| 00:20.37 | *** join/#edev RobotGuy (n=RobotGuy@dsl093-038-120.pdx1.dsl.speakeasy.net) |
| 00:30.39 | cajaka | Evening Robotguy |
| 00:35.10 | *** join/#edev roxfan (n=dunno@252.223-64-87.adsl-dyn.isp.belgacom.be) |
| 00:39.22 | *** join/#edev redblue (i=star@ppp081.109-253-207.mtl.mt.videotron.ca) |
| 00:40.14 | *** join/#edev prpplague (n=dave@70.249.150.251) |
| 00:41.32 | prpplague | ho ho ho |
| 00:41.37 | prpplague | mery freakin monday |
| 00:42.00 | cajaka | Long day at work? |
| 00:42.05 | prpplague | yea |
| 00:43.04 | RobotGuy | Howdy |
| 00:43.11 | prpplague | RobotGuy: greetings |
| 00:43.23 | RobotGuy | prpplague, I just got home a bit ago. |
| 00:44.18 | RobotGuy | I can't believe how many people can't seem to see a blue wheel chair with TWO white flag poles and bright orange flags. |
| 00:44.31 | prpplague | RobotGuy: back from seeing your parol officer? |
| 00:44.55 | RobotGuy | prpplague, Hee hee, ummm, no.. Doctor. :P |
| 00:45.19 | prpplague | RobotGuy: two words: ROAD SPIKES |
| 00:45.28 | RobotGuy | prpplague, Who spiked your oatmeal today? ;) |
| 00:45.58 | prpplague | RobotGuy: hehe, just tired, i get silly when i'm really tired |
| 00:46.23 | CosmicPenguin | he is tired alot |
| 00:46.24 | RobotGuy | prpplague, People tell me I am always silly, so they can't tell. ;) |
| 00:47.23 | prpplague | CosmicPenguin: hehe |
| 00:51.39 | *** join/#edev jerry_smif (i=jerrysmi@166.166.112.221) |
| 00:51.40 | RobotGuy | Dang, I got another e-mail from Robert Sheer, but I can't find it now. |
| 00:51.59 | *** part/#edev jerry_smif (i=jerrysmi@166.166.112.221) |
| 00:52.09 | SpeedEvil | Thunderbird has recently decided for me that the 'all' view in my inbox means 'none'. |
| 00:52.19 | SpeedEvil | Mail clients are annoying. |
| 00:53.46 | RobotGuy | I am using KMail. |
| 00:54.04 | RobotGuy | I didn't care for the way Thunderbird quotes messages. |
| 00:54.56 | prpplague | RobotGuy: CosmicPenguin is the lead linux guy at AMD |
| 00:55.07 | RobotGuy | I want to tinker with Hammer. |
| 00:55.08 | CosmicPenguin | thats not quite accurate |
| 00:55.17 | CosmicPenguin | I'm an embedded linux developer |
| 00:55.20 | CosmicPenguin | I don't lead anything |
| 00:55.54 | RobotGuy | My main PC is AMD (older Athlon XP 2800+). |
| 00:56.05 | *** join/#edev prpplague (n=dave@70.249.150.251) |
| 00:56.26 | prpplague | RobotGuy: CosmicPenguin is the lead linux guy at AMD |
| 00:57.38 | SpeedEvil | I have a processor that sort-of-works under 'athcool' to give some 5-10W idle. But sound and network don't really work. |
| 00:57.52 | RobotGuy | Private Messaging? Python Module? |
| 00:58.19 | RobotGuy | CosmicPenguin, what don't you lead at AMD? |
| 00:58.31 | SpeedEvil | RobotGuy: Power Managment. |
| 00:58.34 | prpplague | power managment |
| 00:58.47 | RobotGuy | Durn, I knew I was forgetting something. |
| 00:59.39 | prpplague | RobotGuy: he don't harras him too bad, he does a pretty darn good job at amd |
| 00:59.58 | RobotGuy | I have two socket A boards I need to build systems around. I just need a case and power supply for one of them. |
| 01:00.14 | prpplague | RobotGuy: CosmicPenguin did a giant amount of work on the olpc |
| 01:00.17 | *** join/#edev igorsk2 (n=dunno@80.200.2.144) |
| 01:00.19 | RobotGuy | prpplague, Naw, I was just asking, because he said he doesn't lead anything. |
| 01:00.39 | prpplague | RobotGuy: ahh |
| 01:00.57 | RobotGuy | If thre is Linux in his title, he's top draw as far as I am concerned. :) |
| 01:01.39 | prpplague | hehe |
| 01:01.39 | RobotGuy | The only reason one of my PCs is not AMD is because I got the system board in a trade. |
| 01:04.10 | RobotGuy | Maybe someday when I grow up. |
| 01:08.21 | ds2 | just do it. |
| 01:08.38 | ds2 | no need to aspire. parts are all over, for any budget |
| 01:08.47 | RobotGuy | I'm not sure I want to grow up. |
| 01:09.14 | RobotGuy | Heh heh, I'm almost 50. |
| 01:09.49 | RobotGuy | ds2, This is why I have a Hammer.. :) |
| 01:11.27 | ds2 | heh... you can get boards for <$20 if needed...hammer is nice and documented and there is someone to pester about it ;) |
| 01:12.14 | RobotGuy | ds2, And a box full of PICs, and another box of assorted MCUs and support chips, including Freescale i.MX21 i.MX27, i.MX31, TI TMS320 series and TMS470 series, Luminary Micro Cortex (some really nice stuff), Analog Devices (I love AD stuff), etc. |
| 01:12.36 | prpplague | oh you have one of those luminary's? |
| 01:12.44 | RobotGuy | Or, and can't forget lots of Dallas-Maxim stuff too. :) |
| 01:13.00 | ds2 | AD is very nice indeed |
| 01:13.03 | prpplague | i;ve been trying to find out what lcd they are using |
| 01:13.04 | RobotGuy | prpplague, Which one you mean? |
| 01:13.26 | prpplague | the little 96x48 oled one |
| 01:13.30 | ds2 | jush wish AD would make a bigger pure arm device instead |
| 01:13.36 | ds2 | just |
| 01:13.42 | RobotGuy | I also have FOUR Renesas starter kits - two H8/36077, and two H8SX/1664 |
| 01:14.05 | RobotGuy | prpplague, You mean on their dev boards? |
| 01:14.18 | prpplague | yea |
| 01:14.40 | RobotGuy | prpplague, I just have a few chips, no dev kits from Luminary Micro yet |
| 01:14.50 | prpplague | ahh |
| 01:15.12 | prpplague | i'd love to use a small lcd like that one, but i can't seem to find one cheap |
| 01:15.17 | joe_bleau | I agree, that LM eval kit is pretty. |
| 01:15.26 | RobotGuy | I know which display Renesas uses. |
| 01:17.15 | RobotGuy | I would like to have one of these: http://www.luminarymicro.com/products/lm3s6965.html |
| 01:17.39 | ds2 | prpplague: what about the sparkfun OLED? |
| 01:17.55 | prpplague | too big |
| 01:18.29 | *** mode/#edev [+o prpplague] by ChanServ |
| 01:18.34 | RobotGuy | This is the dev kit I would like: http://www.luminarymicro.com/products/lm3s6965_ethernet_evaluation_kit.html |
| 01:19.51 | prpplague | pretty good price? |
| 01:19.57 | prpplague | $70 isn't bad at all |
| 01:20.13 | RobotGuy | prpplague, For sure! |
| 01:20.40 | RobotGuy | I need to dig through my samples and see if I have any clock generator chips. |
| 01:21.06 | landley | prpplague: EE still giving you trouble? |
| 01:21.29 | RobotGuy | I was thinking a multi-media application processor would be good for vision on a robot. |
| 01:21.53 | landley | CosmicPenguin: is AMD actually doing anything interesting with Linux? |
| 01:22.01 | prpplague | landley: yea |
| 01:22.54 | prpplague | landley: i've been trying to get the project back on schedule, and he tried to take credit for it today |
| 01:22.56 | landley | RobotGuy: I didn't aspire, I just played. I was doing embedded linux for ~3 years before anybody _paid_ me to do so. :) |
| 01:23.15 | RobotGuy | I would love to have something like, oh, an i.MX32 slaved to my Hammer for computer vision. |
| 01:23.38 | RobotGuy | I mean i.MX31 |
| 01:23.45 | prpplague | .ping |
| 01:24.00 | RobotGuy | Or a TI DaVinci (also can run Linux). |
| 01:24.21 | RobotGuy | .pong |
| 01:24.47 | landley | What's iMX31? |
| 01:26.18 | RobotGuy | landley, http://www.freescale.com/webapp/sps/site/overview.jsp?nodeId=0162468rH3114329733642 |
| 01:27.35 | landley | RobotGuy: what a horribly designed web page. |
| 01:27.45 | landley | Ok, so it's an embedded graphics chip, I think. |
| 01:28.05 | RobotGuy | Unfortunately, the Freescale i.MX processors are all BGA. |
| 01:28.08 | landley | I wonder what kind of lcd you hook it to, and where you get that, and how much it would cost, and the packaging? |
| 01:28.12 | landley | BGA? |
| 01:28.19 | landley | Banana Gate Array? |
| 01:28.20 | RobotGuy | landley, It's more than that - it can run full Linux also. |
| 01:28.28 | RobotGuy | Ball Grid Array |
| 01:28.37 | landley | Ah, mounting thingy. |
| 01:28.38 | landley | Right. |
| 01:28.52 | landley | "We haven't got pins, we've got little _bumps_." |
| 01:29.06 | landley | This is one of those "arm processor glued to rabid weasel" SOC things? |
| 01:29.27 | landley | Ala Marvell 88W8388 network chip thingy? |
| 01:30.03 | ds2 | Mmmmm banana gate arrays |
| 01:30.11 | landley | If it's got an arm processor in it and it runs Linux, why would you want it _slaved_ do your hammer board? |
| 01:30.15 | ds2 | from GorilliaWorks (tm) ;) |
| 01:30.27 | landley | Wouldn't you want a hammer design around a different CPU instead, with the built-in graphics? |
| 01:30.48 | landley | prpplague: you need a blog. |
| 01:31.07 | prpplague | landley: hehe why's that? i rant too much here? |
| 01:31.38 | RobotGuy | I want my Hammer to be a traffic cop type controller sometimes, talking to other MCUs. |
