#milkymist IRC log for Sunday, 2012-02-26

qi-botThe firmware (using branch) build was successful, checkout the VERSIONS for detail, see images here: http://fidelio.qi-hardware.com/~xiangfu/build-milkymist/milkymist-firmware-20120226-0212/02:51
n0carri3revening, all02:54
xiangfun0carri3r, evening02:54
xiangfun0carri3r, btw. I am collecting patches. for next release.02:54
n0carri3rahh! okay, great02:54
n0carri3ri am working on stuff for a gig tomorrow night02:54
n0carri3rafter that, i can send you some new stuff02:55
xiangfun0carri3r, it's better that we upload your patches to the patch club. :)02:55
n0carri3ris MIDIUSB and animated image sequences available on web update now?02:55
xiangfun0carri3r, great.02:55
n0carri3rwhere is the patch club?02:55
xiangfun0carri3r, (MIDIUSB, image ..) that is next release. :)02:55
n0carri3rwhen is that release? :)02:56
Guest25192n0carri3r: good evening02:58
Guest25192(wolfspraul here, my nick is messed up)02:58
n0carri3rahh OK02:58
wolfspraulso, restored02:59
wolfspraulgoing to lunch here :-)02:59
wolfspraulyes sorry, we are extremely bad at delivering things through web update02:59
wolfspraulit must improve on our end...03:00
xiangfun0carri3r, the release is under dailybuild. wait for test and release. it will be soon.03:00
n0carri3ri'm excited :D03:02
n0carri3ri have a gig tomorrow, where i will use the M103:02
n0carri3rand again on march 17th03:03
wolfspraulif you have links to pics or videos, please post03:06
wolfspraulbut more importantly maybe even to get your edited patches back, or feedback on what you liked and what not03:06
n0carri3ri am editing some now03:06
wolfspraulthere has been quite a bit of development, but not in web update yet... hopefully some stuff you will like.03:06
n0carri3ri can share a few03:06
wolfsprauljust post somewhere, we will sort it out03:06
wolfspraulout to lunch now, reading backlog later03:07
n0carri3rok talk soon03:07
n0carri3rxiangfu: i'm preparing some patches now03:12
n0carri3rlet me know who i should email these to03:12
xiangfun0carri3r, xiangfu@sharism.cc :)03:15
n0carri3rok sounds good!03:17
wpwrakbtw, does anyone who knows milkdrop (the original) know what the situation regarding the use of function names for variables is ? e.g., things like  sqrt = sqrt(foo);05:28
wpwrakin FN, we allow this. is it also allowed in milkdrop ?05:28
wpwraklekernel: i have a milkdrop question: do you know if milkdrop allows function names to used as variable names, as we do in FN ? or is this a coincidential "extra feature" only in FN ?10:57
lekernelyes, saw it... and I don't know11:07
lekernelis that important?11:07
lekernelwpwrak: the RTEMS driver doesn't pass the first byte of the MIDI message, right?11:16
lekernel(for USB-MIDI)11:17
lekernelnavre sends 4-byte messages, which are turned into 3-byte messages by the driver11:19
lekernelah, yes I see... cable number + code index number11:20
xiangfuthis build is using latest rtems and wernermisc 4 patches : http://fidelio.qi-hardware.com/~xiangfu/build-milkymist/milkymist-firmware-20120226-0212/11:40
lekerneli'll try to get web updates today or tomorrow11:54
lekerneland all things considered I think it's better to defer FFMPEG for the next update11:55
xiangfulekernel,  you mean next release today or tomorrow?12:01
wpwraklekernel: (usb-midi) yup. sort of a compatibility mode ;-) it may actually make sense to bring some of this information back to allow better identification of device. but there, we also need bus (port) and logical address.13:06
wpwraklekernel: ... so maybe something that better waits for linux, where all such niceties are already implemented :)13:07
wpwraklekernel: (ffmpeg) hmm, okay. oh, i've been wondering: couldn't we also rotate images ? i had a very quick look and didn't see a trivial way to add this. but the TMU should have rotation, no ?13:08
wpwraklekernel: (function names for variables) it's for the  var = expr if cond  syntax. parsing this would be very messy if "if" can be function, variable, AND this kind of modifier (using perl nomenclature)13:10
