**** BEGIN LOGGING AT Fri Aug 08 02:59:59 2014
Aug 08 04:31:38 <Abhishek_> ds2: that's actually the template in WaveDrom
Aug 08 10:03:34 <praveendath92> av500: vvu Testing with video playback - https://www.youtube.com/watch?v=4euqsUcTkfg
Aug 08 10:05:30 <av500> praveendath92: nice!
Aug 08 10:05:32 <av500> this is on BBB?
Aug 08 10:05:55 <praveendath92> av500: Yep.
Aug 08 10:06:01 <av500> very good
Aug 08 10:06:19 <praveendath92> Everything is on BBunless specified otherwise :)
Aug 08 10:06:45 <praveendath92> This is after fixing issues with earlier code.
Aug 08 10:07:01 <praveendath92> Haven't integrated the compression yet.
Aug 08 10:07:24 <av500> oh
Aug 08 10:07:30 <av500> so this is all uncompressed?
Aug 08 10:07:41 <av500> not bad
Aug 08 10:07:43 <praveendath92> Yep.
Aug 08 10:07:45 <av500> what resolution?
Aug 08 10:07:52 <av500> can you resize on the android side?
Aug 08 10:07:53 <praveendath92> 1024x768
Aug 08 10:07:57 <av500> make it use the full screen?
Aug 08 10:08:15 <praveendath92> My tablet screen is 2500x1400
Aug 08 10:08:17 <praveendath92> !
Aug 08 10:08:19 <av500> yes
Aug 08 10:08:27 <av500> but you should be able to resize the surface
Aug 08 10:08:58 <praveendath92> You mean stretch it to the big screen?
Aug 08 10:09:01 <av500> yes
Aug 08 10:09:19 <praveendath92> That is possible. Just an attribute to the imageView :)
Aug 08 10:09:25 <praveendath92> Will test it now.
Aug 08 10:09:27 <av500> :)
Aug 08 10:11:50 <praveendath92> Had issues with reader being garbage collected earlier. So, I had to fix it before jumping to compression.
Aug 08 10:13:54 <vvu> praveendath92: yeyyy we have video
Aug 08 10:14:09 <praveendath92> vvu: Yep :)
Aug 08 10:14:12 <vvu> praveendath92: but what was with the GC ?
Aug 08 10:14:32 * vvu needs to go in the lab to solder soon...
Aug 08 10:15:00 <praveendath92> It was related to threading.
Aug 08 10:15:14 <praveendath92> I'm using a well tested code from rowboat now,
Aug 08 10:15:52 <vvu> 'well tested' unless proven otherwise
Aug 08 10:17:03 <vvu> praveendath92: what hapens if you do not have the device connected but u start the x server ?
Aug 08 10:17:06 <vvu> and stuff like that ?
Aug 08 10:20:12 <av500> GC?
Aug 08 10:20:26 <av500> praveendath92: do I see right that the CPU load on the bone is high?
Aug 08 10:20:40 <av500> but that might be video decode too
Aug 08 10:37:49 <praveendath92> vvu: Well tested as in the android receiver code.
Aug 08 10:38:04 <praveendath92> Not the kernel driver code.
Aug 08 10:38:19 <praveendath92> av500: yes. That's mostly because of video playback.
Aug 08 10:38:41 <praveendath92> I tested wo video playback and the cpu usage wasn't peaking.
Aug 08 10:39:00 <praveendath92> It was calm at the bottom.
Aug 08 10:42:18 <praveendath92> av500: vvu Full screen version - https://www.youtube.com/watch?v=6UwmaWeSPdc
Aug 08 10:43:43 <praveendath92> My Nexus 10 aspect ratio is 16:10 while the video's is 16:9. Hence the side spaces.
Aug 08 11:09:13 <av500> sure
Aug 08 11:09:43 <av500> praveendath92: do you think you can get mouse support in short time?
Aug 08 11:09:46 <av500> pencil down is soon
Aug 08 11:09:56 <av500> and it would make for a nice finale
Aug 08 11:10:12 <av500> and for a nice demo video
Aug 08 11:10:29 <praveendath92> I will see about mouse usage.