| 01:31.58 | prpplague | RobotGuy: generally you want your most powerful cpu to be the cop |
| 01:32.18 | prpplague | RobotGuy: if you get alot of small devices like pics and avr's that'd make more sense |
| 01:32.18 | ds2 | ARM MP!!! |
| 01:32.37 | RobotGuy | prpplague, Generally, yes, but not necessarily if you have an application processor dedicated to a certain task such as computer vision. |
| 01:33.54 | prpplague | RobotGuy: indeed, i'm hope'n that the hammer works well enough to do some vision stuff |
| 01:34.19 | landley | Nah, just in general. |
| 01:34.28 | landley | They're therapeutic. Especially if you don't care about getting fired. :) |
| 01:34.28 | prpplague | RobotGuy: so you could use multiple hammer boards in a some sort of matrix |
| 01:34.58 | RobotGuy | prpplague, Well, the RoboMagellan folks think computer vision is mandatory for that competition. I don't necessarily agree, but I have never built that kind of robot either. |
| 01:35.31 | RobotGuy | OpenCV is a vision library for Linux. |
| 01:35.39 | RobotGuy | There are others also |
| 01:36.13 | RobotGuy | prpplague, Hmmmmm, Hammer Matrix, then there has to be a sequel.. |
| 01:37.06 | prpplague | RobotGuy: yea, i've already tried it on the hammer with a usb camera |
| 01:37.13 | RobotGuy | I'd like to see HammerVision |
| 01:37.22 | RobotGuy | prpplague, How does it work? |
| 01:38.05 | prpplague | i was able to track a pencil and identify some objects |
| 01:38.16 | RobotGuy | prpplague, Not bad. |
| 01:38.17 | prpplague | but i was just doing 320x240 images |
| 01:38.26 | SpeedEvil | hard to do much more on 1.1 |
| 01:38.27 | RobotGuy | Still not bad though. |
| 01:38.29 | prpplague | it was just a quicky test |
| 01:39.01 | prpplague | landley: i care about getting fired, hehe |
| 01:39.04 | RobotGuy | I really want to experiment with computer vision, but most folks sure make it sound difficult. |
| 01:39.32 | ds2 | hammer matrix is the answer to the gum wad? |
| 01:39.39 | SpeedEvil | Doing difficult tasks is hard. |
| 01:39.46 | prpplague | RobotGuy: imho thats what tends to make you an expert, tinkering with the things no one else seems to tackle |
| 01:39.53 | SpeedEvil | Doing stuff like 'track the brighest object inb the frame' isn't too bad |
| 01:40.10 | RobotGuy | Which camera did you use? I want my next webcam to be Linux friendly. |
| 01:40.23 | SpeedEvil | RobotGuy: look for 'driver free' on the box |
| 01:40.29 | prpplague | just some cheapies from geeks.com |
| 01:40.34 | prpplague | i think we made a wiki page |
| 01:40.39 | SpeedEvil | RobotGuy: this means that it has a good chance for a linux uvc driver |
| 01:40.53 | SpeedEvil | as all 'driver free' cameras are usb video class |
| 01:40.55 | RobotGuy | prpplague, I tinker a lot, and have done things people told me could not be done. They did not really like that much. |
| 01:41.22 | *** join/#edev cajaka (n=none@cpe-24-58-193-200.twcny.res.rr.com) |
| 01:41.22 | RobotGuy | SpeedEvil, Ah, OK, so if it doesn't need drivers then. |
| 01:41.40 | SpeedEvil | RobotGuy: The logitech optia AF is 2MP autofocus and outputs 1400*1200 at 10Hz under linux |
| 01:41.50 | SpeedEvil | and about $100 |
| 01:42.07 | landley | prpplague: that would take some of the fun out of it, yes. |
| 01:42.35 | RobotGuy | SpeedEvil, Sounds nice. It's within my price range. |
| 01:42.38 | prpplague | landley: hehe |
| 01:42.55 | landley | How goes the nail board? |
| 01:43.11 | prpplague | landley: i think the boss is gonna get the picture soon, but i think its gonna impact the project pretty heavily in the long run |
| 01:43.15 | SpeedEvil | RobotGuy: that's USB2 of course - you won't get anywhere near that frame rate at usb1.1 |
| 01:44.12 | landley | prpplague: the EE, you mean? |
| 01:44.20 | RobotGuy | SpeedEvil, I figured that. I would not need any where near that resolution though. |
| 01:44.21 | landley | (Context?) |
| 01:44.35 | RobotGuy | prpplague, Which picture? |
| 01:44.43 | prpplague | http://www.geeks.com/details.asp?invtid=C-099A&cpc=SCH |
| 01:44.55 | prpplague | landley: yea |
| 01:45.14 | prpplague | landley: the boss will get the picture that the EE is not doing his job |
| 01:45.32 | RobotGuy | Ahhhhhh, OK. |
| 01:45.42 | RobotGuy | Electric Elephant? |
| 01:45.52 | RobotGuy | Eccentric Eel? |
| 01:46.25 | prpplague | http://www.geeks.com/details.asp?invtid=IR-WEBCAM-PB&cpc=SCH |
| 01:46.34 | RobotGuy | I want a camera I can afford to put a pair of on a robot. |
| 01:46.41 | landley | This impacts nail? Or replying to the blog comment? |
| 01:47.28 | RobotGuy | I think it would be great to have Hammer controlling a pair of cameras. |
| 01:47.45 | prpplague | landley: re the nail , we are doing checks on the layout now |
| 01:47.52 | RobotGuy | It would be even cooler to have Hammer be able to be a complete computer vision processor. |
| 01:47.53 | prpplague | landley: DRC, Design Rule Check |
| 01:48.18 | prpplague | landley: reply to the blog comment |
| 01:48.31 | prpplague | landley: fortunately the hammer/nail stuff is seperate |
| 01:48.41 | RobotGuy | prpplague, WHEW! |
| 01:48.42 | prpplague | landley: the EE for the hammer stuff is great |
| 01:49.11 | RobotGuy | prpplague, Anything new on buildroot and CD images? |
| 01:49.46 | prpplague | RobotGuy: yea the cd has been updated with the new buildroot and all the latest fixes |
| 01:49.57 | prpplague | RobotGuy: i'll be updating the wiki in the morning |
| 01:50.06 | RobotGuy | prpplague, OK, cool. |
| 01:50.27 | RobotGuy | prpplague, Will I need to start a fresh buildroot? |
| 01:50.28 | prpplague | landley: should go to a proto run on friday or monday |
| 01:50.35 | prpplague | landley: 4 days turn around |
| 01:50.41 | prpplague | landley: another day to build up |
| 01:50.54 | prpplague | landley: and then start doing functionality testing |
| 01:54.35 | prpplague | RobotGuy: no |
| 01:54.48 | prpplague | RobotGuy: i figured out a quick fix for your problem |
| 01:54.57 | prpplague | RobotGuy: just do a: make dirclean |
| 01:55.04 | prpplague | RobotGuy: then do your make |
| 01:55.24 | prpplague | RobotGuy: should fix the problem, but it will take a little longer than normal to do the build |
| 01:55.32 | prpplague | RobotGuy: probably around 15 minutes or so |
| 01:56.27 | landley | RobotGuy: the nail board could in theory plug a USB camera into the external USB port. |
| 01:56.53 | landley | RobotGuy: EE is Electrical Engineer, usually. |
| 01:57.07 | prpplague | hehe, and techincally you could daisy change several nail boards, hehe |
| 01:57.09 | landley | The hardware circuit layout person. |
| 01:57.17 | landley | prpplague: nail beowulf! |
| 01:57.27 | RobotGuy | prpplague, Oh, OK! I can do that. |
| 01:57.37 | landley | The difference between theory and practice is that in theory there is no difference, but in practice there is. |
| 01:57.47 | RobotGuy | landley, I know what EE is. :D:D |
| 01:57.58 | landley | In this case, the USB port giving off smoke from the current draw might be a hint of a downside. |
| 01:58.21 | landley | RobotGuy: well you asked. |
| 01:58.43 | landley | It was also the author Cummings's first name, more or less. |
| 01:59.16 | landley | Possibly an insanely large bra size. |
| 02:00.27 | prpplague | landley: yea, you'd have to be able to provide some outside power |
| 02:00.43 | RobotGuy | landley, I was goofing around. |
| 02:00.45 | landley | Plug 'em into a USB hub. |
| 02:00.58 | prpplague | landley: they make a small usb host to usb slave cable that has a wall wart socket |
| 02:01.13 | prpplague | landley: that way you can insert power between the devices |
| 02:01.19 | landley | Yup. You can charge cell phones with 'em. |
| 02:01.35 | landley | The problem with doing that is the devices can't talk to each other directly anymore. |
| 02:01.47 | prpplague | landley: you could do a small openmosix platform |
| 02:01.56 | landley | Break the chain and you haven't got a daisy chain of ethernet connections between 'em. |
| 02:02.16 | landley | Hence the hub: they can still talk to each other through USB, in theory anyway. |
| 02:02.17 | prpplague | landley: or use dist-gcc across all the boards to do native builds |
| 02:02.44 | landley | prpplague: distcc works just as well with a cross compiler. |
| 02:02.56 | landley | All the native vs cross building is handled on the distcc host side. |
| 02:03.13 | landley | If the master machine that's running the distcc build is doing it native, all the slave nodes can use cross compilers and it works out fine. |