wpwraklekernel: (if) on the other hand, if "if" wouldn't be allowed as a variable name, the grammar would be straightforward. with "if" in all those roles, i'd have to basically make the parser behave like LALR(2) when hitting an "if", which would lead to very tricky code.13:13
wpwrakan alternative would be to require semicolons at the end of statements. but that would break compatibility. yet another way would be to disallow "if" as variable only in new-style syntax (where we can make up new rules at will), but that's kinda nasty as well.13:14
lekernelwpwrak: then go for the simpler option and do not allow "if" as variable name13:22
lekernelediting a bit a preset or two is a lesser evil than making the compiler messy just for this13:23
lekerneland we'll want them in the new syntax anyway ...13:23
Action: wpwrak grabs the executioner's sword :)13:27
GitHub178[flickernoise] sbourdeauducq pushed 1 new commit to master: http://git.io/USm0ig13:28
GitHub178[flickernoise/master] Load USB firmware through ioctl interface - Sebastien Bourdeauducq13:28
Fallenouwhat cool console with slides/buttons/rotative buttons would you advice to buy to play with the M1 ?13:28
Fallenouto have nice interactive patches you can "mix" with your own hands13:29
lekernelFallenou: faderfox lv3 is very cool, but perhaps a tad expensive13:29
lekernelotherwise, korg nanocontrol?13:29
Fallenouoh yes the one on the M1 website, it looks very cool indeed, but 250 EUR13:29
lekernelwpwrak: :) can you commit any regression-introducing stuff to a new branch? as I said I'd like to cut a release soon13:30
wpwraklekernel: ok, i'll just let it sit in its topic branch for now then13:31
Fallenouoh nanocontrol seems to be like 50 EUR13:32
Fallenouit's midi-usb ?13:33
wpwrakFallenou: a bit tricky. if you want knobs to control things, LV3, nanokontrol, etc., are suitable. LV3 is very flexible, but also other controllers give you many items to play with.13:33
wpwrakyes, it's mini-usb13:33
wpwrakFallenou: if you want nice interactive performance, you want something with a pad. iCon i-creativ and korg kaossilator pro have this (as do some others). both have their issues, though.13:34
wpwrakFallenou: the icon has a lot of nice functions and feels solid. BUT ... the controls are very coarse. e.g., the faders split the 0-127 range into only 8 distinct values. the x/y pad operates as a 16x8 grid, etc.13:36
wpwrakFallenou: these coarseness if very noticeable with some patches, but hardly at all with others.13:36
Fallenouhm hum !13:36
Fallenouwhen you say pad it's like a touchpad ?13:37
Action: Fallenou noob in dj/vj stuff13:37
wpwrakFallenou: the kaossilator pro doesn't have such issues. BUT ... 1) it's expensive if you just want a midi controller, and 2) it has very few controls. just the pad, two pots, a few buttons, and - depending on mode - the program knob13:37
wpwrakyes, like a touchpad13:37
FallenouI didn't know a pad could be cool to use with M113:38
FallenouI knew about faders/buttons/rotative buttons13:38
wpwrakin my "wheel" video, you see kaossilator and LV3 side by side. there, it's because i use the kaossilator to generate sounds. but they would also make a nice pair just for control. kinda like on old jet airplanes:13:39
FallenouI am searching only for control, not for sound generation (midi keyboard etc)13:40
Fallenouonly to control variables in FNP13:40
wpwrakthe kaossilator would be for the pilot. a few big controls, like the steering column. the lv3 would be for the board engineer: a wall full of little knobs and dials.13:40
wpwrakthe ideal solution may be a tablet with some OSC/MIDI app. but we haven't investigated into that direction yet13:41
Fallenouthe korg controller 2 seems nice, with like 30 2*buttons+fader+rotative-button13:41
wpwrakyes, it has many controls. i don't like the very cheap feel of it. but then, it undeniably *is* inexpensive :)13:41
Fallenouyep that's a good point :)13:42
FallenouI read a review, it says fader is not very heavy *but* is precise13:42
wpwrakyou don't get that many controls for that little money anywhere else :)13:42
Fallenouso feeling is not awesome but it does the job13:42