Aug 08 11:10:44 <av500> should be easier than kbd
Aug 08 11:11:00 <praveendath92> I will also try with a USB hub today.
Aug 08 11:11:06 <praveendath92> Just in case.
Aug 08 11:16:06 <vvu> me gusta
Aug 08 13:26:33 <Abhishek_> praveendath92: wow
Aug 08 13:35:35 <Abhishek_> Gm jkridner
Aug 08 13:36:49 <Abhishek_> jkridner : http://s30.postimg.org/6pyznn2kf/Screenshot_from_2014_08_08_05_35_47.png .  Also fixed that disconnect using setTimeout.
Aug 08 13:37:14 <jkridner> saw your previous note about the disconnect.
Aug 08 13:37:30 <jkridner> looking good!
Aug 08 13:37:44 <av500> jkridner: - https://www.youtube.com/watch?v=6UwmaWeSPdc
Aug 08 13:38:21 <jkridner> praveendath92: super!
Aug 08 13:40:28 <jkridner> isn't that copywritten material? ;-)
Aug 08 13:41:28 <jkridner> praveendath92: should share some of the videos to the G+ group in the GSoC section
Aug 08 13:42:12 * Abhishek_ decoded his own voice using BeagleLogic on the hangout to work around using copyrighted stuff
Aug 08 13:42:47 <jkridner> Abhishek_: :-)
Aug 08 14:39:38 <praveendath92> jkridner: I wanted a video to quickly test it. The 1st thing I found was that.
Aug 08 17:20:56 <praveendath92> panto: ping
Aug 08 17:41:17 <panto> praveendath92, pong
Aug 08 17:43:52 <praveendath92> Hi
Aug 08 17:44:27 <praveendath92> I'm looking for a quick implementation of a USB driver that can go with my framebuffer driver.
Aug 08 17:45:12 <panto> what does that mean?
Aug 08 17:45:16 <panto> what kind of usb driver?
Aug 08 17:45:28 <praveendath92> Sorry. I meant a mouse driver.
Aug 08 17:45:42 <praveendath92> Just a couple ops like mouse pos and clicks.
Aug 08 17:46:45 <panto> just copy bits from drivers/input/mouse
Aug 08 17:47:25 <praveendath92> Data will be sent from USB (by Android)
Aug 08 17:47:32 <panto> irrelevant
Aug 08 17:47:43 <panto> you just need to inject input events
Aug 08 17:47:46 <praveendath92> Okiee.
Aug 08 17:48:24 <praveendath92> And about injecting the input events?
Aug 08 17:48:40 <panto> take a look at the drivers in drivers/input/mouse
Aug 08 17:50:10 <praveendath92> I will try that.
Aug 08 17:50:43 <mranostay> fair use has to come into play right? :)
Aug 08 18:16:03 <vvu> Abhishek_: you have a logo for beaglelogic ?
Aug 08 18:16:05 <Abhishek_> mranostay: yeah, IMO it counts as fair use but you never know when DMCA can strike :)
Aug 08 18:16:19 <Abhishek_> we are not lawyers
Aug 08 18:16:54 <Abhishek_> vvu: Not yet, but do you have suggestions? {I mean you must be looking for BARD logos right?]
Aug 08 18:17:04 <vvu> nop, just for my flasher
Aug 08 18:17:07 <vvu> wanna make a small icon
Aug 08 18:17:13 <vvu> but cannot find a Beagle icon :)
Aug 08 18:17:19 <vvu> in photoshop or smth that i can edit
Aug 08 18:19:04 <Abhishek_> vvu: ping jkridner
Aug 08 18:19:14 <Abhishek_> I guess he isn't here right now
Aug 08 18:19:28 <vvu> nop
Aug 08 18:20:56 <vvu> i think i will always hate windows development
Aug 08 18:21:06 <Abhishek_> unless you use C# :)
Aug 08 18:22:20 <Abhishek_> seriously, what I mean to say is Windows-only development is damn easy but it is more difficult to develop something that is consistent across all platforms (Win / Mac / Linux)
Aug 08 18:22:38 <vvu> that is what i'm trying to do
Aug 08 18:22:41 <vvu> i am one man army here
Aug 08 18:22:56 * Abhishek_ also considers himself a one man army
Aug 08 18:23:16 <Abhishek_> vvu: Why not use mingw32 on Linux to compile for Windows?
