#milkymist IRC log for Monday, 2011-06-06

CIA-29flickernoise: Xiangfu Liu master * rb264ce5 / src/oscsettings.c : oscsetting: add autobuild feature - http://bit.ly/kauoDW05:21
terpstralekernel, there is no carry flag on the LM3207:18
terpstrayou use a comparison between the two summands and use add the result into the high word07:18
terpstraie, to sum two 64-bit values in [r1 r2] and [r3 r4]07:22
terpstraadd      r2, r4, r207:22
terpstracmpgu    r4, r4, r207:22
terpstraadd      r1, r3, r107:22
terpstraadd      r1, r4, r107:22
xiangfulekernel: if this patch ok? http://dpaste.com/551016/09:32
xiangfuthe only problem for me, is the same key give me different value some time.09:32
xiangfuand single press, there are 2~3 values sometimes09:33
lekerneler, what is it supposed to do?09:33
lekernelremove the capture button?09:34
lekernelwhy?09:34
lekernelif you have inconsistent values, either your remote isn't RC5 or there is a bug somewhere09:34
lekernelconcealing the problem by hiding the values is obviously not ok09:35
xiangfuthen if there is consistent values. we don't need 'capture' button at all. because you will need that when you open the windows09:35
lekernel?09:36
lekernelthe purpose of the "capture" button is to grab the code that is sent from a particular key on the remote09:36
lekernelso you don't have to look up RC5 specs and just press the button you want instead09:36
lekernelah, sorry, I didn't see you were always enabling capture mode09:39
xiangfuhmm.. I mean. 1. when open the  IR window, we should just add the 'ir input event'   2. press keys in remote controller 3, add file name. 4. press 'add/update'09:39
lekerneluhm, well, ok then09:39
lekernelbut do the same for MIDI09:39
xiangfuoh(sorry for bad/slow English)09:39
xiangfuok. I will commit IR first. then MIDI09:40
wolfspraulwe renamed something to 'ssd'09:49
wolfspraulI'm confused about that, I don't think that name is very good.09:49
wolfspraulssd = solid state drive?09:49
lekernelyes09:50
lekernelit's better than "flash" which people confuse with the macromedia thing09:51
lekernelssd either means nothing or "solid state drive", which is pretty close to what the flash filesystem is09:51
wolfspraulxiangfu: can you explain more about the inconsistent remote control values you see?09:52
wolfspraullekernel: OK I understand the 'flash' problem, but ssd...09:52
wolfspraulmaybe just data or so?09:53
wolfspraulwell, it's a detail09:53
wolfspraulthis kind of cryptic thing is bad, generally. people feel they are too stupid to use the product.09:53
xiangfu(inconsistent value) when I press one button many times. it's give me 0c, 08, 2c.09:54
wolfspraullet's do a test. press the same button 20 times, how many 0c, how many 08, how many 2c?09:54
wolfspraullekernel: fyi - these inconsistencies are with the 'gold standard' rc-5 control Adam is using for testing too, so I'm surprised we see this kind of inconsistencies. Want to find out what's going on.09:55
xiangfu0x0C: 10, 0x2C: 6, 0x08: 4.09:57
wolfspraulhow about other buttons?09:58
xiangfusome keys have 4,5 different values.09:58
wolfspraulthis looks very random to me. does it work at all?09:58
wolfspraulxiangfu: obviously this is not working right now ;-)09:58
wolfspraulxiangfu: can you press a few more buttons 20 times and tell us what you see?10:06
wolfspraulare you using the remote through the acrylic case? can you try to remove the acrylic on that side? just want to rule out that the acrylic is causing any problem...10:09
xiangfuok10:12
lekernelwhat is the value supposed to be, according to the rc5 standard?10:14
lekernelalso it works reliably for me (as usual) with the remote I have...10:15
lekernelssd isn't cryptic, it's all over computer advertisements10:15
xiangfuok. then the remote controller is not RC5 probability10:18
wolfspraulno way. this is the remote controller Adam sent you months ago for testing. and he uses for testing too.10:19
wolfspraulAdam is off on a public holiday today, we'll ask him tomorrow to test on his end too10:20
wolfspraulxiangfu: how many remote controllers do you have that you think are rc-5?10:20
wolfsprauldo all of them show these inconsistencies?10:21
wolfspraulcan you try to press the same key 20 times on each of them, and tell us what you get?10:21
xiangfu4. all have inconsistencies10:21
xiangfuthere is not same key. all remote controller different10:21
lekernelxiangfu, the rc5 standard specifies a code that is associated to a particular key10:22
lekernelfor example keys 0-9 have codes 0x00-0x0910:22
lekerneldo you get those codes sometimes, or something totally different?10:22
wolfspraulxiangfu: from now on. if there is an inconsistency, the remote control does not work!10:22
wolfspraulinconsistency obviously is unacceptable to an end user :-)10:23
wolfspraulso you have 0 remote controls right now that work10:23
wolfspraulgood point - do you have any control that gives you a value of 0x05 when you press '5'?10:24
xiangfuno10:26
xiangfulekernel: something totally different10:27
lekerneliirc there is a "dump remote" command in the demo firmware, try that10:28