wpwrakit lacks the niceness of a pad, though. think of a DJ scratching with a turntable (= pad) vs. the same DJ clicking on the progress bar in xmms or such (= no pad)13:43
Fallenouoh something I noticed, with a video projector, sometimes when going into performance mode, I don't see the entire "screen"13:43
Fallenouthe right hand part is not displayed13:43
FallenouI have to unplug + plug again the projector and then I can see everything13:43
wolfspraulFallenou: I haven't followed closely for about a week or so - any mmu news?13:43
wolfspraullast I saw some pastebin success story, now party planning :-)13:44
Fallenouwolfspraul: Should I send updates when I make progresses on the ML ? or would it make too much noise ?13:44
wolfspraulseems m1 leads to good things :-)13:44
Fallenouhehe :)13:44
wolfsprauloh definitel not too much noise13:44
wpwrak(projector) hmm. so the screen is x-shifted ? or is the whole screen larger, such that it doesn't fit ? is the aspect ration still 1:1 ?13:44
wolfspraulhow many gpl licensed mmu's exist in the world?13:44
Fallenouwpwrak: maybe x-shifter but not sure :/13:45
wolfspraulany progress update is a piece of pioneering documentation and may help or encourage however else to enter the dark world of IC design...13:45
lekernelwolfspraul: there's GRLIB which certainly works nicely13:45
wolfsprauland already learnt something again! :-)13:45
lekerneland something from Openrisc, but not very good13:46
wolfspraulbut you would probably agree that an occasional status update is definitely not noise, given how rare such efforts still are13:46
Fallenou14:48 < wolfspraul> Fallenou: I haven't followed closely for about a week or so - any mmu news? < well it seems the data part is working, load and stores work in simulation, "data store" has been tested successfully (pastebin) on the FPGA but I didn't wrote a test case for the "data load" part13:46
FallenouI will focuse on this today13:46
lekernelFallenou: you should actually write dozens of test cases, not just one13:46
lekernelwhat happens when the pipeline is stalled etc.13:46
FallenouI wrote a mmu.c in libhal abstraction to clean my test case (basically replace all the inline assembly by a C function call)13:47
wpwrakyeah, a quick note every 1-2 weeks wouldn't be perceived a noise by any but the most easily offended :)13:47
lekernelotherwise we'll bump into these problems when putting the Linux kernel into the equation and introducing a lot of complexity13:47
Fallenouyes I agree (more tests)13:47
Fallenoudifferent "stall" conditions + accesses already cached, not cached yet etc13:48
Fallenoulekernel: I noticed the BIOS is displayed on the VGA, pretty cool, is there a framebuffer for this somewhere ? how is it done?13:49
FallenouI thought bios puts() and printf() would only show up in UART13:49
wolfspraulFallenou: what do you think is the best way to start making use of the mmu? try with the Linux kernel right away? or some intermediate steps?13:49
Fallenouwolfspraul: I don't really know, for the moment I trigger mmu features using test commands in the M1 BIOS13:50
wolfspraulmaybe just more test cases, and then try to use it in Linux?13:50
Fallenouwolfspraul: but afterthat yes I think when everything works properly on a lot of test cases it could be nice to start to implement support for it in Linux, but we are not there yet ^^13:50
wolfspraulexcept for more testing, it's fully working?13:51
wolfspraul(to the degree that you can tell right now)13:51
Fallenouthere are mainly three parts13:51
Fallenoudata part, instruction part, exception handling13:51
lekernelwpwrak: you still support imagefile1/imagefile2 ?13:51
Fallenoufor now the part which seems to be close to be OK is the data part13:51
Fallenouin my tests I only try to access mappings which are "defined", which exists, I didn't try accessing unexisting mappings which would fail, there is no "exception" triggering for now13:52
FallenouI think it will be the next move13:52
Fallenouafter data part OK13:52
lekernelseems so :) ok...13:53
wpwraklekernel: yes. all the old stuff should work without change.13:53
Fallenouyesterday I modified the Lekernel - Sailing patch which makes use of imagile1 and 213:53
