**** BEGIN LOGGING AT Thu Feb 16 02:59:58 2012 Feb 16 04:40:17 hi Feb 16 04:40:52 av500 i'm gonna chuck this satellite next month -- i'm getting unlimited 3g device for $50 a month, no contract Feb 16 06:45:20 hmmmmm Feb 16 07:02:19 hi Feb 16 08:18:20 av500: whats on video? you trolling the arm presentation, or your arm trolling something completely different? :P Feb 16 08:25:25 av500 trolls anything he can :) Feb 16 08:26:15 i know, i was there. Feb 16 08:52:47 mranostay: the tizen people prevented effective trolling by holding the talk in the smallest room available Feb 16 08:53:11 av500: we're both thinking of the ARM BOF, right? Feb 16 08:54:06 LetoThe2nd: yes Feb 16 10:33:00 I'm hitting this issue when trying to build angstrom, following the instructions on http://www.angstrom-distribution.org/building-angstrom, http://build.sakoman.com/packages/2828/ Feb 16 14:20:15 hi guys...how do i install devkit8000 driver for my windowxp host virtualbox ubuntu? so I can use adb Feb 16 14:24:01 googe usb_android.ini Feb 16 14:32:27 av500: is googe than new clone of google? Feb 16 14:32:55 pr0n Feb 16 14:32:59 hehe Feb 16 14:33:02 try and find out Feb 16 14:33:05 www.googe.com Feb 16 14:35:06 guest___: sorry, usb_adb.ini Feb 16 14:35:38 av500 thanks...looking into it now.. Feb 16 14:36:35 so the ini file...how do i use it? i need to install it in the xp 1st ? point the device to the .ini files? Feb 16 14:36:57 no Feb 16 14:37:02 you need to google for it and read Feb 16 14:57:50 koen: looks like i'll be at EW Feb 29 Feb 16 16:10:29 LetoThe2nd: sweet! Feb 16 16:15:16 ok thanks...got my adb running! **** ENDING LOGGING AT Thu Feb 16 16:19:53 2012 **** BEGIN LOGGING AT Thu Feb 16 16:20:20 2012 **** ENDING LOGGING AT Thu Feb 16 16:24:35 2012 **** BEGIN LOGGING AT Thu Feb 16 16:26:17 2012 Feb 16 16:31:21 https://lh6.googleusercontent.com/-2JGcDu9vItc/Tz0vJhp6AGI/AAAAAAAAIqg/XgPOnhcoMp8/w402/12%2B-%2B1 Feb 16 16:31:27 I got my bone case :) Feb 16 16:50:52 Hmm can you open /dev/kmem, do an mmap() to the address if an omap peripheral (in global address space), and read/write said peripheral directly from userspace? Feb 16 16:50:59 assuming you had file permissions, of course Feb 16 16:51:33 or maybe it's /dev/mem not kmem Feb 16 16:58:51 SilicaGel, I think devmem2 is the tool of choice for that type of fiddling Feb 16 16:59:00 that's a library? Feb 16 16:59:30 command line tool Feb 16 16:59:38 ok Feb 16 16:59:45 so i just asked this on the google group but Feb 16 16:59:48 I'm asking here anyway. Feb 16 16:59:55 Why couldn't I skip the PRU loader entirely, Feb 16 17:00:22 and just use this devmem2 program to push data into the L4 address space of the IMEM for a particular PRU? Feb 16 17:01:11 SilicaGel: well, why use an OS at all? :) Feb 16 17:01:14 of course you can Feb 16 17:01:28 Good question. I'm uninstalling linux and replacing it with main() right now. Feb 16 17:01:31 :) Feb 16 17:01:42 apt-get main Feb 16 17:01:45 hehe Feb 16 17:02:16 I am going to do some experimenting this way just for giggles Feb 16 17:02:34 I don't intend to do this long term, but the uio_pruss stuff on the bone just does NOT seem to be ready for prime time Feb 16 17:07:09 SilicaGel: I use /dev/mem to talk directly with i2c omap controller Feb 16 17:07:33 with mmap() ? Feb 16 17:08:30 right Feb 16 17:08:43 siegen: http://tldp.org/LDP/khg/HyperNews/get/devices/fake.html do you really have to do that malloc() first? Feb 16 17:09:00 so you malloc() the memory, then you map it somehow? And taht doesn't cause some kind of