Aug 08 18:23:21 <Abhishek_> *cross-compile
Aug 08 18:23:40 <Abhishek_> That's what the guys at sigrok do
Aug 08 18:24:36 <vvu> no worries here but usb stuff is more tricky
Aug 08 18:24:53 <vvu> i can't get the drivers installed
Aug 08 18:24:58 <Abhishek_> you using Pete Batard's library?
Aug 08 18:24:58 <vvu> winusb drivers
Aug 08 18:25:06 <vvu> libusb
Aug 08 18:25:07 <Abhishek_> Zadig?
Aug 08 18:26:52 <vvu> yes
Aug 08 18:27:18 <vvu> it says that the driver is installed right
Aug 08 18:27:24 <vvu> but have some awkward errors
Aug 08 18:30:06 <Abhishek_> like?
Aug 08 18:31:28 <vvu> http://pastebin.com/BjTjcKVA
Aug 08 18:40:20 <Abhishek_> vvu: did you try the libusb-win32 driver?
Aug 08 18:49:12 <vvu> all of them :)
Aug 08 18:49:22 <vvu> winusb libusb-win32 libusbK
Aug 08 18:49:25 <vvu> same error
Aug 08 18:49:37 <vvu> the problem is that i do not know the procedure exactly
Aug 08 18:49:44 <vvu> and there is no documentation explicit for this
Aug 08 18:55:10 <vvu> Abhishek_: so any hints ?
Aug 08 18:55:27 <vvu> it sees the winusb driver right, there is a call from there that it claimed intf 0
Aug 08 18:55:31 <Abhishek_> I do not yet know
Aug 08 18:56:16 <Abhishek_> but then when you try and claim interface 1 it's already claimed, right?
Aug 08 18:56:57 <vvu> don't think so
Aug 08 18:57:13 <vvu> the problem is that it cannot read some configuration descriptor
Aug 08 18:57:19 <vvu> don't even know what that is
Aug 08 18:57:33 <vvu> need to ask on libusb mailing list
Aug 08 18:57:43 <vvu> maybe i can get some help there
Aug 08 18:57:51 <Abhishek_> which line is that in the log?
Aug 08 18:58:33 <Abhishek_> vvu: join #libusb
Aug 08 18:59:26 <vvu> last line
Aug 08 18:59:30 <vvu> pasted there the log
Aug 08 18:59:35 <vvu> but nobody answered
Aug 08 19:24:22 <Abhishek_> vvu: which interface is which? I mean, how are the interfaces mapped (0: RNDIS, 1: Mass storage) etc ?
Aug 08 19:28:12 <vvu> no no
Aug 08 19:28:17 <vvu> intf 0 is config interface
Aug 08 19:28:20 <vvu> and 1 is the RNDIS one
Aug 08 19:28:26 <vvu> i am using the AM335x ROM
Aug 08 19:28:38 <vvu> 0 is always the config interface
Aug 08 19:34:47 <vvu> Abhishek_: i have 2 interfaces Config(1 endpoint which is interrupt) and RNDIS interface(2endpoints IN/OUT Bulk)
Aug 08 19:36:32 <Abhishek_> do you have an extra network connection in the "network connections" ?
Aug 08 19:37:03 <vvu> nop
Aug 08 19:37:05 <vvu> replaced the driver
Aug 08 19:37:08 <vvu> with the winusb one
Aug 08 19:37:38 <Abhishek_> strange.
Aug 08 19:38:39 <Abhishek_> I think the network driver might be holding back the endpt
Aug 08 19:41:01 <Abhishek_> oh, the problem seems to be with interface 0, configuration descriptor
Aug 08 19:54:06 <vvu> Abhishek_: it puts the BBB in a unknown state i think so
Aug 08 19:54:12 <vvu> because the rom code
Aug 08 19:54:25 <vvu> resets the usb bus if it did not find anything in a specified timeout
Aug 08 19:54:38 <vvu> and then it should disconnect and connect again
Aug 08 19:54:46 <vvu> but when i install winusb that does not happen
Aug 08 19:55:34 <Abhishek_> does it expect "magic bytes" to be sent over?