wpwraklekernel: not sure if that actually makes much sense, though13:54
FallenouI put a different background (water) and a MiNET logo instead of milkymist :D13:54
FallenouMiNET guys found it cool13:54
Fallenouwolfspraul: there is not much you can do with the MMU for now it's a very big work in progress, I wish I could work more on it, but I have limited time :/13:55
wolfsprauloh sure, understand13:57
lekernelwpwrak: shall I remove midiX ?13:57
wolfspraulbut it's moving, sounds good!13:57
Fallenouwolfspraul: yes it is !14:01
FallenouI try to keep simulation git repository up to date so that anybody can join the mmu project14:02
Fallenouwith as few setup time as possible14:02
Fallenouit still needs ISE installed :/14:02
lekernelFallenou: maybe you can have a look at verilator14:03
lekernelif you feel like trying a new simulator... should be a relatively easy task14:03
Fallenouyes I could give it a try14:03
lekernelI don't know if it can simulate lm3214:03
lekernelbut it's been moving pretty fast14:03
Fallenouok cool :)14:04
Action: Fallenou noticed yesterday there is a MIC on the M114:04
Fallenoubut it's hard to make it hear something14:04
Fallenouxiangfu: you no longer build the FNP handbook in the web updates ?14:06
FallenouI can only see it in the "doc" directory of one of the updates14:06
Fallenouin this one : http://fidelio.qi-hardware.com/~xiangfu/build-milkymist/milkymist-firmware-20110509-0000-Release/14:06
xiangfuFallenou, yes. I think when the 'doc' changed to a repo. I forget add it to the build script file.14:08
xiangfuFallenou, will add that to build script file tomorrow.14:08
Fallenoucool thx :)14:09
wpwraklekernel: (midiX) hmm. better not. else, we should also update the GUI.14:09
wpwrak(if) i'll just make it switch with the syntax. so no regression with respect to MD :)14:10
wpwrakno .. where are our VJs who can tell us whether such things have to be allowed in the first place ? maybe mailing list material ...14:10
lekernelwpwrak: yes, including GUI update ofc14:11
lekernelso I'll do it14:12
Action: Fallenou just tried a wireless mouse with a bluetooth USB dongle : does not work14:13
Fallenoushould I send the usb descriptor on the ML?14:13
wpwrakdoes the dongle identify itself as BT or as HID ?14:13
FallenouI would say HID14:14
Fallenoulet's check14:14
Fallenouthe string name is "Logitech USB receiver"14:15
FallenouI try to attach it to a debian VM to get the lsusb -vvv :)14:15
wolfspraulI lost my only working mouse to mechanical failure14:17
Fallenoudamn, mac os does not want to pass the control through to virtualbox14:17
lekernelwpwrak: hmm... tabs in the patch editor do not look very nice. better indent with spaces...14:19
wpwraklekernel: better support tabs in the patch editor :)14:19
wpwraklekernel: this is an item i have waiting for the next time when xiangfu is looking for something small to fix ;-)14:20
wpwraklekernel: there are more issues in MTK. such as allowing the cursor to leave the visible area (editor, file selection) when using cursor keys to move it14:21
lekernelok, do you mind if we have spaces until this is fixed?14:23
Fallenouhttp://pastebin.com/ANHdPAd0 usb descriptor (under windows *errrrm*)14:25
lekernelby the way, do { and } keys work with all layouts?14:26
wpwrakmmh. i'd just use tabs then and imagine the spaces in my head. better than having to remember all the places where spaces should be converted back to tabs.14:26
Action: Fallenou back with a wired logitech usb mouse14:27
Fallenouwhich works14:27
wpwrakFallenou: hmm, it can be used as HID or BT. so at least in theory, we should be able to handle it. but ... FN may not find the report format agreeable.14:28
wpwrak(braces) no idea :) all i use if US. not even sure if is have spanish with USB. i do have some PS/2 spanish ...14:29
lekernelwpwrak: yes, but I'm talking about potential FN bugs here14:32
lekerneldo those braces work with the US layout when you connect your keyboard to the M1?14:33
wpwrak(just tried it :)14:33
Fallenouright after booting it starts compiling patches to jump into performance mode, if I click on "Close" button the window disapears, but it keeps compiling patches in background task, and then jumps in performance mode anyway14:34