| 02:03.41 | landley | So a distcc cluster might as well use x86-64. Best price/performance ratio still. :) |
| 02:04.00 | landley | (of course if you want an insanely large cluster power consumption starts becoming an issue, and then arm sounds good for running it longer than 1 day.) |
| 02:04.34 | landley | So I could see a monster arm based render farm having a total cost of ownership advantage. |
| 02:04.48 | SpeedEvil | not really. |
| 02:04.55 | landley | Especially since you could fit way more nodes into a densely packed area, with heat becoming a primary limiting factor. |
| 02:04.57 | SpeedEvil | The arm isn't great. |
| 02:05.09 | landley | But x86-64 being an order of magnitude faster than arm does start to cut into this. :) |
| 02:05.11 | SpeedEvil | in terms of power/compute |
| 02:05.17 | *** join/#edev prpplague (n=dave@70.249.150.251) |
| 02:05.25 | landley | What's king of the power to performance ratio these days? |
| 02:05.34 | landley | That was always arm's strong point, I thought. |
| 02:05.59 | landley | (Admittedly it had to do with nobody else paying ATTENTION to this before about 2003...) |
| 02:06.04 | prpplague | landley: yea ARM is still at the top of the power to performance |
| 02:06.13 | RobotGuy | landley, Isn't that like comparing apples to oranges? |
| 02:06.26 | landley | RobotGuy: ? |
| 02:06.31 | SpeedEvil | prpplague: even including the 'energy saving' multi-core chips? |
| 02:06.41 | SpeedEvil | prpplague: even including the 'energy saving' multi-core 4Ghz+ chips? |
| 02:06.54 | prpplague | SpeedEvil: oh yea |
| 02:07.16 | RobotGuy | The x87-64 thing |
| 02:07.18 | landley | Does power consumption increase linearly with clock speed? |
| 02:07.23 | RobotGuy | x86-64 |
| 02:07.25 | landley | There was a square in there somewhere last I checked. |
| 02:07.47 | prpplague | SpeedEvil: although the geode stuff is right there |
| 02:08.04 | SpeedEvil | prpplague: how much is arm over the best server chips? |
| 02:08.09 | landley | The Pentium M and its successors (Core X Duo) suck much much less than the P4 based crap that proceeded them, from a power consumption standpoint. |
| 02:08.13 | landley | But you could almost spot weld with a pentium 4. |
| 02:08.37 | landley | Multi-core is _not_ energy saving. |
| 02:08.47 | landley | The ability to switch _off_ extra circuitry is energy saving. |
| 02:08.56 | landley | But having multiple cores in the first place is a performance hack. |
| 02:08.59 | prpplague | landley: it does relate pretty linearly, but the main port of stuff like x86 is just simply the arch was never designed for low power in mind |
| 02:09.09 | landley | Yup. |
| 02:09.21 | prpplague | landley: its like trying to make a lincoln towncar get 45 mpg |
| 02:09.23 | landley | The core of arm is still somewhere in the neighborhood of 43000 transistors, last I checked. |
| 02:09.31 | prpplague | landley: it can be done, but at what cost? |
| 02:09.41 | landley | You add cache and memory management and such to that, but the arm core is small and simple. (Elegant.) |
| 02:10.02 | landley | x86 is layers of nested hackery pipelined for speed and with backwards compatability back to the 8080.. |
| 02:10.24 | landley | (Although I believe they finally ditched the BCD crap put into the 8086 for COBOL support. I think.) |
| 02:10.24 | cajaka | Doesn't gumstix have a chart comparing it's early products vrs a blade server for gflops per watt? |
| 02:10.58 | RobotGuy | prpplague, buildroot did not finish. The makedevs script did not find some files. |
| 02:10.59 | landley | The point is, AMD only started _measuring_ this stuff back around 2002-ish, and Intel around... 2004? 2005? |
| 02:11.13 | landley | Maybe they noticed earlier and it took a while to work through engineering. |
| 02:11.32 | ds2 | landley: AMD did have more "fires" then Intel |
| 02:11.33 | landley | I know the guys in Isreal had a clue, and the guys in bangalore very much did not. |
| 02:11.54 | landley | The K7 had nothing to do with anything that came before. |
| 02:12.09 | landley | When Compaq bought the corpse of DEC, the Alpha design team didn't wind up at compaq, they wound up at AMD. |
| 02:12.25 | landley | The Athlon was the result of asking the Alpha designers "what should an x86 chip look like?" |
| 02:12.47 | landley | Opteron isn't really a descendant of K6 or K5. |
| 02:13.04 | landley | Am I cursing prpplague's router again? |
| 02:13.05 | ds2 | you could almost say the K6 wasn't a decendant of the K5 either |
| 02:13.16 | *** join/#edev prpplague (n=dave@70.249.150.251) |
| 02:13.27 | landley | prpplague: am I cursing your router again? |
| 02:13.34 | RobotGuy | prpplague, buildroot did not finish. The makedevs script did not find some files. |
| 02:13.57 | prpplague | landley: hehe seems so |
| 02:14.11 | prpplague | RobotGuy: i've have never had so much trouble with buildroot |
| 02:14.19 | prpplague | RobotGuy: you did a make dirclean |
| 02:14.24 | ds2 | prpplague: you should've gotten one those routers that flyback was plastering links all over a few months ago |
| 02:14.29 | RobotGuy | prpplague, You said that before. This is ME though. ;) |
| 02:14.37 | RobotGuy | prpplague, I did dirclean |
| 02:14.46 | prpplague | ds2: yea gonna go this week and get a new one |
| 02:15.06 | SpeedEvil | prpplague: hammer + 2* USB network cards |
| 02:15.11 | ds2 | prpplague: you are actually getting one of those? |
| 02:15.21 | prpplague | ds2: no, just a new router |
| 02:15.28 | ds2 | oh =) |
| 02:15.41 | prpplague | SpeedEvil: ? |
| 02:15.47 | SpeedEvil | (for a router) |
| 02:16.05 | prpplague | SpeedEvil: hehe yea i could do that |
| 02:16.34 | prpplague | RobotGuy: so all you changed was the project name? |
| 02:16.35 | RobotGuy | /me is starting to get a complex |
| 02:16.51 | RobotGuy | prpplague, Project name, hostname, and banner. |
| 02:17.23 | prpplague | RobotGuy: and it was working properly before you changed the project name? |
| 02:17.36 | RobotGuy | # |
| 02:17.36 | RobotGuy | # Project Options |
| 02:17.36 | RobotGuy | # |
| 02:17.36 | RobotGuy | BR2_PROJECT="WALTER" |
| 02:17.36 | RobotGuy | BR2_HOSTNAME="walter" |
| 02:17.37 | RobotGuy | BR2_BANNER="W.A.L.T.E.R. welcomes you to The Dynaplex Network" |
| 02:17.39 | RobotGuy | # |
| 02:17.54 | prpplague | hmm |
| 02:18.00 | RobotGuy | prpplague, I have never successfully built a rootfs |
| 02:18.20 | prpplague | RobotGuy: OH, i thought you told me that it built the other day with the default config |
| 02:18.38 | RobotGuy | prpplague, I should say I have never successfully built a rootfs that works. |
| 02:18.55 | RobotGuy | prpplague, Nope. |
| 02:19.01 | ds2 | this sounds like a good opportunity to learn to build rootfs's by hand! |
| 02:19.07 | prpplague | RobotGuy: it was building last night |
| 02:19.17 | ds2 | none of this silly buildroot stuff |
| 02:19.31 | RobotGuy | prpplague, My rootfs has never worked. |
| 02:19.46 | prpplague | RobotGuy: yea it was, you sent me two rootfs's |
| 02:19.55 | prpplague | RobotGuy: but it did build it |
| 02:20.11 | RobotGuy | prpplague, Well, OK, it built, but they never work. |
| 02:20.24 | prpplague | RobotGuy: so whats the error now? |
| 02:21.17 | RobotGuy | makedevs: line 46: regular file '/src/hammer/buildroot/project_build_arm/WALTER/root/etc/shadow' does not exist: No such file or directory |
| 02:21.17 | RobotGuy | makedevs: line 47: regular file '/src/hammer/buildroot/project_build_arm/WALTER/root/etc/passwd' does not exist: No such file or directory |
| 02:21.17 | RobotGuy | makedevs: line 52: regular file '/src/hammer/buildroot/project_build_arm/WALTER/root/usr/share/udhcpc/default.script' does not exist: No such file or directory |
| 02:22.46 | prpplague | RobotGuy: yea, do another make dirclean |
| 02:23.13 | prpplague | RobotGuy: were you doing the build as root before? |
| 02:23.19 | prpplague | RobotGuy: or maybe as a different user? |
| 02:23.27 | RobotGuy | prpplague, I copied Hammer-config back to .config and started fresh with no changes |
| 02:23.51 | RobotGuy | prpplague, I just reset all my permissions to what they should be. |
| 02:24.45 | landley | ds2: That's why I did http://landley.net/code/firmware |
| 02:24.49 | landley | It's a bash script. |
| 02:24.58 | landley | You want to see what it's doing... it's a shell script. |
| 02:25.22 | prpplague | landley: btw as to the openlist() function |