Aug 08 19:56:11 <vvu> it sends to the PC some bootp messages
Aug 08 19:57:18 <vvu> joel_: ping
Aug 08 20:19:25 <vvu> Abhishek_: eh just posted on libusb ML
Aug 08 20:19:30 <vvu> maybe i can get some help there
Aug 08 20:33:07 <alexanderhiam> hey jkridner
Aug 08 20:46:27 <Abhishek_> jkridner: were you able to test the web interface with live data?
Aug 08 22:16:46 <ds2> alexanderhiam: is botspeak done?
Aug 08 22:20:05 <alexanderhiam> ds2: karki has more long term goals for it, but it's definitely usable now
Aug 08 22:22:01 <ds2> alexanderhiam: 'k - was trying to see if we need to do any last minute technical brainstorming to get GSoC14 wrapped up
Aug 08 22:22:41 <alexanderhiam> I'm not sure if karki's planning any new features before Monday
Aug 08 22:26:39 <alexanderhiam> ds2: actually, I don't believe the ADC has been implemented yet. I'm not sure if that's been eliminated from the gsoc goals, but if no it might be a scramble to get that done
Aug 08 22:32:57 <ds2> alexanderhiam: ADC via PRU or ADC in general?
Aug 08 22:33:13 <ds2> ADC is pretty trivial
Aug 08 22:34:24 <alexanderhiam> ds2: ADC from PRU while keeping the kernel happy
Aug 08 22:34:40 <ds2> what do you mean by keeping kernelhappy?
Aug 08 22:35:00 <ds2> in general or for the hacked up PoS's floating around?
Aug 08 22:35:34 <alexanderhiam> :P
Aug 08 22:36:05 <ds2> let me clarify that -
Aug 08 22:36:14 <ds2> the driver floating around for the ADC is total crap
Aug 08 22:36:28 <ds2> tried using it and wind up having to redo it entirely
Aug 08 22:36:46 <ds2> so if you mean working around it w/o chucking it, that is harder then just chucking it
Aug 08 22:37:10 <alexanderhiam> yeah, it would be best to disable the driver and interface straight to the ADC at the register level, but then that's a problem for anything using the driver
Aug 08 22:37:33 <alexanderhiam> and there's not much time to write a new driver...
Aug 08 22:37:52 <ds2> indeed
Aug 08 22:38:15 <ds2> it is really too bad folks are insisting on IIO when people aren't using that
Aug 08 22:38:40 <ds2> in anycase, if it comes to it - I have a (crappy code) driver that will send out regular samples on a /dev entry
Aug 08 22:39:19 <alexanderhiam> ok
Aug 08 22:39:44 <alexanderhiam> of course going through a kernel driver brings up timing issues as well
Aug 08 22:39:45 <ds2> i needed audio input
Aug 08 22:39:49 <ds2> what timing issues?
Aug 08 22:40:02 <ds2> I am doing audio and there are no timing issues... noise issues, yes but you get that regardless
Aug 08 22:41:15 <alexanderhiam> I mean as far as the event loop goes, though I guess it could just grab values as fast as it can and keep them in the PRU memory
Aug 08 22:41:35 <ds2> eh?
Aug 08 22:41:38 <ds2> there is a FIFO on the ADC
Aug 08 22:41:48 <ds2> and the ADC can be configured to sample automatically based on its ICLK
Aug 08 22:41:51 <ds2> (or FCLK)
Aug 08 22:42:08 <ds2> it is a matter of clearing the FIFO on interrupt
Aug 08 22:42:44 <ds2> now if you can't keep up with interrupts for a FIFO filling at 8KHz, you got bigger issues given the Cortex can be running at up to 1GHz
Aug 08 22:43:15 <alexanderhiam> right, but for the PRU to set up the ADC sampler would be bypassing kernel drivers wouldn't it?
Aug 08 22:44:07 <ds2> sure
Aug 08 22:44:34 <alexanderhiam> maybe that's inevitable
Aug 08 22:44:44 <ds2> just don't use the kernel driver OR make sure there is no overlap of the FIFOs
Aug 08 22:44:49 <ds2> there are 2 FIFOs
Aug 08 22:44:55 <ds2> but why run it though the PRU?