lekernelif you still get inconsistent values there, it's either a fpga design problem or an incompatible remote10:29
xiangfuwhere is demo firmare? not in github/flickernoise.git10:29
wolfspraulmaybe none of his remote controls are actually rc5?10:29
lekernelin milkymist.git10:30
lekernelsoftware/demo10:30
xiangfuwolfspraul: probability10:30
wolfsprauland Adam also is using a non-rc5 control for his testing, and sent Xiangfu the wrong one too?10:30
wolfspraulwell, we have to clear this up fast :-) good that we are focused on this now10:30
wolfspraulxiangfu: from now on - please help us find _WORKING_ remote controls :-)10:30
wolfspraulI will go shopping now, try to find some too.10:30
wolfspraulI have 3 here, none of them work.10:30
lekernelphilips brand should be ok10:30
xiangfuthe one Adam send to me. not working.10:31
wolfspraulwe didn't really focus on this before, but now we do10:31
wolfspraulmaybe so far Adam's testing was really just to see _any_ value10:31
wolfspraulI think I vaguely remember seeing 0x2B/2C values in the rc2 testing, let me check the logs...10:32
xiangfu(_any_ value) for me it is. :(10:32
wolfspraulwith the 3 controls I have here I also get 0x2A/0x2B values sometimes, but obviously overall the thing is not working for me10:32
wolfspraulthe test log only says "press the remote's power button" http://downloads.qi-hardware.com/hardware/milkymist_one/production/rc2/test_results/0a-results10:33
lekernelwolfspraul, read the source ? :)10:34
wolfspraulyes it checks for value 0x0c10:36
wolfspraulxiangfu: can you press the power button on the remote Adam sent you? what value do you get?10:36
wolfspraulif you press the power button 20 times, are there inconsistencies?10:37
xiangfuno power button, only 'standby'10:37
xiangfuI take a picture of the remote controller.10:38
xiangfuone memont.10:38
wolfspraulmaybe just press that button...10:38
xiangfuit give. 0C, 26, 0E10:39
wolfsprauldo you have the acrylic on or removed?10:39
xiangfuremoved.10:39
wolfspraulxiangfu: which image are you running? can you try to go back to some older images, or the msd ones from Sebastien, to see whether some of them work?10:41
wolfspraulmaybe focus on that power (standby?) button, and try to get a consistent 0x0c10:42
wolfspraulyou can also try the testing images, we probably still have them somewhere10:43
wolfspraulwhen testing the 40 boards from rc2, I don't remember ever running into this test failing10:43
wolfspraulso if we had the 0x0c check back then, there shouldn't be a way that an inconsistency slipped through somehow, I should have seen that at least a few times on those 40 boards (which were tested several times too)10:44
xiangfuwolfspraul: I amd using test image now.10:48
xiangfuwolfspraul: some keys give [PASSWD]. mute keys always give [PASSED]10:48
xiangfumute keys. 10 times, 9 times give [PASSED].10:49
xiangfuI never got a [FAILED], most keys have no raction.10:50
xiangfuonce test image raction, it give [PASSED]10:52
xiangfuok. there is no [FAILED] in source code. :_10:53
xiangfu:)10:53
wolfspraulwait wait wait10:54
wolfspraulI don't understand10:54
wolfspraul10 times, 9 times pass is also not enough. it needs to be 100%10:55
wolfspraulwhy do 'some other' keys pass as well?10:55
lekernelbecause the RC5 decoder gets some random garbage10:55
wolfspraulso you are saying with the test software at least in 90% you get a 0x0c value?10:56
lekerneland this garbage happens to contain 0x0c10:56
lekernelperiod10:56
wolfspraullekernel: you think none of those controls are rc5, and that explains the problems?10:56
lekerneldo you have a logic analyzer? check at the output of the IR sensor that you get RC5 frames10:56
lekernelmaybe10:56
xiangfuI don't have that. :(10:56
lekernelyou can try xilinx chipscope, if you dare ...10:56
lekernelbut I never got it to work10:57
lekernelmaybe Adam has a proper one?10:57
wolfspraulxiangfu: when you bought some of your remotes, did they specifically say 'rc-5'?10:57
lekernelor just a digital oscilloscope should be ok ...10:57
xiangfuwolfspraul: no specifical say 'rc-5'10:57
xiangfuwolfspraul: now. I just order three. all them specitical rc-5. just it's not the small one. all big remote controllers10:59
wolfspraulok10:59
wolfspraulyou mean you just ordered them?10:59
wolfspraulso they arrive in a few days?10:59
xiangfuyes. I just ordered them on taobao. should be arrive in 2~3 days.11:00
xiangfuthe controller adam send me is like:11:07
xiangfuhttp://en.qi-hardware.com/wiki/File:Adam_remote_controller_2.jpg11:07
xiangfuhttp://en.qi-hardware.com/wiki/File:Adam_remote_controller_1.jpg11:07
xiangfulekernel: please check the midi patch(remove capture):  http://dpaste.com/551058/11:30
xiangfuI don't have a device that can test MIDI. :(, I found there is a Channel 0~15, is that also from MIDI? or must input by user?11:31
wpwrak(rc5) why rc5 ? there's a gazillion different protocols. may be easier to just work with general patterns after AM demodulation.11:53
rohwpwrak: why rc5? i think because its 'a standard' not 'none' .)11:59