| 02:25.58 | ds2 | IMO, everyone needs to be able to build at least 1 rootfs by hand before using any of the other tools |
| 02:26.00 | prpplague | landley: i still think option 1 is the best, although option 2 is more robust, the whole goal is small effcient code |
| 02:27.15 | landley | Yup. I'm leaning towards that myself. |
| 02:27.27 | landley | Poking at sed just now though, so I can punt on it for the moment. |
| 02:28.25 | Mirell | landley: Picked up the book of igNobel Prizes. |
| 02:30.46 | *** mode/#edev [+o prpplague] by ChanServ |
| 02:32.21 | prpplague | ho ho hum |
| 02:32.55 | prpplague | landley: we debated alot on adding a barrel connector on the nail so that you could provide an external power source |
| 02:33.05 | prpplague | landley: but that sort of goes against the whole goal |
| 02:33.53 | cajaka | Aren't there quite a few usb wall worts out there if power is needed? |
| 02:34.22 | prpplague | cajaka: yea there are a few, but they are mainly for charging |
| 02:34.23 | Mirell | USB for power would be quite cool. |
| 02:35.01 | cajaka | Right... current... duh |
| 02:37.05 | prpplague | http://www.geeks.com/details.asp?invtid=TB-168&cpc=SCH |
| 02:37.45 | prpplague | all the usb chargers like that usually do about 550mA |
| 02:37.56 | prpplague | there are a few that you can get that are up to around 1.5A |
| 02:38.14 | prpplague | but it still kinda defeats the purpose of the nail board |
| 02:39.46 | flyback- | http://deesinbox.com/2007/12/15/the-kittens/ |
| 02:41.07 | ds2 | prpplague: I have seen ones that are rated up to 1A |
| 02:41.12 | cajaka | Think of it this way it may defeat the planned purpose of it but you never know what people will use it for when they sink there teeth into it. |
| 02:41.34 | prpplague | cajaka: indeed |
| 02:41.48 | prpplague | flyback-: those are good eating |
| 02:41.59 | flyback- | really? |
| 02:42.01 | flyback- | insteresting |
| 02:42.37 | prpplague | cajaka: these boards are pretty inexpensive to develop |
| 02:42.41 | landley | Mirell: it's a good book. |
| 02:42.48 | prpplague | cajaka: we had planned to do several others |
| 02:43.17 | landley | prpplague: the _ability_ to use an external power source is nice. The requirement to do so would be bad. |
| 02:43.31 | landley | And paying extra for it or making the device larger would be dubious. |
| 02:43.39 | prpplague | landley: indeed, this would be an optional source, not required |
| 02:44.05 | prpplague | indeed |
| 02:44.06 | landley | How much design tweaking can you do between the initial 100 and the production run? |
| 02:44.13 | Mirell | prpplague: Ya PINGED moi! |
| 02:44.21 | landley | Personally I'd use the initial 100 to try to see the market, and take preorders. |
| 02:44.48 | prpplague | Mirell: hehe, you got used to make sure my router was still alive, hehe |
| 02:44.55 | prpplague | landley: very very little |
| 02:44.56 | Mirell | Ohnos. |
| 02:45.14 | prpplague | landley: the main changes will be between the first 20 and the 100 beta run |
| 02:46.09 | landley | Ok. |
| 02:46.50 | landley | A sufficiently small external power connector would be a nice option to have. |
| 02:46.57 | landley | How much would it add to the price? |
| 02:47.02 | landley | (Charger sold separately.) |
| 02:47.14 | prpplague | about $0.20 |
| 02:47.22 | landley | Would it make the device larger? |
| 02:47.38 | landley | 20 cents is probably going to remain a rounding error even at 5k unit volume. |
| 02:48.18 | prpplague | evey cent counts |
| 02:48.35 | prpplague | hehe, we generally keep track down to $0.0001 |
| 02:48.48 | landley | What I meant was it wouldn't make a difference in the final price large enough to affect purchasing decisions for end users. |
| 02:49.29 | landley | What _would_ be necessary would be making it clear in the docs that you do not need to plug the device into a charger when it's plugged into USB. |
| 02:49.40 | landley | And in fact that plugging it into a charger and usb at the same time is bad juju. |
| 02:49.48 | landley | There's the potential downside of somebody doing that. |
| 02:50.38 | Mirell | That's what magic EE voodoo is for. |
| 02:51.00 | prpplague | landley: actually no, there would be a mosfet that would cut off the 5v from the usb in favor of the wallwart |
| 02:51.19 | landley | As, so daisy chaining would work. :) |
| 02:51.23 | ds2 | IIRC, that mosfet is required for the spec |
| 02:51.42 | prpplague | landley: the switch over would only occur when the power was leveled off and wouldn;t case a reset |
| 02:51.53 | landley | Actually, what you probably want to do is get the little rubber stopper thing and plug up the charger hole by default. Then mention in the docs that you can _get_ a charger, and where to yank the rubber plug if you do. |
| 02:52.07 | landley | How much would the rubber stopper thing add? |
| 02:52.53 | prpplague | landley: yea that would be too costly |
| 02:53.10 | prpplague | landley: plus open barrel connectors are pretty standard on devices |
| 02:53.26 | prpplague | landley: i.e. like to connector on a cell phone for charging |
| 02:53.43 | prpplague | landley: specifically like motorola's |
| 02:54.03 | Mirell | prpplague: Motorola settled on the mini-USB connector |
| 02:54.08 | Mirell | Which I love. |
| 02:54.27 | prpplague | Mirell: for charging? |
| 02:55.06 | Mirell | prpplague: Yep |
| 02:55.18 | prpplague | interesting |
| 02:55.23 | prpplague | i wasn't aware of that |
| 02:55.31 | Mirell | That's what both my RAZR and previous Motorola headset used |
| 02:56.09 | landley | How much does a wall wart->usb connector cost, and how much does a charger to plug into the barrel thingy cost? |
| 02:56.48 | ds2 | prpplague: Motorola did something nonstandard with their miniUSB handling |
| 02:57.35 | prpplague | landley: standard wall-wart for 5v 500mA would be about $8.00 |
| 02:58.03 | flyback- | bah |
| 02:58.03 | prpplague | landley: the wallwart to usb adapter about $2 |
| 02:58.09 | flyback- | you can find switchers for that much now |
| 02:58.31 | ds2 | prpplague: http://wiki.openezx.org/EMU <-- motorola USB "enhancmenets" |
| 02:58.52 | prpplague | flyback-: we are talking commercial items |
| 02:58.52 | landley | So wallwart to usb is actually cheaper? Or do you mean the adapters chain? |
| 02:59.14 | flyback- | I am talking commercial :P |
| 02:59.16 | prpplague | landley: yea if you do the usb adapter you need the wall-wart in addition |
| 02:59.21 | landley | An advantage of having a _second_ power source is that you can chain the suckers. |
| 02:59.36 | landley | It would also add 20 cents to save the end user $2... |
| 02:59.43 | flyback- | prpplague: |
| 02:59.48 | flyback- | quick question about the walnut board |
| 02:59.56 | landley | I'd have to make it explicit in the documentation, with some pictures, but that's ok. |
| 02:59.57 | flyback- | are those caps solid or electrolytic |
| 03:00.12 | landley | Have a documentation section on "chaining hammer boards", and "running a hammer board standalone". |
| 03:00.28 | flyback- | you don't have one?!?! |
| 03:00.29 | flyback- | wow |
| 03:00.32 | agaffney | I do, but I have no idea :P |
| 03:00.43 | flyback- | cause I thought about making that my car pc |
| 03:00.43 | prpplague | landley: yea |
| 03:00.44 | landley | Can you chain 2 of them from a USB port, or do you need a wall wart for each hammer? |
| 03:00.55 | flyback- | need one for each |
| 03:00.55 | prpplague | landley: two would be fine |
| 03:00.59 | flyback- | usb spec is 500ma max |
| 03:01.06 | flyback- | oh it must be like 250 then |
| 03:01.13 | landley | Hmmm... |
| 03:01.19 | prpplague | landley: probably could get away with three, but it would be close |
| 03:01.28 | landley | If a hammer is powered from a wall wart, is it providing the full 500ma in its downstream USB? |
| 03:01.39 | landley | That would mean 3 hammers per wall wart, and 2 from a laptop. |
| 03:01.58 | landley | Of course if you wanted to be _really_ crazy, you could loop 'em. :) |
| 03:02.10 | prpplague | landley: yea if we were to have a wall-wart i'd make it 1.5A |
| 03:02.23 | landley | Dunno _why_ you'd want to loop 'em. |
| 03:02.29 | landley | Ok, so you don't need a wall wart for _each_ hammer. |
| 03:02.29 | prpplague | landley: would would roughly supply power for around 8 nail boards |
| 03:02.39 | landley | I'd also document plugging several into a USB hub. |
| 03:02.50 | prpplague | landley: that would work too |
| 03:02.57 | landley | prpplague: yeah, but the USB port wouldn't output enough for more than 2-3. |