leak? Feb 16 17:10:53 SilicaGel: i get filedesc opening /dev/mem and then mmap passing that fd Feb 16 17:11:07 but your first argument to mmap is NULL right? Feb 16 17:11:08 returns me the pointer to that memory Feb 16 17:11:15 right Feb 16 17:11:25 ok, then i'm going to ignore that svgalib example Feb 16 17:11:41 it really works Feb 16 17:11:43 :) Feb 16 17:11:54 yeah. Feb 16 17:12:09 I think it might be a really quick/dirty way to help me debug what's going on with the PRU Feb 16 17:12:39 someone from T.I. sent me a ton of patches for pru clock configuration, it's crazy stuff Feb 16 17:12:58 in my case i am using for better performance to talk with some i2c peripherials Feb 16 17:13:17 avoiding some more ioctls Feb 16 17:13:24 one of the things the code refers to is something called ICSS, I don't even know what that is Feb 16 17:13:34 - .main_clk = "icss_fck", Feb 16 17:13:34 + .main_clk = "pruss", Feb 16 17:13:51 I wonder if ICSS_FCK is something on an omap or something, i.e. not sitara Feb 16 17:14:24 dont think its omap Feb 16 17:16:10 morning Feb 16 17:16:22 * mranostay curses CA-85 Feb 16 17:16:36 heheh Feb 16 17:16:44 85 does indeed suck Feb 16 17:20:00 Hi, I'm having trouble changing the resolution on Angstrom. I'm using the following link: http://elinux.org/BeagleBoardFAQ#Display_resolutions_.231 Feb 16 17:20:31 I believe I'm on build 2.6.32, so I should be using the 3rd section but it is not working. Feb 16 17:24:59 I can't find "ICSS" in the sitara TRM, so I guess I'm ignoring it. Feb 16 17:25:24 those naming conventions aren't very good btw… the omap35xx is also called "sitara" Feb 16 17:25:30 oops Feb 16 17:25:41 what's the best way to refer to it then, am335x ? Feb 16 17:25:49 i think that's least ambiguous Feb 16 17:25:58 shit, I probably made the same mistake on the mailing list, then :( Feb 16 17:26:08 i think it's TI's fault Feb 16 17:26:52 AM33XX_IRQ_ICSS0_1 ahh HAH what is THAT Feb 16 17:27:37 I wonder if it's something like Interrupt Controller Subsystem Feb 16 17:28:07 they usually call interrupt controller INTC Feb 16 17:29:01 So, when you unmount the BeagleBone USB storage, the USB-to-Ethernet interface should automatically start, says the README. Am I supposed to see a new link (in ip link) on the host machine? The host machine is linux. Feb 16 17:29:06 you're trying to understand the clock tree for pruss? Feb 16 17:29:35 some dude named amit shah at t.i. sent me some patches, i'm tyring to figure them out beacuse they don't line up perfectly with my kernel Feb 16 17:29:53 they seem to do a few things, ike add a clocks to the platform so you can get them with clk_reg() Feb 16 17:29:58 which is good i guess Feb 16 17:30:05 but they REMOVE one called icss and I don't know waht taht is Feb 16 17:30:12 static struct omap_hwmod am33xx_icss_hwmod = { ... } Feb 16 17:30:15 sounds scary doesn't it? Feb 16 17:31:00 "uio_pruss driver was unable to load due to gpmc driver bug wherein the gpmc_irq for AM33XX was being set to INT_34XX_GPMC_IRQ." Feb 16 17:31:08 I'm not sure what the hell that means, either Feb 16 17:31:44 If you don't want to hear it from me any more it's ok to tell me to shut up Feb 16 17:32:38 * LetoThe2nd inserts an old aerosmith record and cranks up "shut up and dance" to full volume :) Feb 16 17:32:52 .oO( Thank God it wasn't Dude Looks Like a Lady ) Feb 16 17:36:47 gpmc is for talking to flash memory usually Feb 16 17:37:11 oh yeah i see that ... that's pretty nice!!! Feb 16 17:37:19 is there a separate channel for the beaglbone? Feb 16 17:37:22 I don't