wpwrakFallenou: see, so you can clean up your "desktop" without having to wait for compilation to finish :)14:34
Fallenouthe thing is usually I want to go in the menues, the settings, the editor etc14:35
Fallenouso I go in performance mode anyway and then I go back in desktop mode14:35
wpwrakyou could just change the configuration to "None"14:36
wpwrak(Setting menu)14:36
Fallenouoh !14:36
Fallenouso that's what it does :p14:36
wpwrakit has a second function: it lets you pick some confusing keyboard layout ;-))14:36
Fallenouright it works thanks :)14:36
lekernelwpwrak: cool, thanks!14:41
wpwraknow go all the perl way with ... var = expr unless cond  or keep it at  var = expr if !(cond)  ?14:53
GitHub154[flickernoise] sbourdeauducq pushed 1 new commit to master: http://git.io/sMTx9A14:55
GitHub154[flickernoise/master] patches: cleanup syntax - Sebastien Bourdeauducq14:55
lekernelif is shorter and clearer imo14:55
wpwrak(syntax cleanup) ah, nice :) but .. do you really want to do it this way ?14:57
wpwrake.g., we should also change "if" -> "? :", "above" -> ">", and so on14:58
wpwrakbut then it gets a bit intrusive. my longer-term plan is to make a parallel set and compare the equivalence of the generated code14:58
wpwrak(or at least a well-known difference, in cases where it's not trivially equivalent)14:59
wpwrakhere's a (simple) example of such a rewrite check: https://github.com/milkymist/flickernoise/blob/master/src/compiler/test/rewrite15:01
lekernelwell... sure, but that will be good enough for now15:08
wpwrakwhat i'm concerned about is that we don't have a test for equivalence yet. so if you made any mistakes, it'll be hard to find them15:10
wpwraki.e., one would have to go and fish them out of the revision history15:11
wpwrakif you left a copy somewhere, that would allow easy checking at a later time15:11
lekernelwell, then me or someone else will notice the difference on the screen at some point...15:13
lekernelequivalence checking can be quite difficult, not sure if it's really worth doing it (except from a "playing with formal systems" point of view, ofc)15:13
wpwrakmost of the syntax changes produce exactly the same AST15:15
wpwrakand ptest happens to be able to dump the AST :)15:15
wpwrakalright, the  var = expr if cond  stuff only changes the syntax for new-style code. can we consider that as sufficiently non-regressive ?15:17
GitHub88[flickernoise] wpwrak created if-modifier (+5 new commits): http://git.io/U6R99Q15:22
GitHub88[flickernoise/if-modifier] compiler: give access to patch_compile but without prologue and epilogues - Werner Almesberger15:22
GitHub88[flickernoise/if-modifier] ptest: compile to PFPU without patch framework with -c -c -c - Werner Almesberger15:22
GitHub88[flickernoise/if-modifier] compiler: disallow "if" as a variable name in new syntax - Werner Almesberger15:22
wpwrakthe ptest enhancements weren't actually necessary for this. but they have other uses.15:23
Fallenouwheel video is really awesome :)15:28
lekernelwpwrak: for the "rain dance MIDI RMX", any better idea than mapping the eight LV3 faders to the current midi1..midi8?15:49
wpwrakoh yes !15:51
wpwraklemme find it ...15:51
GitHub68[flickernoise] wpwrak pushed 1 new commit to master: http://git.io/9nzJrQ15:57
GitHub68[flickernoise/master] demo/raindance/: Tornado Raindance, with stronger MIDIfication (WIP) - Werner Almesberger15:57
GitHub175[flickernoise] wpwrak pushed 1 new commit to if-modifier: http://git.io/g0HWiA15:57
GitHub175[flickernoise/if-modifier] wip - Werner Almesberger15:57
wpwrakhere it is. still needs a few smaller improvements15:57
wpwrakah, crap. didn't mean to push the if-modifier work in progress. ah well.15:58
larscquick! rewrite history!15:59
lekernelgreat. do you want it included in the bag of web updates?15:59
wpwraklarsc: that's kinda nasty :)15:59
larscwpwrak: i know15:59
lekernelnote that in theory, simple mode is supposed to skip MIDI patches16:00
lekernelI can include the pacman one as well... the rotating wheel has a bit too many files, better wait for GIF/MPG support16:01