| 03:02.57 | prpplague | landley: using a 4 port powered hub |
| 03:03.09 | landley | A 4 port powered hub might sustain 8 hammer boards? |
| 03:03.15 | prpplague | yea |
| 03:03.26 | prpplague | landley: each port is suppose to supply 500mA |
| 03:03.35 | prpplague | landley: two on each power |
| 03:03.36 | landley | We'd need software for letting the hammer "downstream" hammer boards talk out through the upstream ones. Network routing rules, I suppose. |
| 03:03.37 | prpplague | port |
| 03:03.47 | prpplague | yea |
| 03:03.54 | prpplague | its completely doable |
| 03:03.56 | landley | The hammer board can detect upstream power and provide the safe amount of downstream power automatically? |
| 03:04.02 | prpplague | yea |
| 03:04.21 | landley | With wall wart, it provides 500 downstream, on a laptop it provides 250 downstream, and plugged into 250 it disables downstream entirely? |
| 03:05.04 | landley | (Or provides like 10ma, and only lets you use self-powered devices. I dunno how that works.) |
| 03:06.49 | Mirell | http://arstechnica.com/staff/fatbits.ars/2008/01/06/mwsf-2008-keynote-bingo |
| 03:09.31 | prpplague | landley: into a single laptop port you get 500mA |
| 03:09.54 | prpplague | landley: and that would be passed down to the devices connected to the nail |
| 03:10.31 | prpplague | landley: but if you use a wall-wart onto the nail board, we'd get 1.5A that would be passed down to the devices |
| 03:10.50 | prpplague | landley: each nail board will use about 150mA |
| 03:11.15 | prpplague | landley: so you could get about 9 or 10 hammer boards daisy chained |
| 03:12.04 | prpplague | landley: and unpowered hub gets the base 500mA and distributes it to the other ports |
| 03:12.17 | prpplague | landley: so if its a 4 port hub, each port would get 100mA |
| 03:13.37 | prpplague | landley: it can very confusing |
| 03:17.50 | landley | prpplague: you could pass all 500mA into the downstream USB connector the nail provides, because it's eating 250ma of the power it's getting from upstream. |
| 03:18.10 | landley | If it's eating 150mA, then in theory it could provide 350 downstream, enough for 2 more nail boards. |
| 03:18.18 | landley | Does power consumption change based on CPU load, or is this peak? |
| 03:18.42 | landley | 4 port _powered_ hub could then power 12 nail boards. |
| 03:18.49 | landley | 4 port unpowered hub could power 3. |
| 03:19.24 | prpplague | correct |
| 03:19.42 | landley | But if we program them to daisy chain via the net gadget and some routing rules, the hub becomes optional. |
| 03:19.46 | landley | The unpowered hub. |
| 03:19.55 | prpplague | landley: that too is correct |
| 03:20.16 | landley | must be documented, with pictures. |
| 03:20.21 | prpplague | yep |
| 03:20.27 | landley | I need a digital camera... |
| 03:20.42 | prpplague | once we got some back from proto shop i'll do some tests |
| 03:20.46 | prpplague | landley: they are cheap these days |
| 03:20.53 | landley | I know. |
| 03:20.59 | prpplague | landley: i got a 2.1mp cameras for the kids for just $40 |
| 03:22.05 | landley | Fade has one somewhere, but it's packed. |
| 03:22.19 | landley | I need to take pictures of the nail board plugged into my laptop for the docs. |
| 03:22.34 | landley | "This is uncle ted at the side of the house, but you can see the front of the house". |
| 03:22.48 | prpplague | hehe |
| 03:25.29 | prpplague | ho ho hum |
| 03:35.48 | MonMotha | dammit, I want a 12-13" laptop with UXGA, and I cannot find it |
| 03:35.52 | MonMotha | at any price |
| 03:36.09 | MonMotha | my 15" from 4.5 years ago has UXGA, I'd think they'd have them down at the 12-13" size by no |
| 03:36.10 | MonMotha | w |
| 03:36.53 | Mirell | Woot, my .screenrc is cool now. |
| 03:39.16 | Mirell | Hmm...need to test something |
| 03:41.45 | *** join/#edev Mirell (n=mirell@quiche.ling.utexas.edu) |
| 03:42.36 | Mirell | Ah, excellent. |
| 03:54.53 | prpplague | landley: what kind of stuff would you want to do with multiple nail boards? |
| 03:57.13 | Mirell | Beowulf cluster. |
| 03:57.13 | Mirell | Duh. |
| 03:59.36 | landley | prpplague: dunno yet. :) |
| 03:59.47 | landley | It's more cool than useful at the moment. |
| 04:00.07 | prpplague | hehe |
| 04:04.23 | Mirell | Have a massive Beowulf cluster with RAID-5 NAS |
| 04:05.23 | prpplague | hehe |
| 04:05.30 | prpplague | all via usb |
| 04:06.00 | *** join/#edev samuelr (n=samuelr@dsl093-061-094.pit1.dsl.speakeasy.net) |
| 04:06.55 | Mirell | prpplague: Of course |
| 04:07.13 | Mirell | Hmm...I might get my company to pay for my cell phone and Internet |
| 04:08.03 | prpplague | Mirell: low power computing system for things like disaster sites |
| 04:08.19 | prpplague | Mirell: need more processing power, plug another nail in |
| 04:08.45 | prpplague | Mirell: lower power enough to run off a small generator or even a 12 or 24v battery |
| 04:10.20 | MonMotha | prpplague: know much about current laptops? |
| 04:10.30 | MonMotha | I'm looking for a 12-15" with UXGA or I'll take WUXGA |
| 04:11.28 | MonMotha | lighter is better, but I do want an optical drive |
| 04:43.36 | *** join/#edev DrZimmerman (n=theo@2001:5c0:8fff:fffe:0:0:0:6af1) |
| 04:45.19 | flyback- | http://www2.gmer.net/mbr/ <-- this could be bad for linux as it could be a nice excuse to implement full mode on trusted computing and drm |
| 05:04.53 | *** join/#edev FireEgl (i=FireEgl@Sebastian.Atlantica.US.TO) |
| 05:48.00 | *** join/#edev RobotGuy (n=RobotGuy@dsl093-038-120.pdx1.dsl.speakeasy.net) |
| 06:05.35 | *** join/#edev redblue (i=star@ppp042.108-253-207.mtl.mt.videotron.ca) |
| 06:09.32 | Mirell | 'night flyback- |
| 08:46.48 | Mirell | I definitely need to get a wall-mount thing for my bike. |
| 08:58.03 | *** join/#edev hw (n=hw@p578b3905.dip0.t-ipconnect.de) |
| 10:00.44 | *** join/#edev lyakh (n=lyakh@p57BD2812.dip0.t-ipconnect.de) |
| 12:27.00 | *** join/#edev lyakh (n=lyakh@p57BD2812.dip0.t-ipconnect.de) |
| 14:04.44 | *** join/#edev T0mW (n=Tom@24.238.86.158.res-cmts.sth.ptd.net) |
| 14:22.02 | *** join/#edev prpplague (n=dave@mail.americanmicrosystems.com) |
| 14:22.03 | *** join/#edev dijenerate (n=dijenera@66.54.118.16) |
| 14:24.42 | *** join/#edev chouimat|work (n=mchouina@kde/developer/chouinard) |
| 14:25.38 | chouimat|work | morning |
| 14:29.41 | *** mode/#edev [+o prpplague] by ChanServ |
| 14:33.33 | *** join/#edev sjhill (n=sjhill@real.realitydiluted.com) |
| 14:34.04 | prpplague | sjhill: greetings |
| 14:34.12 | prpplague | sjhill: whats cookin in mips world? |
| 14:35.22 | *** join/#edev linac (n=lin@121.41.43.232) |
| 14:46.55 | *** join/#edev g1powermac (n=g1powerm@unaffiliated/g1powermac) |
| 14:47.09 | *** join/#edev GPSFan (n=kenm@12.10.255.246) |
| 14:47.09 | *** mode/#edev [+o GPSFan] by ChanServ |
| 14:47.15 | prpplague | GPSFan: greetings |
| 14:47.25 | GPSFan | prpplague: howdy.. |
| 14:47.33 | prpplague | ~quote scoxq.pk |
| 14:47.40 | prpplague | GPSFan: whats cookin? |
| 14:48.17 | sjhill | prpplague: not much |
| 14:48.24 | GPSFan | prpplague: not much, all I got done yesterday is plowing the driveway. |
| 14:48.36 | prpplague | GPSFan: hehe |
| 14:49.03 | sjhill | prpplague: i got a huge bonus this past year and another raise...and my responsibilities have shifted to architect duties...i miss coding |
| 14:49.32 | GPSFan | prpplague: sounds like you have got quite an interest from the robotics folks in the Hammer. ;>) |
| 14:49.42 | prpplague | sjhill: wow nice |
| 14:49.54 | GPSFan | sjhill: congrats.. |
| 14:49.57 | sjhill | thx |
| 14:49.59 | sjhill | i think |
| 14:50.00 | prpplague | GPSFan: so far it looks that way |
| 14:50.14 | prpplague | GPSFan: still working on getting people interested in trying the hammer |
| 14:50.50 | prpplague | sjhill: hehe, guess you didn't need to the extra work for xmas money after all |
| 14:51.07 | sjhill | heh, yeah |
| 14:51.16 | GPSFan | I liked the picture someone posted of the hammer & gummstick side-by-side. |
| 14:51.24 | sjhill | but i will if i attempt to buy any more Thomas the Train stuff...good grief |
| 14:51.33 | prpplague | sjhill: HA |
| 14:51.39 | prpplague | sjhill: btdt |
| 14:51.47 | prpplague | GPSFan: yea that was a nice pic |
| 14:52.31 | GPSFan | prpplague: now you need a table comparing the pro's & cons of each. for the wiki. |
| 14:52.42 | prpplague | GPSFan: yea |
| 14:53.30 | *** join/#edev CosmicPenguin (n=nobody@163.181.251.103) |
| 14:53.31 | *** mode/#edev [+o CosmicPenguin] by ChanServ |