wpwrakroh: well, it's a bit like setting up an international web shop that accepts only payments in uruguayan pesos. sure, it's a well-enough established "standard", but ...12:00
rohcomeon... rc5 isnt at all 'seldomly'12:00
rohits rather like 'euro' or 'us$' compared to stuff avail12:01
wpwrakroh: some time ago, i experimented a bit with IR remotes. they basically all exhibited wildly different patterns. i didn't even bother to try to figure out what higher layers could be there, just treat this as a stream on on/off intervals.12:04
rohsure. just makes for more ugly decoders. stuff ends up looking like lirc (detecting on-off pattern times ;)12:05
wpwrakroh: maybe 1-2 in ten were rc5. maybe 1-2 in then "sony". the bottom line is that there's not really any standard.12:05
wpwrak(ugly decoders) yes, ugly decoders that work ;-)12:05
rohwpwrak: true. there are multiple ones. get one from philips and it will work12:05
wpwrakvs. theoretically perfect decoders that will work with a perfect remote control. if xiangfu buys a batch of "rc5" remotes and none of them works, does that suggest something about the fun prospective end users will have ? ;-)12:06
roheh. simple: dont buy batches of untested stuff ;)12:07
lekernelwpwrak, then maybe you'd like to port your wonderful remote control recognition software to MM? :)12:08
wpwraklekernel: hihi, it's actually just record and playback. and runs on a PIC. one of my very first experiments with an MCU ;-)12:12
lekernelthere are tons of stuff in MM which are "actually just xxx" :)12:12
wpwrakmaybe LIRC already has what you need in terms of remote control recognition ? they ought to be quite flexible on the driver side12:14
rohwpwrak: they detect space and mark, and do patterns from there.12:14
rohflexible yes. but pain to confiugure. in the end it doesnt matter. deliver a remote that works.12:15
rohcompat. with 10-20% of the market is already more than usual for ir-stuff.12:15
wpwrakroh: space/mark. yup, i think that's about the lowest common denominator. well, besides fun with the carrier frequency. (there are all sorts of carriers, several in 30-39 kHz, 56 kHz, and i'm sure there's more)12:16
wpwrakroh: picking one "standard" remote to ship with the device is an entirely different story. there you're free to put whatever requirements you fancy.12:17
rohanyhow. need to run now. but thats a 'enough battles, this one isnt worth it' issue ;)12:19
rohyou can make anything better. sometimes it doesnt matter. its good enough for now (and doesnt need any changing) .. btw.. the 38khz or what filter is already in the ir-reciever usually, so that is the only thing fixed by hw12:20
wpwrak(38 kHz filter) yeah. usually not much choice there, if you use a fully integrated receiver.12:23
wpwrak"roh withdraws from battle without admission of defeat" so it shall be recorded ;-)12:25
wpwrakroh: whee, oddly on topic: http://www.heise.de/newsticker/meldung/Apple-will-Datenuebertragung-via-infrarotem-Licht-patentieren-lassen-1255791.html12:27
wpwrakroh: remotes are just a special case ;-)12:27
xiangfuif we can modify the Rc-5 to another protocol in SOC code? I checked the datasheet of IR receiver in M1, it seems support all protocaols.12:32
lekernelyes of course but as nearly always in this project: just fucking do it.12:33
lekernelthere are more important things however ...12:34
xiangfulekernel: ok. just for understand.12:36
xiangfuwolfspraul: check this  ^, I though we can not do this before :(12:36
lekernelwell it's not the first time that companies like apple come up with this kind of idea12:38
lekernelsony and macrovision are also champions on that12:39
xiangfulekernel: please check the midi patch(remove capture):12:41
xiangfu  http://dpaste.com/551058/12:41
wpwrakthe nice thing about patents is that they're not elitist at all. even the dullest engineer will have a dozen ideas every day that would be worth a patent or two ;-)12:41
lekernellooks ok12:41
lekernelwell, you need a significant sum of money to have patents12:42
lekernelthe application itself costs, plus the lawyers you need to speak legalese and get the necessary amount of red tape12:42
wpwraklekernel: i was trying to be sarcastic :) attacking the aspect of "an idea worth excessive amounts of privilege"12:50
lekernelpatents aren't even that, they're just yet another tool that bureaucrats use to get power12:51
terpstralekernel, is it possible you took the wrong power supply with you?14:18
lekernelmh, probably not... back home, unpacked everything and didn't find anything that wasn't mine14:20
lekernelwhat does yours look like?14:21
terpstrait's 12v +center14:28
terpstraand i have an unknown power supply on my desk14:28
terpstra16v +center ... that isn't mine14:28
terpstrabut, anyway, maybe one of my colleagues borrowed it. nevermind.14:29
lekernelI don't have anything 16V14:30
wpwraki think i'll keep that quote for my "famous last words / self-fulfilling prophecy" collection ;-)14:58
--- Tue Jun 7 201100:00

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