know how taht impacts this though Feb 16 17:37:28 yeah there is, but it's really really quiet Feb 16 17:37:41 any advice for the Angstrom resolution? ^^ Feb 16 17:37:57 The PRU's memory, it is static ram isn't it? the DMEMs and shared memory? Feb 16 17:38:36 quantumi: not that I know of Feb 16 17:38:52 and it would be the same situation as for the mailing list Feb 16 17:39:02 yeah, there's a #beaglebone on freenode. Like I said, though, it's super quiet. Feb 16 17:39:16 for the same reason that the bone ml is queit Feb 16 17:39:18 quiet Feb 16 17:39:28 it's probably sram if you don't have to do anything to make it work Feb 16 17:42:09 #define AM33XX_PRUSS_SHAREDRAM_BASE 0x4a310000 Feb 16 17:42:10 YAY Feb 16 17:42:37 ICSS doesn't seem to be documented in the TRM, is there a register address you can cross reference? Feb 16 17:42:46 let me see Feb 16 17:43:07 perhaps its related to that undocumented M3 power control stiff Feb 16 17:44:21 static struct omap_hwmod am33xx_pruss_hwmod = { .clkdm_name = "icss_ocp_clkdm", } Feb 16 17:44:24 i found that snippet Feb 16 17:45:54 maybe it's an old name for something Feb 16 17:46:12 * This file is automatically generated from the AM33XX hardware databases. Feb 16 17:46:16 #define AM33XX_DMA_ICSS0_7 0 Feb 16 17:46:50 mach-omap2/omap_hwmod_33xx_data.c:/* PRUSS/ICSS */ Feb 16 17:46:56 could it be ... an old name for the pruss? Feb 16 17:47:46 Webinar topics include Industrial Communications (ICSS PRU, robust analog interface devices), Motor Control (analog motor control, embedded processors for motor control), PLC I/O Modules, Human-Machine Interfaces, and Industrial Sensors. Register here. Feb 16 17:47:48 AhhHAH Feb 16 17:48:58 yeah sounds like they changed the name Feb 16 17:50:39 would the beaglebone or beagleboard be better for openCV? Feb 16 17:50:53 depends Feb 16 17:51:16 ok are you ready for this jay: Feb 16 17:51:16 >This is the OCMC-RAM in TRM - 64KB outside PRUSS. IMEM is the PRU code space - these are mapped to load the firmware to PRUs from ARM application. DMEM is data memory with low latency access from PRUs -again used for PRU internal processing and sharing data with ARM application. Feb 16 17:51:24 >In the previous generation of PRUSS - DMEM within PRUSS was very limited (512 bytes each) so it did make sense to use OCMC-RAM outside PRUSS. But here PRUSS DMEM is quite big and for most applications this shall be good. Feb 16 17:51:28 what the hell is he talking about Feb 16 17:51:34 i'm looking for eventual integration into a product, and they claim that the beaglebone is easier to replicate since it doesn't use POP memory, but i have my doubts about that Feb 16 17:51:44 seems like routing DDR signals is more challenging to me Feb 16 17:51:49 you know Feb 16 17:51:56 i did some work with this freescale dsp Feb 16 17:51:58 and i thought the same thing Feb 16 17:52:15 jmole: pop is easy these days Feb 16 17:52:24 av500: that's what I'm saying Feb 16 17:52:27 if you can solder .4mm bga, you can add pop Feb 16 17:52:27 but the pinout of that particular dsp aligned with standard sdram chips in a way that I was able to route the whole data and address bus on a single layer with no wire crossing Feb 16 17:52:32 jmole, the actually memory is apparently hard to get Feb 16 17:52:47 jmole, the components on the beaglebone should be easier to source Feb 16 17:52:49 Russ: you sure? Feb 16 17:52:58 They shoudl be easier to source *eventually* Feb 16 17:53:03 eh? Feb 16 17:53:05 that's what jkridner said Feb 16 17:53:14 ddr2 vs lpddr Feb 16 17:53:14 