| 14:57.38 | chouimat|work | hi prpplague GPSFan CosmicPenguin sjhill |
| 14:58.14 | GPSFan | chouimat|work: howdy. |
| 14:58.50 | prpplague | sjhill: hehe, sounds like you are better off now than you were are timesys or broadcom |
| 14:59.17 | CosmicPenguin | GPSFan: the news tells me you are buried under like 100 feet of snow or something |
| 14:59.52 | GPSFan | CosmicPenguin: no, it's 200' haha |
| 15:00.00 | chouimat|work | snow?? where??? it's melting here :( |
| 15:00.26 | GPSFan | actually only about 10", the first storm was ~1/2" of rain. |
| 15:00.58 | GPSFan | it was melting yesterday morning, today it's 13F. |
| 15:01.38 | CosmicPenguin | I hear the san juan's are getting pummelled |
| 15:02.35 | prpplague | CosmicPenguin: greetings |
| 15:02.45 | prpplague | chouimat|work: greetings |
| 15:03.32 | *** join/#edev dijenerate (n=dijenera@72.51.111.182) |
| 15:03.48 | GPSFan | Wolf Creek got 49" |
| 15:04.52 | CosmicPenguin | 49" is a curse - you have to dig out your damn lifts |
| 15:05.01 | prpplague | hehe |
| 15:05.07 | chouimat|work | :)\ |
| 15:05.17 | GPSFan | CosmicPenguin: their snowfall ytd is 249" !!!!! |
| 15:05.33 | CosmicPenguin | yeah, I saw a picture of silverton - you couldn't see the front of the buildings |
| 15:05.38 | prpplague | CosmicPenguin: yea well, telluride is soo expensive now, its not like normal people can afford it |
| 15:05.42 | chouimat|work | wow this more than 6m |
| 15:09.19 | sjhill | CosmicPenguin! |
| 15:10.30 | CosmicPenguin | sjhill! |
| 15:27.34 | wmat | prpplague: did you ping me last night about something? |
| 15:28.46 | prpplague | wmat: yea but i don't recall what it was about |
| 15:30.40 | wmat | prpplague: ok |
| 15:33.48 | prpplague | wmat: oh i remember now |
| 15:33.59 | prpplague | wmat: i was considering talking to tim bird about the nail board |
| 15:35.25 | T0mW | we are supposed to hit a high of 66F today |
| 16:03.45 | wmat | prpplague: talking to him about it in what context? |
| 16:05.02 | prpplague | wmat: basically just showing him the proto design, the specs, maybe sending him one for an eval |
| 16:05.36 | prpplague | wmat: just don't know if thats the kind of item he'd be interested in, and in addition with him basically working for sony, not sure how that would work looking at our products |
| 16:06.22 | CosmicPenguin | prpplague: I know what he'll say... |
| 16:06.27 | CosmicPenguin | bring it to the CELF conference |
| 16:06.34 | wmat | heh |
| 16:06.35 | CosmicPenguin | so, cut out the middle man |
| 16:06.41 | wmat | prpplague: can't hurt to ping him |
| 16:07.10 | prpplague | CosmicPenguin: yea |
| 16:07.18 | prpplague | CosmicPenguin: you attended CELF last year right? |
| 16:07.24 | CosmicPenguin | I didn't go last year |
| 16:07.57 | CosmicPenguin | We have pretty aggressive travel restrictions on right now, I have to pick and chose my battles |
| 16:08.22 | CosmicPenguin | so, cut out the middle man |
| 16:08.29 | CosmicPenguin | oops - sorry |
| 16:08.31 | CosmicPenguin | irssi is being wierd |
| 16:09.13 | prpplague | CosmicPenguin: have you ever attended CELF? |
| 16:09.22 | CosmicPenguin | yeah, I spoke there two years ago |
| 16:09.28 | wmat | prpplague: on an unrelated note - my new years resolution is to get a beer related site out there |
| 16:09.47 | prpplague | wmat: ?? your own? |
| 16:09.58 | wmat | prpplague: yeah, beerhacker.net |
| 16:10.00 | prpplague | CosmicPenguin: ahh i remember now |
| 16:10.15 | wmat | prpplague: home brew stuff |
| 16:10.25 | prpplague | wmat: ahh, don't like ratebeer? |
| 16:10.43 | CosmicPenguin | mmm - homebrew |
| 16:10.49 | wmat | prpplague: no, i do actually |
| 16:11.17 | CosmicPenguin | wmat: you know about http://www.opensourcebeerproject.com/, right? |
| 16:11.19 | prpplague | CosmicPenguin: what conferences do you have on your agenda then? |
| 16:11.53 | CosmicPenguin | prpplague: I'm going to the LinuxBIOS portion of the High Performance Computing Week, but thats pretty simple because its in Denver this year |
| 16:11.53 | wmat | CosmicPenguin: I do now, thx |
| 16:12.08 | CosmicPenguin | wmat: its a really cool idea, but the beer they ended up with isn't _that_ great |
| 16:12.10 | CosmicPenguin | its okay |
| 16:12.20 | CosmicPenguin | prpplague: I'm thinking about CELF |
| 16:12.22 | CosmicPenguin | and probably OLS |
| 16:12.42 | CosmicPenguin | But the baby is due on Feb 12, so that might affect things greatly |
| 16:12.45 | wmat | my thinking is more geared toward beer brewing projects with a tech component |
| 16:12.56 | prpplague | CosmicPenguin: baby? |
| 16:13.25 | chouimat|work | wmat: you mean something like this? http://www.asciimation.co.nz/bender/ |
| 16:13.31 | CosmicPenguin | wmat: like the bender brewer? :) |
| 16:13.37 | CosmicPenguin | beat me! |
| 16:13.39 | wmat | chouimat|work: yes, i saw that yesterday actually |
| 16:13.59 | wmat | and like this: http://www.lemis.com/grog/brewing/ |
| 16:14.22 | prpplague | CosmicPenguin: uh your boss lady preggy? |
| 16:14.48 | CosmicPenguin | prpplague: yeah |
| 16:15.00 | CosmicPenguin | oh - cool: http://www.blogaboutbeer.com/2008/01/02/new-belgiums-fat-tire-is-for-the-fishes/ |
| 16:15.16 | prpplague | CosmicPenguin: wow, don't know how i didn't know about that |
| 16:15.18 | CosmicPenguin | yay beer + science |
| 16:15.37 | wmat | CosmicPenguin: nice |
| 16:15.53 | wmat | CosmicPenguin: also, congrats on the pending baby. They're lots of fun. |
| 16:17.11 | *** part/#edev koolatron (n=sean@cpe-066-057-115-047.nc.res.rr.com) |
| 16:18.12 | CosmicPenguin | fun isn't the word I keep hearing |
| 16:18.20 | wmat | heh |
| 16:20.58 | chouimat|work | CosmicPenguin: more like "lack of sleep" ? |
| 16:24.08 | Mirell | Apparently, my company health care will be $52 per pay period. Sweet. |
| 16:24.58 | CosmicPenguin | prpplague: apparently, you need to read my blog more |
| 16:26.42 | flyback- | prpplague: |
| 16:26.46 | flyback- | you will enjoy this |
| 16:26.46 | flyback- | hang on |
| 16:26.58 | flyback- | http://dailymotion.alice.it/video/x3wrzo_fabrication-dune-lampe-triode_tech |
| 16:27.36 | flyback- | I have a isa rtl8019 now I just need to flash the webpal |
| 16:27.50 | flyback- | I wish I could figure out how to setup a dialin without a carrier |
| 16:28.08 | flyback- | so I could use the trick someone came up with that uses the webpal's build in os flash routines to flash to linux |
| 16:29.11 | flyback- | actuallY I made a +5 ttl serial card and found what appears to be a discreit diodes and transistors level converter thing from a low end pda but dunno what voltages it works with |
| 16:57.57 | *** join/#edev RobotGuy (n=RobotGuy@dsl093-038-120.pdx1.dsl.speakeasy.net) |
| 16:59.32 | *** join/#edev ccherret1 (n=chris@S0106000c413c715f.ed.shawcable.net) |
| 17:18.49 | *** join/#edev samuelr (n=samuelr@dsl093-061-094.pit1.dsl.speakeasy.net) |
| 17:26.01 | *** join/#edev RobotGuy (n=RobotGuy@dsl093-038-120.pdx1.dsl.speakeasy.net) |
| 17:27.06 | RobotGuy | Hey prpplague, are you awake?? |
| 17:27.42 | Mirell | He died, come back next awakening. |
| 17:28.39 | RobotGuy | G'Morning, Mirell |
| 17:31.15 | RobotGuy | I have to reboot into Windows. :( BRB |
| 17:50.29 | prpplague | CosmicPenguin: hehe |
| 17:50.50 | prpplague | CosmicPenguin: yea i guess i do |
| 17:57.14 | Mirell | Hmm... |
| 17:57.29 | Mirell | I should go eat lunch here soon. |
| 17:59.21 | *** join/#edev RobotGuy (n=RobotGuy@dsl093-038-120.pdx1.dsl.speakeasy.net) |
| 18:00.26 | prpplague | RobotGuy: pong |
| 18:00.57 | RobotGuy | prpplague, I got Squeak working on my PC, and I am going to a joint Smalltalk/PARTS meeting tonight, taking Hammer with me. |
| 18:01.16 | prpplague | RobotGuy: yea thats the one scheer is holding? |
| 18:01.43 | RobotGuy | prpplague, It's a joint meeting with the local Smalltalk group and PARTS (Robert Sheer). |
| 18:02.25 | prpplague | ahh |
| 18:02.44 | RobotGuy | prpplague, IF this meeting is a success, there will most likely be more, especially since Roboert Sheer uses Squeak. |
| 18:03.12 | RobotGuy | I just got Squeak running, but have no clue what to do with it yet. |
| 18:03.25 | SpeedEvil | Build a better mousetrap? |
| 18:03.44 | RobotGuy | SpeedEvil, Maybe my kitty would like to play with it. |
| 18:03.44 | SpeedEvil | One that hunts mice using vision. |
| 18:06.30 | RobotGuy | prpplague, Two PARTS members have either ordered a Hammer or are interested in using it. I am not a member yet, but will join next month. |
| 18:07.42 | *** join/#edev alec_v (n=alec@monitor2.gin.spb.ru) |