wpwrakmaybe we can have .zip and .tar support ? ;-))16:01
wpwrakpacman is pretty much finished. or will be once if-modifier is merged and i've tested the wip16:02
wpwrakrain dance still needs a bit of love. i need to make the clipping mode optional. (simple) and i also want to use buttons to have alternate settings. need to merge the solution i found to actually produce usable results (not all do, particularly not the "nice" ones :)16:04
wpwrakhmm. if i merge master into if-modifier (to get the usb ioctl) and then the merged if-modifier branch back into master (to get the new if construct), will git know how to sort this out ? maybe better to do this in a separate branch ...16:06
larscshould work i guess16:09
Action: wpwrak is too chicken to try :)16:11
wpwrakpacman works with trailing "if" :)16:12
lekernelwpwrak: cool. if you have no regression wrt the other patches, then just merge everything into master... no need to keep tons of branches around16:16
wpwraklet's see if my Ministry of Truth operation was subtle enough ...16:16
GitHub60[flickernoise] wpwrak pushed 2 new commits to master: http://git.io/USGy_Q16:16
GitHub60[flickernoise/master] demo/pacman/pacman.fnp: update to use  ... if cond; - Werner Almesberger16:16
GitHub60[flickernoise/master] Merge branch 'if-modifier' - Werner Almesberger16:16
wpwrakhmm. it rejected non-ff patches ;-(16:16
wpwrakah, but that was in if-modifier. excellent16:17
wpwrakso .. if ffmpeg isn't coming yet, then i'll close imaginarium, too16:19
wpwrakhmm. would be nice to be able to change the line width of the wave. sometimes you get nice overlaps in raindance if the circle is small but when growing things, the lines get too thin to produce the same effect16:25
lekerneltried wave_thick ?16:28
wpwrakah, interesting. learned yet another thing :) let's see how it goes ...16:30
wpwrakcool. thanks !16:42
Action: Fallenou bying a Korg nanocontrol216:43
Fallenoublack :)16:43
GitHub139[flickernoise] sbourdeauducq pushed 1 new commit to master: http://git.io/WDXSPA16:55
GitHub139[flickernoise/master] Remove legacy midiX variables - Sebastien Bourdeauducq16:55
wpwrakwe should actually add some preprocessor capabilities to FN. like macros.16:56
GitHub103[flickernoise] wpwrak pushed 4 new commits to master: http://git.io/zel8vg17:10
GitHub103[flickernoise/master] raindance.fnp: enc1_push corrections - Werner Almesberger17:10
GitHub103[flickernoise/master] raindance.fnp: use enc3_push to control the line thickness - Werner Almesberger17:10
GitHub103[flickernoise/master] Merge branch 'master' of github.com:milkymist/flickernoise - Werner Almesberger17:10
Fallenouhttps://twitter.com/#!/yannsionneau/status/173831122457862146 picture of data load+store test running18:07
lekernelhmm... navre tends to crash sometimes, eg when booting with MIDI controllers connected18:23
wpwrakyes, that's one of the yet unsolved USB problems18:39
wpwrakhmm. we have an interesting memory corruption in the symbol table as well. need to make those pointers more long-lived  ...18:43
wpwrakhits patches with a lot of user-defined variables. such as raindance.fnp :)18:44
GitHub141[flickernoise] wpwrak pushed 2 new commits to master: http://git.io/IW_MjQ18:56
GitHub141[flickernoise/master] raindance.fnp: green button of channel 8 controls whether to wrap or to smear - Werner Almesberger18:56
GitHub141[flickernoise/master] raindance.fnp: alternate settings for the first four faders - Werner Almesberger18:56
lekernelwpwrak: have you run into this USB problem with only the keyboard/mouse connected?19:01
wpwrakno, but recently i've had usb-midi connected almost all the time19:02
wpwrakso that may not mean a lot19:02
wpwrakthe problem may also come from full-speed, not usb-midi per se19:02
GitHub193[flickernoise] wpwrak pushed 1 new commit to master: http://git.io/Wswgqw19:08
GitHub193[flickernoise/master] compiler: quick and dirty work-around for having too many identifiers - Werner Almesberger19:08
wpwrakFN should be happy again. even with the raindance monster19:09
wpwraki should probably use an RB tree for these user identifiers than a linear list anyway. well, maybe after a nap.19:10
