkymist
<Thihi> Gonna try the review unit for the first time soonish.
<Thihi> When I'm finished wiith other things.
elldekaa joined #milkymist
lekernel joined #milkymist
<kristianpaul> you enjoy this isnt it? :)
<wpwrak> the mail-bombing of lurkers ? ;-)
<kristianpaul> yes
<wpwrak> actually, 4/7 got held up because it was too large for the list's limit. but lekernel already gave it safe passage :)
<wpwrak> expect more of the same, though. now things are pretty much where i want them. so can start with the real changes.
<kristianpaul> nice
<wpwrak> lekernel: ah, and there are a lot more non-ansi prototypes in flickernoise.git. they're at places i didn't touch yet, but i guess it would make sense if i fix them too ?
<kristianpaul> lekernel: can you tell us a bit more about compiler-pfpu interface, how the code gets there, how the data gets in and how the data gets back out
<kristianpaul> (taking wpwrak words :))
<wpwrak> context is that kristianpaul wants to use the pfpu for non-graphical tasks. i haven't looked at how pfpu and the rest of the graphics system communicate. i suppose per-frame equations send most of their output to the cpu and it gets handled there. but then, that's just a guess.
<wpwrak> (and it doesn't matter for me at the moment. my interest is elsewhere :)
<kristianpaul> Thanks for the intro, indeed, for example i wonder (snippet) how i could use the atan pfpu instruction on my aplications
<wpwrak> is there an atan ?
<wpwrak> nyet. no atan. phew. you had me worried for a jiffie :)
<kristianpaul> oops no, moment no argh, was nap fault :)
<kristianpaul> s/atan/sin
<kristianpaul> doing some rtfm seems compilation is required on runtime
DJTachyon joined #milkymist
kristianpaul joined #milkymist
kristianpaul joined #milkymist
kristianpaul joined #milkymist
kristianpaul joined #milkymist
kristianpaul joined #milkymist