thats what i also heard, the sitara stuff should be longterm catalog Feb 16 17:53:17 yes, but that is cargo cult Feb 16 17:53:19 SilicaGel: they're talking about the OCM (on chip memory) SRAM the SoC uses for bootstrapping Feb 16 17:53:19 If you actualyl go looking though, the AM335x still isn't available from most places Feb 16 17:53:22 and the PMIC = awful Feb 16 17:53:27 no problems getting them Feb 16 17:53:32 I was originally looking at tegra before beagle, and you can't get those unless you're in volumes of 100k+ Feb 16 17:53:32 The PMIC is listed on a few places but at qty.2500 Feb 16 17:54:02 So I *hope* the PMIC + AM335x will be somethign I can really buy from avnet, digikey, whatever, but I'm a little concerned. Feb 16 17:54:13 Well, that's just a ramping issue Feb 16 17:54:15 nvidia is too secretive anyways Feb 16 17:54:16 SilicaGel: it is supposed to be orderable. I'll go bug the marketing folks to get the right ordering information. Feb 16 17:54:18 jay6981: after bootstrapping, is it not used? Feb 16 17:54:31 maybe i'll just buy both the bone and the board and figure out which one works best Feb 16 17:54:32 jay6981: exactly Feb 16 17:54:38 You'll be able to source them eventually :) Feb 16 17:54:42 SilicaGel: it's still available for use… not sure if anything actually uses it Feb 16 17:54:43 jkridner: it has been really tough even getting information on the PMIC. Mid month I think they finally posted the full data sheet Feb 16 17:54:50 guys like nvidia and QC dont want you anyway Feb 16 17:55:11 TI will want you if you promise to buy many level shifters Feb 16 17:55:22 I do Feb 16 17:55:26 the fancy kind of level shifters, too Feb 16 17:55:28 a500, no different than the OMAP guys Feb 16 17:55:28 the kind that work on MMC Feb 16 17:55:59 mdp: maybe the MDBU will change that :) Feb 16 17:56:23 what the speed difference between the bone and the xm? Feb 16 17:56:23 hi, anyone experience with ubuntu on beaglebone? it boots for me, I get a console, see the boot log, but can't type at the login prompt, anyone an idea ? Feb 16 17:56:34 effem ask ubuntu Feb 16 17:56:43 SilicaGel: they're saying the old PRU had to use memory outside the PRUSS because the DMEMs were too small Feb 16 17:56:54 oh i see Feb 16 17:57:10 jay6981: you're good at figuring things out. and helping me figure things out. thanks. Feb 16 17:57:16 no prob Feb 16 17:57:16 woglinde: I would expect this to be specific to the beaglebbone as I used RCN's preconfigured image Feb 16 17:57:31 eFfeM: if you stick to http://elinux.org/BeagleBoardUbuntu#Demo_Image and actually read what it says, not just copypasta things, it should work fine :) Feb 16 17:57:38 hm didnt see rcn for a while Feb 16 17:57:41 * SilicaGel Do Not Eat Feb 16 17:57:50 food Feb 16 17:57:53 LetoThe2nd: rcn did a bone image? Feb 16 17:58:13 eFfeM: where is my picture frame btw? Feb 16 17:58:19 its like 3ys now!!!!!!!! Feb 16 17:58:28 av500: rcn did an image including setup script that supports the whould beagle/panda/bone shebang Feb 16 17:58:29 SilicaGel: regarding those patches... which kernel are you currently using? Feb 16 17:58:51 LetoThe2nd: ah Feb 16 17:58:53 av500: activity stopped (actually lost the job at the company I developed part of it) Feb 16 17:58:58 Currently i-sdk-am335x-evm-05.03.01.00/board-support/linux-3.1.0-psp04.06.00.03.sdk but I've also bitbaked an angstrom 3.1 Feb 16 17:59:04 There are reasons to use both, which is really unfortunate. Feb 16 17:59:05 av500: check the link, maybe scroll down a bit. Feb 16 17:59:12 av500: I'm now working on amahi (www.amahi.org) was trying to see if I could get it on the bone too Feb 16 17:59:13 The DCAN stuff AFAIK still isn't in T.I. linuxsdk Feb 16 17:59:14 LetoThe2nd: I dont check links Feb 16 17:59:20 It's in angstrom, though people tell me it has problems Feb 16 17:59:24 SilicaGel: 'cos I've seen similar patches for the d_can and they applied to 3.2 iirc Feb 16 17:59:34 av500: hrhr Feb 16 17:59:34 oh? Feb 16 17:59:40 eFfeM: ah, mahi Feb 16 17:59:44 but maybe it was 3.1 and I was using 3.0 Feb 16 17:59:49 homse server Feb 16 17:59:49 anyway they touched the clock settings Feb 16 17:59:50 oh my Feb 16 17:59:52 dwery: do said patches have to do with routing clocks around, mostly? Feb 16 17:59:53 ahhh Feb 16 17:59:57 that's useful information Feb 16 18:00:05 clock names and similar things Feb 16 18:00:11 d_can was buggy as hell :D Feb 16 18:00:20 no chances it would have worked before Feb 16 18:00:29 when i talked to folks on the t.i. forums about DCAN a while back (more than a month) they were telling me look at angstrom, git clone off the trunk and you'll be further ahead than our stuff Feb 16 18:02:38 ahead with bugs :D Feb 16 18:02:39 the DCAN is on my list of stuff to do, but it can wait probably another month, so I was really hoping to defer worrying about it and see if the drivers matured Feb 16 18:02:39 they made nasty things with a string buffer, like overwriting it Feb 16 18:02:39 this of course all means kernel gyrations for me, and the more friggin drivers/modules I have to patch, recompile, etc. the worse life will be Feb 16 18:02:39 oh geeze, so you're talking general code quality issues Feb 16 18:02:39 yes. but it's fixed now Feb 16 18:02:39 did anyone get flogged? Feb 16 18:02:45 someone did not understand proper use of memory allocation vs stack Feb 16 18:03:14 I guess they have different dewvelopes on that Feb 16 18:03:15 .o(I bet the india team) Feb 16 18:03:27 developers* Feb 16 18:03:45 dwery: well, if there's one thing we learned from star wars, it's that such failures should be met with an invisible choking Feb 16 18:03:53 :D Feb 16 18:04:10 corporal punishment shoudl be allowed in such circumstances Feb 16 18:04:53 the rest of the code seems fine Feb 16 18:04:59 I'll review it when I get some spare time Feb 16 18:05:17 It's a PF_CAN socket interface from the userspace side right Feb 16 18:05:20 I hate that interface Feb 16 18:05:27 SocketCAN. works nicely Feb 16 18:05:31 does it? Feb 16 18:05:34 I hate it :( Feb 16 18:05:44 any particular reason? Feb 16 18:05:54 you'll make fun of me :( Feb 16 18:06:22 I'll be good :D Feb 16 18:06:36 I made a PC/104 dual CAN + J1708 board out of an FPGA and wrote a kernel driver for it, but I rpesented a character-mode device to userspace Feb 16 18:06:58 whoa! that's a great piece of work! Feb 16 18:07:01 Yeah. Well ... my complaint is that all our health management software is written in java, and you can't access PF_CAN sockets Feb 16 18:07:05 from java. Feb 16 18:07:13 uhm.. strange you can't Feb 16 18:07:21 it's a socket anyway Feb 16 18:07:23 yeah java only understands TCP sockets and UDP sockets Feb 16 18:07:30 o.O Feb 16 18:07:31 it doesn't have the ability to open some other kind of socket Feb 16 18:07:32 wrong Feb 16 18:07:36 unixsocket Feb 16 18:07:37 uh oh Feb 16 18:07:39 SilicaGel: raw sockets? Feb 16 18:07:45 from matthew Feb 16 18:07:49 libmatthew yeah Feb 16 18:07:51 you need that fot dbus Feb 16 18:08:00 i'm familiar with that. Feb 16 18:08:06 I meant to say it can't access it without JNI (or