wpwrakby the way, if you write the "logic analyzer", then i'll adapt the usb decoder for it :)19:19
wpwraki'm thinking of the following data format: 8 data lines. an 8 bit tag. and a 16 bit idle counter. data is only output if one of the inputs changes, if the counter reaches 65535, or if a tag is set. the tag would be a register or similar that can be written by whatever subsystem we're playing with. by default, perhaps through CSR.19:21
wpwrakthe counter would count one per cycle in which nothing else happens. that way, the data rate would drop by a factor of 65535 is the inputs are idle (e.g., between messages)19:23
wpwrakwith such a thing, it should be a pleasure to debug nasty little usb problems19:23
Fallenouyou would use a M1 to build such a sniffer ?19:29
Fallenouto debug another M1 ?19:29
kristianpaulsouds scary ;-)19:31
wpwrakor to debug itself19:36
lekernelwpwrak: where is midi_dev freed in parser.y ?19:37
wpwrakit's not. this variable passes data inside the parser. the data is freed by stim_db_free19:40
wpwrak(called from compiler.c, when verything is done)19:41
lekernelwpwrak: struct sym -> stim != NULL means that the variable is coming from the stimulus (atm only MIDI) system right?19:51
lekernelwow, raindance is super cool19:57
wpwraknow it's the control freak edition ;-)19:58
wpwrakuses something like 33 midi controls19:59
wpwrakwould be interesting to use images instead of a wave as the pattern "seed". that way, we could also fill the remaining buttons ;-)20:00
kristianpaulall this new midi stuff is very cool, but does it work with all consoles or i can buy one with my eyes closed?20:01
wpwrakraindance uses differential decoders. you'd have to hack things a little to achieve the same functionality with non-differential ones20:03
lekernelkristianpaul: buy one with your eyes closed, and send us bugfixes if it doesn't work :P20:03
wpwrakalso, some controller may not have enough controls :)20:03
wpwrakbut the main question is how the device feels20:04
wpwrake.g., pacman uses the joysticks for coordinates. you can of course do the same with a pair of faders or pots, but will it fell the same ?20:04
kristianpaullekernel: ;-)20:05
kristianpaulfeels is a good point, i guess pots by it self are not enought cool :-)20:06
lekernelgrmbl, seems we have a regression in the web update... "failed to download bitstream"20:06
lekernelthe version numbers are downloaded correctly though ...20:07
kristianpaulwpwrak: a touch screen is a must have :-)20:07
kristianpaulwpwrak: or what do you think about touch feelings?20:07
GitHub169[flickernoise] sbourdeauducq pushed 1 new commit to master: http://git.io/7pmJrQ20:08
GitHub169[flickernoise/master] Flag patches requiring stimulus and exclude them from simple mode - Sebastien Bourdeauducq20:08
wpwrakah, good point :)20:09
wpwrakkristianpaul: touch screen would mean a lemur or a tablet. there are a few choices for touch pad, but they're either expensive or have odd issues (such as the super-low resolution of the icon i-creativ)20:10
GitHub59[flickernoise] sbourdeauducq pushed 1 new commit to master: http://git.io/B8iaIg20:12
GitHub59[flickernoise/master] webupdate: fix download folder - Sebastien Bourdeauducq20:12
kristianpaulha, nanoonte could do HID isnt.. :-)20:16
kristianpaulbut feels..20:16
kristianpaulwhat about this http://articulo.mercadolibre.com.co/MCO-20260948-controlador-midi-korg-nanokontrol-_JM#questionText ?20:17
wpwrakit has a successor, the nanoKONTROL220:20
wpwraksupposedly, that one is better20:20
wpwraki have it (the nK2). it's cheap, in every sense of the word20:20
wpwraksome people like it, though20:21
kristianpaulha, found it http://articulo.mercadolibre.com.co/MCO-20778505-controlador-korg-nano-kontrol-2-negro-_JM20:22
kristianpaulchepear also :)20:22
Action: Fallenou just bought it on ebay20:39
Fallenoufor 50 EUR you can't expect much from it20:40
FallenouI guess it's great to have so much controls for that less money20:40
Fallenouthat few*?20:40
kristianpaulindeed lots of controls20:42
kristianpaulis that all about, isnt?20:42