Aug 08 22:45:56 <alexanderhiam> that's what I was getting at before, the pruspeak driver could just grab values from the adc driver as fast as it can and stick them in shared memory for the PRU to read when it needs to
Aug 08 22:46:17 <alexanderhiam> I don't think there's a real need for high speed ADC sampling in botspeak
Aug 08 22:46:58 <ds2> are you guys trying to do real processing on the PRU? (vs using PRU for acquisition/outputs)?
Aug 08 22:47:44 <ds2> for that, I wonder if you can write a kernel driver to setup DMA between them
Aug 08 22:47:51 <alexanderhiam> botspeak doesn't lend itself to doing any sort of real processing
Aug 08 22:48:12 <ds2> hmm I am confused...why would the PRU itself want the ADC values then?
Aug 08 22:49:07 <alexanderhiam> the idea (as I understand it) is more to quickly write custom I/O drivers that run on the PRU, then interface to them from userspace programs to do processing
Aug 08 22:50:15 <alexanderhiam> you need to be able to respond to ADC values in the botspeak code, which runs entirely on the PRU
Aug 08 22:50:47 <alexanderhiam> e.g. 'IF AIN[0] > 100 GOTO 10'
Aug 08 22:51:14 <ds2> oh so kind of like having an analog trigger?
Aug 08 22:54:55 <alexanderhiam> sure. With an IR range finder and some DC motors you could easily have a complete object avoiding robot writtin entirely in botspeak.
Aug 08 22:55:41 <alexanderhiam> Because variables are all shared between the running script and the userspace API, you could then easily change its speed by setting a variable while the script is running
Aug 08 22:56:54 <ds2> interesting... didn't think there was enough code space to do that
Aug 08 22:57:06 <alexanderhiam> it's a bit tight
Aug 08 22:57:16 <ds2> heh
Aug 08 22:57:22 <alexanderhiam> the software pwm is done in a separate event loop in the other pru
Aug 08 22:57:31 <alexanderhiam> adc could be done there too
Aug 08 22:57:52 <ds2> i just don't see much value in doing ADC on the PRU
Aug 08 22:58:05 <ds2> it is going over the L3 and that'll screw up the PRU's timing anyways
Aug 08 22:59:06 <alexanderhiam> yeah, that's why I'm thinking non-real-time whenever-thery're-ready ADC values written to shared memory by the botspeak kernel driver would be fine
Aug 08 22:59:16 <ds2> *nod*
Aug 08 22:59:24 <ds2> it is worse then that
Aug 08 22:59:39 <ds2> the L3 access is blocking, AFAIK, so that'll screw up the PWM
Aug 08 23:00:11 <alexanderhiam> ah, good point
Aug 08 23:00:29 <ds2> now if there is a way to explicitly make it nonblocking, I would love to know... I have a similar issue - generating video output AND trying to go over the L3 to write to DDR :(
Aug 08 23:00:54 <alexanderhiam> fun
Aug 08 23:01:23 <ds2> hence the comments about lack of space
Aug 08 23:01:47 <alexanderhiam> what's that for?
Aug 08 23:02:48 <ds2> I am doing a HMD thing using NTSC video so I got a capture chip and NTSC out w/overlay running on the PRU
Aug 08 23:03:05 <ds2> PRU0 generates video, PRU1 samples the 27MHz bus from teh acquisition
Aug 08 23:03:16 <ds2> not much room left for nonRT access to DDR :(
Aug 08 23:03:22 <alexanderhiam> oh wow
Aug 08 23:03:26 <alexanderhiam> poor PRUSS!
Aug 08 23:03:38 <ds2> that's what it is made for!
Aug 08 23:04:58 <alexanderhiam> what's the input?
Aug 08 23:05:24 <ds2> there is a TI chip that takes analog in and converts it to ITU-656 video stream
Aug 08 23:05:40 <ds2> which is an 8 bit YUV formated parallel bus running at 27MHz
Aug 08 23:07:43 <alexanderhiam> so the overlay is applied to the ITU-656 stream?