| 18:07.49 | prpplague | RobotGuy: yea dave hylands has already ordered his |
| 18:07.54 | RobotGuy | Dave Hylands is creating a buildroot package for Squeak. |
| 18:08.07 | prpplague | yea |
| 18:08.35 | RobotGuy | prpplague, I know you know because you read PARTS now, but others may not know. |
| 18:08.46 | prpplague | ahh ok |
| 18:09.27 | RobotGuy | I am very excited about the interest in Hammer within PARTS. |
| 18:09.44 | prpplague | me too, hope it translates into sales |
| 18:09.55 | RobotGuy | prpplague, Me too! |
| 18:12.52 | *** join/#edev redblue (i=star@ppp249.108-253-207.mtl.mt.videotron.ca) |
| 18:24.20 | RobotGuy | Got to get a haircut. BBIAB. |
| 18:35.28 | *** join/#edev landley (n=landley@adsl-76-197-33-46.dsl.austtx.sbcglobal.net) |
| 18:51.12 | prpplague | davidc__: hehe they missing you at SFE yet? |
| 18:55.59 | ds2 | prpplague: Calibrating delay loop... 397.31 BogoMIPS (lpj=1986560) |
| 18:57.24 | prpplague | ds2: yea that matches what i found on the net |
| 18:59.28 | landley | I take it high resolution timers with short enough periods to actually use for this sort of delay aren't commonly available? |
| 19:00.34 | prpplague | landley: i'm sorry, i didn't follow your question |
| 19:01.09 | landley | bogomips is busy waiting. I dunno why "schedule timer interrupt, HLT, receive timer interrupt" isn't more commonly used for those sort of things. |
| 19:02.49 | prpplague | landley: ahh, i was mainly looking at the bogomips for the pxa255 verses the s3c2410 |
| 19:03.11 | prpplague | landley: of course its not very relevant, but many people still seem to use it as a guide to performance |
| 19:04.17 | landley | It's one of the biggest sources of bootup delay on an optimized system. |
| 19:04.42 | landley | Feed the bogomips loop iteration value thingy in as a kernel command line option. |
| 19:04.47 | landley | Yank the darn floppy driver already. |
| 19:04.55 | landley | Tweak IDE so it doesn't probe for nonexistent disks... |
| 19:07.35 | prpplague | landley: indeed, i have that as a default argument in apex bootloader |
| 19:11.03 | landley | Which naturally leads to the question "what needs this"? |
| 19:11.26 | landley | Fundamentally, it's used to turn a busy-wait into a number of nanoseconds. |
| 19:11.39 | landley | Because actually using a _timer_ would be inadvisable somehow. |
| 19:12.12 | landley | I admit that some timers only have milisecond resolution (or worse). |
| 19:12.23 | landley | I'm just wondering where in the code actually _needs_ micro-waits. |
| 19:12.35 | landley | (Several of which were using a port 80 write anyway.) |
| 19:15.50 | SpeedEvil | For example, some hardware is unreliable at back-back write speeds. |
| 19:16.17 | landley | SpeedEvil: I know that. |
| 19:16.34 | landley | And some hardware is (strangely) much faster with a slight delay between transactions. (Go figure.) |
| 19:16.46 | prpplague | landley: yea, i've wonder myself if there is any real need for that level of accuracy |
| 19:17.10 | landley | I'm not saying we don't _need_ the delay, I'm just wondering where specifically it's used and whether there's a better alternative using a timer chip. |
| 19:17.19 | SpeedEvil | not really. |
| 19:17.20 | landley | Thomas Gleixnenr was pounding on timers, I should bug him. |
| 19:17.29 | SpeedEvil | you end up using the timer chip resources |
| 19:17.35 | landley | Considering it's something like 3 am in germany, I doubt he's awake right now though. :) |
| 19:17.36 | SpeedEvil | they typically only have so many timers |
| 19:17.59 | landley | SpeedEvil: so you have a kernel subsystem that keeps track of the expiration requests and sets the timer chip for the soonest one. |
| 19:18.08 | landley | You only ever _need_ one timer per processor. |
| 19:18.14 | landley | The rest can be handled in software. |
| 19:18.20 | chouimat|work | 3 am in germany??? it's 8pm |
| 19:18.25 | landley | Is it? |
| 19:18.29 | SpeedEvil | Consider what else you might do. |
| 19:18.53 | SpeedEvil | Such as drop back out of the interrupt handler and start executing user code |
| 19:19.10 | chouimat|work | landley: it's +6hours from where I am |
| 19:19.10 | SpeedEvil | the overheads of this can be several hundred cycles |
| 19:19.32 | SpeedEvil | (on cisc processors with large caches |
| 19:19.37 | landley | SpeedEvil: sure. |
| 19:20.01 | SpeedEvil | all you do is make the timer less accurate, for a vanishingly small slice of work. |
| 19:20.09 | landley | But in that case, can't you configure bogomips by figuring out what the smallest possible _timer_ overhead is? |
| 19:20.16 | landley | And only busy-wait if the delay is smaller than that? |
| 19:20.29 | landley | A bogomips value over a couple thousand says to me "could have used a timer". |
| 19:20.30 | SpeedEvil | Sure. In principle. |
| 19:20.35 | landley | Then the boot time delay goes away. |
| 19:21.00 | SpeedEvil | In some cases, it makes driver code a bit more complex. |
| 19:21.31 | landley | Why isn't it a library function? |
| 19:21.31 | SpeedEvil | and will increase memory traffic, which may hurt power use |
| 19:21.36 | landley | wait_this_long(nanoseconds) |
| 19:22.10 | landley | spinning >1000 cycles takes more power than bouncing off a timer chip and going to sleep? |
| 19:22.25 | landley | There should be one wait function that decides whether or not to spin or sleep. |
| 19:22.27 | SpeedEvil | but you're not going to sleep - in most cases. |
| 19:22.50 | landley | So that would be this function deciding to spin instead. |
| 19:22.59 | SpeedEvil | you're going to user mode, reseting all the various in/out-of-kernel bits, fetching more code into the cache, ... |
| 19:23.15 | landley | SpeedEvil: what the...? |
| 19:23.19 | landley | bogomips is used in the kernel. |
| 19:23.30 | landley | Why would you switch to userspace? |
| 19:23.48 | SpeedEvil | Oh sorry |
| 19:24.06 | SpeedEvil | I thought you were arguing against the use of busy-wait delays |
| 19:24.12 | landley | I am. |
| 19:24.25 | landley | But "halt the CPU until timer says hi" != schedule() |
| 19:24.56 | SpeedEvil | many CPUs have non-deterministic behaviour on coming back from halt. |
| 19:25.32 | SpeedEvil | If you were to submit a patch for hardware that made it do that on hardware that supported it, I don't think it'd get rejected. |
| 19:26.09 | SpeedEvil | I just doubt the time spent in udelay is a significant cpu usage in any real case. |
| 19:26.34 | SpeedEvil | As to calculating the bogomips - that delay - I'm not arguing that that's a good thing. |
| 19:27.23 | landley | SpeedEvil: I think tglx has been doing that. |
| 19:27.28 | landley | It's part of the timer rewrite patches. |
| 19:27.32 | landley | I'm just wondering what's _left_. |
| 19:27.51 | SpeedEvil | For powersaving stuff you mean? |
| 19:28.10 | landley | SpeedEvil: in general. |
| 19:28.12 | *** join/#edev adyer (n=adyer@adsl-69-223-13-78.dsl.chcgil.ameritech.net) |
| 19:28.15 | SpeedEvil | I noted on the OM wiki that some user-mode stuff can help. |
| 19:28.24 | landley | Remember the timer and irq patches that went in... 9 months ago, maybe? |
| 19:28.28 | SpeedEvil | For example, turning X to 8-bit, or lower, not 24 |
| 19:28.49 | SpeedEvil | To decrease memory bandwidth and hence power. |
| 19:28.51 | landley | SpeedEvil: there's 1000 things you can do, but that's a different issue. |
| 19:29.01 | landley | Bogomips calculation is one of the big boot time delays. |
| 19:29.17 | landley | In many systems, _the_ biggest remaining one, since it's a fixed time delay and the other scale with speed of the system. |
| 19:29.24 | SpeedEvil | How precise is it at the moment? |
| 19:29.29 | prpplague | landley: indeed, and the worst part is people use that as a guide to performance |
| 19:29.33 | landley | You can wire around it, but only in a very platform-specific way. |
| 19:29.41 | landley | prpplague: which it's never been. |
| 19:30.09 | SpeedEvil | I can see rewriting it so that it gets 1% error forex. |
| 19:30.23 | SpeedEvil | which may be significantly faster on some systems. |
| 19:30.39 | prpplague | SpeedEvil: the big thing is just doing the calibration once, and then using that value over and over as a kernel argument |
| 19:31.08 | prpplague | SpeedEvil: just add something like: lpj=1986560 |
| 19:31.10 | SpeedEvil | The trivial option would be a bootloader argument |
| 19:31.17 | prpplague | SpeedEvil: to the kernel argument and you are good |
| 19:31.42 | SpeedEvil | Also, modules can help |