JNA) Feb 16 18:08:14 you can easily bridge can over udp Feb 16 18:08:23 yeah, I could do that Feb 16 18:08:44 Ultimately I WILL do something like that, because the whole world has decided that sockets are the API to interface to CAN Feb 16 18:08:45 java can't do raw sockets or even ICMP Feb 16 18:08:48 LetoThe2nd: went over the page again, it is definitely not responding to omap over the network, I have console output but no input (this means it *is* the proper tty), network does not get an ip addr. no idea where I went wrong Feb 16 18:09:06 and, while I don't agree with that decision, complying with it probably gives me greater flexibility in the future. Feb 16 18:09:10 SilicaGel: did you pay royalties to Mr Bosch too? :D Feb 16 18:09:22 for when I want to use an off the shelf PEAK adapter, N.I. adapter, or the like. Feb 16 18:09:37 oh well, I hate gnome 3 but I guess I'll have to live with it.. Feb 16 18:09:42 eFfeM: no idea right now, i used that just a few hours ago and it worked like a charm :) Feb 16 18:09:54 ME TOO Feb 16 18:10:05 I am going to miss Kubuntu very much :((( Feb 16 18:10:09 ah ok, good Feb 16 18:10:10 effem pull the sd and set network manually Feb 16 18:10:14 SilicaGel: which FPGA did you use? Feb 16 18:10:17 bbiab Feb 16 18:10:24 a Spartan Feb 16 18:10:33 Oh that's why you asked about royalties hehe Feb 16 18:10:36 * SilicaGel whistles Feb 16 18:10:47 some says the patent has expired Feb 16 18:10:59 but I'm not willing to check.. Feb 16 18:11:05 woglinde: yeah just decided to so do Feb 16 18:11:31 yeah Feb 16 18:11:44 we actually took tha tboard to production but replaced the all fpga one with a couple of SJA1000s Feb 16 18:11:54 to free up logic space for doing a better job with the J1708 decoding Feb 16 18:11:58 I'd love to have a 3.3v version of the sja Feb 16 18:13:21 dwery: all I use around here for programmable logic is xilinx, because the tools are cheap and in many cases free. I've done a little bit of stuff with alterra but their software is pretty terrible Feb 16 18:13:31 agreed. Feb 16 18:13:38 I've come to like modelsim though Feb 16 18:13:57 I used modelsim a lot before I knew about VHDL test benches Feb 16 18:14:18 I used TCL to model real world I/O :-) Feb 16 18:14:22 here let me show you an example!!! Feb 16 18:14:41 :D Feb 16 18:14:57 for me was fun to programm with mitrion-c Feb 16 18:14:58 http://www.autofrog.com/~chrisp/Projects/Electronics/NixieClock/Sim.jpg Feb 16 18:15:12 That's modelsim running a nixie tube clock I built out of a xilinx CPLD Feb 16 18:15:18 ouch! Feb 16 18:15:26 the outputs of the CPLD drove the elements of the nixie tubes Feb 16 18:15:44 nice stuff Feb 16 18:15:57 so I wrote something in modelsim that would let me draw pictures of the actual tubes, and those buttons are fakning the real physical buttons to set the clock Feb 16 18:16:19 It ran something somewhat close to 'real time' but with modelsim you can of course probe even internal signals. It was fun Feb 16 18:16:29 I bet it took more than writing the VHDL Feb 16 18:16:40 actually i twasn't bad! Feb 16 18:17:04 modelsim has (had?) a tcl interface, and you could directly see cpld signals (pins as well as internal) as if they were tcl variables Feb 16 18:17:06 by their names Feb 16 18:17:15 so like my signal dsp_hrs_l I could refer to from tcl Feb 16 18:17:30 then based on those, it was just a big case stement to decide which .jpg to display in which of the six positions Feb 16 18:17:34 never got to grasp tcl Feb 16 18:17:37 TCL is pretty awful. Feb 16 18:17:53 Back