FallenouATM I have 0 control :p20:43
FallenouI have to start with something20:43
kristianpauli hate experiment FN since i had to set variable values by hand20:44
kristianpaulso this could change my experience20:44
kristianpaulFallenou: not zero, you forgot mid, treb and bass ;-)20:45
Fallenouhehe sure20:45
Fallenouand time20:45
kristianpaulan yes !20:45
FallenouI still have time too20:45
kristianpaulwell, OSC too but is too generic so well..20:46
wpwrakyou can route MIDI through OSC :)20:47
wpwrakFallenou: "so little money" would probably be best20:47
kristianpaulah well, for remote control is that good option wpwrak20:47
Fallenouwpwrak: ah yes :' too much wine I don't speak english anymore20:48
wpwrakkristianpaul: also works controller---(usb-midi)-->pc---(ether)-->m120:49
wpwrakFallenou: ah, already celebrating the mmu success. good :)20:49
kristianpauli dont have the money yet, but Fallenou let us know how it goes when it arrives !20:49
kristianpaulwpwrak: why such re-routing midi that way?20:49
Fallenou(celebrating mmu) ahah too early to celebrate, but yes anything is worth celebrating :p20:50
wpwrakkristianpaul: e.g. to remap control numbers (before we had the new midi system)20:51
wpwrakkristianpaul: or to merge devices and save usb ports ;-)20:51
kristianpaulah ! HUB :-)20:52
kristianpaulgood idea20:52
lekernelok, didn't find major bugs... so two problems (1) USB crash (2) symtab initial alloc20:58
lekernelnothing else?20:59
wpwrakUSB has three issues, as i posted to the list. that crash if but one of them.21:00
wpwrakthere's may still be a code generation issue. but not i'm not sure if it isn't another face of the symtab allocation. it caused some variable updates the be lost when reordering the equations in the wheel demo21:02
wpwrakinterestingly, the variant that cause trouble was less correct than the current one :)21:02
lekernelok. let's wait a bit and you take a look at it?21:03
wpwrakin terms of correctness, the symtab initial alloc is as good as solved with my hack. you'd run out of registers before hitting the new limit21:03
wpwrakalright. i'll have a go at it tomorrow21:03
lekernelcool, thanks!21:04
lekernelah, another bug: the "full screen" button in video-in doesn't use the current settings21:19
wpwrakthat kinda seems to defeat its purpose ?21:20
wpwrak05:21 for xiangfu. he'll have almost all day before we catch up ;-)21:21
lekernelwhat do we do for the MIDI USB crash? push the web update out, but do not announce the MIDI-USB support yet?21:23
lekerneland keep it for people who like to try experimental and unstable stuff?21:24
lekernelwith mouse/kbd only it seems ok... but MIDI crashes pretty often21:24
wpwrakwe can label it "experimental"21:25
wpwrakusb still has 1-3 issues anyway. e.g., all those rx timeouts are some full-speed issue.21:25
lekernelyes, but those aren't as bad as a complete loss of USB when plugging something21:26
lekernelit seems like the M1 completely froze or something21:26
wpwrakdid it ? i see the following behaviours:21:26
lekernelwell, the softusb core did21:26
wpwrak1) M1 comes up just fine with kbd and midi (often)21:27
lekernelbut the RTEMS shell etc. are still working21:27
wpwrak2) M1 comes up without midi but with kbd (rare)21:27
wpwrak3) M1 comes up, rendering, but without any USB (often)21:27
lekernelyeah, but if you don't have rendering configured at power up, it looks like a freeze21:27
wpwrak4) M1 stays in in bios, while an endless stream of usb rx timeouts scrolls by (often)21:28
wpwrakyup. if it's not doing anything on its own and you have no means to tell it what to do, it does indeed look frozen :)21:29
lekernelnote that hotplugging USB-MIDI can also make it go into #321:29
wpwrakor from 2) to 1) :)21:29
wpwraklekernel: hmm, you should send bug #35 to xiangfu. it's his code. or did that go to everyone on milkymist ? (the mail only shows me as a recipient)23:05
Fallenouit's sent to everybody I think23:10
FallenouI received it as well23:10
wpwrakah, then it's just github being sneaky :)23:29
Fallenougn8 !23:33
--- Mon Feb 27 201200:00

Generated by irclog2html.py 2.9.2 by Marius Gedminas - find it at mg.pov.lt!