Aug 08 23:07:51 <ds2> no
Aug 08 23:07:55 <alexanderhiam> oh
Aug 08 23:07:58 <ds2> it is done on the analog side
Aug 08 23:08:16 <ds2> the ITU-656 stream is to feed the cortex for doing facial recognition, OCR, etc
Aug 08 23:08:21 <ds2> or to save as a photo
Aug 08 23:08:30 <alexanderhiam> ah
Aug 08 23:08:35 <alexanderhiam> very cool
Aug 08 23:08:46 <ds2> overlay on the digital size is going to be too laggy
Aug 08 23:09:06 <ds2> can cause motion sickness if that happens on a HMD
Aug 08 23:09:12 <alexanderhiam> right
Aug 08 23:09:53 <ds2> i think for a lot of our problems, getting DMA sorted out would simplify things
Aug 08 23:11:43 <alexanderhiam> so all the data to overlay is going over the L3?
Aug 08 23:13:18 <alexanderhiam> with little space to buffer it on the pru
Aug 08 23:13:33 <ds2> no, I am using the DataRAM on the PRU as a framebuffer
Aug 08 23:13:39 <ds2> so it is a tad bit resolution limited :(
Aug 08 23:14:09 <ds2> right now, it is B&W. the real fun is when I can a colorburst seperator chip and have the PRU phase lock with the color burst to produce color output
Aug 08 23:14:48 <ds2> B&W also keeps the fb size down
Aug 08 23:17:23 <ds2> what would neat is if someone can implement all the DSS functionality from the OMAP2+ on the PRUSS ;)
Aug 08 23:18:58 <alexanderhiam> what's the application?
Aug 08 23:20:12 <ds2> just would be neat... the DSS does so much more then the dumb fb on the 335x
Aug 08 23:20:30 <alexanderhiam> oh, I mean of your HMD
Aug 08 23:20:41 <ds2> oh... experimental platform
Aug 08 23:20:53 <alexanderhiam> good answer
Aug 08 23:21:00 <ds2> problem is the existing HMD products are not easy to experiment and developwith
Aug 08 23:21:10 <ds2> everyone tries to lock you in to their crap
Aug 08 23:21:18 <alexanderhiam> yup
Aug 08 23:21:21 <ds2> glass likes to lock you to google ToS's, same with the others
Aug 08 23:22:20 <ds2> so far, most folks that try are just wrapping cloud services
Aug 08 23:22:39 <ds2> so more ToS handcuffs
Aug 08 23:23:44 <alexanderhiam> are you running Linux on the ARM?
Aug 08 23:23:50 <ds2> yep. stock linux
Aug 08 23:24:11 <ds2> well.. somewhat stock...right now, it is just a angstrom distro w/my own kernel
Aug 08 23:24:27 <alexanderhiam> API to push stuff to the HMD?
Aug 08 23:25:29 <ds2> that's a userland ducktape right now
Aug 08 23:25:43 <ds2> wanted to have something working by makerfaire/sv
Aug 08 23:25:52 <ds2> eventually, it should be a /dev/fb device
Aug 08 23:25:59 <alexanderhiam> awesome!
Aug 08 23:26:24 <ds2> the basic features work...but the adcs are noisy enough that the voice recogonition is blowing major chunks :(
Aug 08 23:27:09 <ds2> need to find time to do an analysis on the noise and if it can be worked around with a better ADC config or better filtering on the outside
Aug 08 23:27:25 <alexanderhiam> why not an external ADC and I2S for audio?
Aug 08 23:27:58 <ds2> cuz our friendly hardware designer exposed limited number of pins on the port and...
Aug 08 23:28:17 <alexanderhiam> ah
Aug 08 23:28:30 <ds2> it is useful to understand the limitations on the ADC
Aug 08 23:28:45 <ds2> so far everyone is hung up on driver issues
Aug 08 23:29:37 <ds2> actually... a thought just ocurred to me
Aug 08 23:30:18 <ds2> run the mic into some op-amps to generate a PWM'ed version of hte audio...then use the QCAP for digitizing
Aug 08 23:31:41 <alexanderhiam> interesting... you'd have the analog side separated so potentially lower noise
Aug 08 23:32:04 <ds2> yep
**** ENDING LOGGING AT Sat Aug 09 02:59:59 2014