in the day though, a ton of EDA tools used tcl as a built in scripting language Feb 16 18:18:22 they never got to grasp PERL :D Feb 16 18:18:44 i guess Feb 16 18:18:47 I'm captain java now though Feb 16 18:18:57 so my dream is that everything has groovy built into it Feb 16 18:19:02 :heart: groovy Feb 16 18:20:28 I've got this truck simulator I built ... let me show you: https://picasaweb.google.com/108380917869049424169/ImagineRIT2011?authkey=Gv1sRgCMK0873ii5aO8wE#5604692174577057986 Feb 16 18:21:02 It has groovy built into it as a scripting language to react to events. So it simulates the truck's CAN databus, and it spits out things like coolant temp, engine rpm, wheel speed, oil pressure ... Feb 16 18:21:09 !! Feb 16 18:21:12 what Feb 16 18:21:20 Don't ask me who that girl is driving, I have no idea Feb 16 18:21:31 oh man holy crap yes I do, it's my wife Feb 16 18:21:32 hahaha Feb 16 18:21:34 <-- idiot Feb 16 18:21:35 :D Feb 16 18:22:15 so you do that for a living? Feb 16 18:22:40 That simulator has an engine and weight model built into it that's accurate, I developed it as a training tool to demonstrate to drivers how they can drive the same course and get 6 mpg, or 3 mpg, just based on how they accelerate and shift gears. Feb 16 18:22:52 and they cared? Feb 16 18:22:59 Oh we make them care :) Feb 16 18:23:02 haha Feb 16 18:23:14 actually I work for a university, I'm a research engineer. I work on a lot of different things. Feb 16 18:23:27 nice Feb 16 18:23:29 The most consistent of those things over the past 10 years has been vehicle diagnostic systems Feb 16 18:23:41 which CURRENTLY are mostly x86, Intel atoms and crap Feb 16 18:23:52 I'm on a one man crusade to push them all to arm Feb 16 18:23:55 I know of a bone with CAN bus :D Feb 16 18:24:05 I keep telling everyone, you need to think like a cell phone, not like a small personal computer Feb 16 18:24:19 silicagel where is jwinnipeg? Feb 16 18:24:30 he's in Tampa on vacation! Feb 16 18:24:33 He'll be back on Monday Feb 16 18:24:41 ah Feb 16 18:24:48 any decision made? Feb 16 18:24:55 for java? Feb 16 18:25:33 Not really. We're doing a lot of performance testing right now with the oracle jre 6 headless vfp Feb 16 18:25:42 the performance is spectacular Feb 16 18:25:47 on fosdem I attend the openjdk arm-asm talk from andrew haily Feb 16 18:25:55 but we're still hoping to *not* use that Feb 16 18:26:10 Oh yeah? What are your feelings on its readiness for prime time? Feb 16 18:26:31 just use android, it has fast java :) Feb 16 18:26:42 fast non-java :) Feb 16 18:26:51 hm some features missing like safepoints and voilatile Feb 16 18:27:02 which ed didnt care about or didnt had the time Feb 16 18:27:09 woglinde: do you want to see what jwinnebeck looks like? https://picasaweb.google.com/lh/photo/MaoeouaPCl4HyO6gmdsWtc-wfef3cGFziAjyAyijx1o?feat=directlink Feb 16 18:27:10 but they will be adressed Feb 16 18:27:48 * woglinde fights with the fucking not threadsafeness of java webservicestack Feb 16 18:27:48 That's him driving our truck simulator in my lab. He actually did 99% of the work on that truck simulator, he knows about 3D graphics and OpenGL and triangles and all that mumbo jumbo. Feb 16 18:28:14 that lab is too clean Feb 16 18:28:21 its a sim Feb 16 18:28:23 oh no it's not, the dirty parts are behind you Feb 16 18:28:25 yep Feb 16 18:28:39 SilicaGel: how do you know what my lab looks like? :) Feb 16 18:28:43 haha Feb 16 18:28:52 are those supposed to be xeyes? Feb 16 18:28:56 Hahahaha Feb 16 18:28:58 yes hahahahahahahahahaha Feb 16 18:29:09 hahahahahahahahahhahahahahahahahahahhahahahaahhahahahhaaha Feb 16 18:29:13 you probably can't see it, but if you zoomed in, there's actually an X "close" box to the upper right of the eyes Feb 16 18:29:25 nice touch Feb 16 18:29:29 haha Feb 16 18:29:33 that's my office btw, with the xeyes Feb 16 18:30:04 they told me I could live in the cube farm, or I could build my own cube in my lab, so I did the latter Feb 16 18:30:49 hahaha i am so happy right now, because you recognized them as xeyes. Most people around here think I'm just plain insane. They have no dea why I put eyeballs on my wall. Feb 16 18:31:18 yes, true: http://i.imgur.com/NqqS8.png Feb 16 18:31:28 HAHAHA Feb 16 18:31:41 damn, I've been so brainwashed, I was looking at the upper left Feb 16 18:31:45 damn, I forgot to apply the CSI filter... Feb 16 18:32:14 yeah, those csi filters can make something out of "nothing" Feb 16 18:32:39 in that picture, that little screen on the right is what gets installed in trucks Feb 16 18:32:49 it gives the drivers feedback and coaches them to drive better to conserve fuel Feb 16 18:33:16 Some day that will be driven by a little arm that will look suspiciously like the beaglebone reference design :D Feb 16 18:33:33 the driver will be replaced by an arm? Feb 16 18:33:55 yes, we don't need him to have legs or a torso once we move the pedals up next to the steering wheel Feb 16 18:33:56 boo perl :) Feb 16 18:35:47 woglinde: hardcoding an interface works like a charm, guess my usb is somewhat flakey, this time it stopped halfway the bootlog (butit got up completely) Feb 16 18:36:22 effem usb is a mess on bone Feb 16 18:36:33 ahok Feb 16 18:36:35 ah ok Feb 16 18:36:58 s/on bone// Feb 16 18:38:26 er musb is on the bone? Feb 16 18:39:04 What's the deal with USB disconnets being a total failure on the bone (usb port never works again) - has that been fixed? Last time I asked someone said (with some snark) "The smartest TI people india are working on that" Feb 16 18:39:53 SilicaGel: same status Feb 16 18:40:20 SilicaGel jkridner may know Feb 16 18:49:48 I fixed that! Feb 16 18:49:54 http://gitorious.org/clonetty Feb 16 18:50:51 this problem we're seeing isn't just tty devices though, it's anything. When you plug in a device once it enumerates just fine. When you pull it out, and plug it back in, it refuses to enumerate ever again Feb 16 18:51:05 It happens with tty type devices, but it also happens with things like our usb accelerometers and PEAK can adapters Feb 16 18:52:02 Russ: that does not help me. on my dev PC I lose about one ttyUSBx per day Feb 16 18:52:13 for ever since the usb bus reset and that tty stays allocated or so Feb 16 18:52:19 av500, I remember that Feb 16 18:52:22 Russ: what you fix? Feb 16 18:52:32 av500, you need a fixed kernel Feb 16 18:52:50 Russ: I guess so, but I refuse to upgrade my 2008 linux install :) Feb 16 18:53:11 mranostay, the problem where /dev/ttyUSB0 goes away everytime you reset or power cycle your board Feb 16 18:53:23 yeah udev rules? Feb 16 18:54:00 no, it clones /dev/ttyUSB0 to a pty and reopens it as necessary Feb 16 18:54:07 maybe I should merge it with ttywatch Feb 16 18:55:42 * mranostay check that git repo Feb 16 18:55:46 *cheks Feb 16 18:55:49 gah checks Feb 16 19:20:20 maybe the USB problem hasn't been fixed because they're busy working on this: http://www.dellchallenge.org/projects/humanure-power-project Feb 16 19:21:28 I'm sorry. That project is probably someone's life's work, and there I went and made fun of it. Feb 16 19:21:57