| 19:31.43 | prpplague | SpeedEvil: yea, if you just add it to your default kernel args in the bootloader it works great |
| 19:32.05 | SpeedEvil | cat /proc/bogomips or something |
| 19:32.10 | SpeedEvil | to do the current computation |
| 19:32.23 | prpplague | SpeedEvil: for the s3c2410 is was able to shorted the boot time by almost 10 seconds by simply passing lpj as an argument |
| 19:32.55 | SpeedEvil | what's it at? |
| 19:32.59 | SpeedEvil | (boot time) |
| 19:33.20 | SpeedEvil | for a minimal kernel, say supporting nand, and nothing else |
| 19:33.38 | prpplague | SpeedEvil: well, with the correct config i boot from power on to prompt in just under 3 seconds |
| 19:33.51 | SpeedEvil | prpplague: that is nice. |
| 19:34.26 | SpeedEvil | where are the big delays remaining - uncompressing? |
| 19:34.45 | prpplague | SpeedEvil: yea, the copy from flash to sdram, and decompressing |
| 19:34.53 | prpplague | SpeedEvil: about 1 second each |
| 19:35.07 | prpplague | SpeedEvil: and about 1 second to boot |
| 19:35.36 | SpeedEvil | I wonder if you can fiddle the decompress algorithm |
| 19:36.45 | landley | I've seen boot times under 1 second. |
| 19:36.59 | landley | My complaint is "why do you need to coustom configure this to get a fast boot time?" |
| 19:37.15 | landley | Hence asking "what is bogomips still used for, is there any way to just not have it?" |
| 19:37.20 | landley | Seems worth asking, at least... |
| 19:37.42 | prpplague | landley: indeed |
| 19:37.47 | SpeedEvil | What is the inaccuracy if you shorten the period of calculation to 10ms? |
| 19:38.46 | landley | SpeedEvil: depends on the board probably. |
| 19:38.55 | landley | But 10ms should be plenty. |
| 19:39.05 | landley | If you _need_ a 10ms wait, even on a slow board, USE A TIMER CHIP. |
| 19:39.48 | landley | Spins are for nanosecond waits. |
| 19:40.37 | landley | A milisecond is a _million_ nanoseconds. |
| 19:45.09 | prpplague | another item for 2.6 kernels that drives me nuts is the initramfs detection |
| 19:45.32 | prpplague | usually on my inhouse kernels we patch that as a kernel arg as well |
| 19:46.05 | prpplague | so you can specify that the image is either initramfs or initrd |
| 19:46.09 | SpeedEvil | what's bad about it? |
| 19:46.09 | prpplague | no checking is done |
| 19:46.26 | prpplague | SpeedEvil: it usually takes about 3-5 seconds |
| 19:46.41 | SpeedEvil | oh. /me doesn't boot very often. |
| 19:47.02 | prpplague | SpeedEvil: well on handheld device, you power off and on alot |
| 19:47.13 | SpeedEvil | prpplague: you shouldn't though. |
| 19:47.26 | SpeedEvil | prpplague: if suspend-RAM works properly and fast. |
| 19:48.40 | SpeedEvil | suspend-RAM + self-refresh of course |
| 19:48.47 | prpplague | SpeedEvil: one sec |
| 19:51.01 | prpplague | SpeedEvil: even with suspend we have to be pretty power aware |
| 19:51.17 | prpplague | SpeedEvil: if the device boots in under 3 seconds, suspend isn't always needed |
| 19:51.23 | SpeedEvil | True - but suspend-RAM should be on the order of a milliwatt. |
| 19:51.24 | prpplague | SpeedEvil: therefor we longer battery life |
| 19:51.47 | SpeedEvil | If your application suite can come up in a second too, sure. |
| 19:51.52 | prpplague | SpeedEvil: indeed it is, but when we are talking 10-15 hours of usage on a single charge, every milliwatt counts |
| 19:52.50 | SpeedEvil | But, an extra 2 seconds doing moderate CPU booting and 2s doing application booting is 400s in suspend-RAM |
| 19:53.12 | prpplague | SpeedEvil: sorry didn't follow that |
| 19:53.36 | SpeedEvil | 4s*100mw = 400s*1mw |
| 19:54.09 | SpeedEvil | if you spend cpu time - and hence energy - booting, then you could as easily spend it on staying in suspend-RAM, which you can come back from faster. |
| 19:54.22 | SpeedEvil | which may save energy. |
| 19:54.29 | prpplague | SpeedEvil: indeed, but when it's sitting unused overnight on the desk, that advantage goes away |
| 19:54.34 | SpeedEvil | depending on times. |
| 19:55.23 | SpeedEvil | A 1200mAh li-ion cell gives you some 5000 hours of suspend. |
| 19:55.52 | prpplague | SpeedEvil: hehe 1200mAh also only gives us about 4 hours of usage |
| 19:56.30 | SpeedEvil | in what? |
| 19:56.38 | SpeedEvil | prpplague: not the hammer I assume |
| 19:56.45 | landley | prpplague: you should only have to decompress the first 4k or so. |
| 19:56.48 | prpplague | SpeedEvil: no, in our handheld devices |
| 19:56.51 | landley | With gzip, that shouldn't be too expensive. |
| 19:57.15 | SpeedEvil | prpplague: are the handhelds running any UI? |
| 19:57.19 | prpplague | landley: yea, i didn't spend alot of time tweaking the uncompression |
| 19:57.23 | prpplague | SpeedEvil: yea |
| 19:57.28 | SpeedEvil | prpplague: and if so how fast does it come up? |
| 19:57.38 | prpplague | SpeedEvil: 3 seconds |
| 19:57.41 | landley | prpplague: I wrote my own bunzip2 for busybox. (And a bzip2 i never merged, but plan to put into toybox.) |
| 19:57.42 | prpplague | from power on to usage |
| 19:57.47 | landley | I should do the same for gzip. |
| 19:57.55 | landley | Using the existing library verbatim isn't always the best choice. |
| 20:04.49 | SpeedEvil | prpplague: if you're getting 4h usage, you're at around a watt, so 2s*1w saving - if you can resume in 1s - it can run some 40min in suspend-RAM and save power. |
| 20:05.02 | SpeedEvil | (based on the first 128M RAM chips numbers I found) |
| 20:06.33 | prpplague | SpeedEvil: indeed, but you are missing the point on several areas |
| 20:06.39 | SpeedEvil | probably. |
| 20:06.55 | prpplague | SpeedEvil: first we have to get atleast 8hours usage with 12-15 prefered |
| 20:07.01 | SpeedEvil | If your average time between boots is under that, then it's a net loss. |
| 20:07.11 | SpeedEvil | over that rather |
| 20:07.12 | prpplague | SpeedEvil: second, often times these devices sit on a desk overnight unused |
| 20:07.20 | prpplague | SpeedEvil: they have to be ready in the morning |
| 20:08.01 | prpplague | SpeedEvil: in addition you are assuming that the boot up power requirements are the same as normal operations |
| 20:08.05 | SpeedEvil | Turning off suspend-RAM to off mode after x seconds isn't really hard in principle. Yes it's extra coding. |
| 20:08.05 | prpplague | SpeedEvil: which they aren't |
| 20:08.17 | SpeedEvil | display a big chunk? |
| 20:08.26 | prpplague | SpeedEvil: wifi and barcode reader |
| 20:09.17 | SpeedEvil | the big chunk of course is if you can go to sleep while the user is thinking. |
| 20:09.33 | SpeedEvil | Without them noticing. |
| 20:10.04 | SpeedEvil | Sorry, please feel free to ignore me, and I'll run down sooner or later :) |
| 20:10.25 | prpplague | SpeedEvil: hehe, np |
| 20:11.17 | prpplague | SpeedEvil: http://www.elinux.org/M5900 |
| 20:13.11 | SpeedEvil | I'm just doing other stuff while trying to digest the philosophy of all the datasheets. |
| 20:13.11 | prpplague | landley: yea i didn't tweak the decompress stuff, just used the stock stuff |
| 20:13.40 | prpplague | landley: our current boot time was acceptable, so i didn't dig any further |
| 20:13.58 | prpplague | landley: there are several good OLS papers on boot time tweaking |
| 20:14.06 | prpplague | landley: alot of it is post on the elinux wiki |
| 20:16.39 | landley | And a presentation on it at CELF in 2006. |
| 20:17.12 | prpplague | landley: yea iirc thats also one of the ones posted on the wiki |
| 20:19.51 | prpplague | http://yro.slashdot.org/yro/08/01/08/1641209.shtml |
| 20:19.54 | prpplague | hehe |
| 20:20.11 | prpplague | i had a laptop confiscated and destroyed when returning from canada in 2001 |
| 20:20.35 | prpplague | they thought it was a fake laptop since it didn't boot windows |
| 20:21.20 | prpplague | never did get paid for it |
| 20:21.59 | Mirell | ... |
| 20:22.59 | agaffney | government-- |
| 20:23.04 | prpplague | i waited at customs for almost 12 hours, missed my flight |
| 20:23.05 | landley | prpplague: I drive all over the place, largely because the airlines are unusable with the Office of Domestic Surveilance being in charge of airport "security". |
| 20:23.16 | landley | Just issue us a Linus blanket and move on already. |
| 20:23.17 | prpplague | landley: indeed |
| 20:23.19 | agaffney | I just don't travel :P |
| 20:23.28 | CosmicPenguin | Linus makes blankets now? |
| 20:23.31 | prpplague | hehe |
| 20:23.41 | landley | I drive to michigan for penguicon. I drive to pennsylvania to visit Eric. Drive to florida to visit my grandparents. |
| 20:24.02 | prpplague | agaffney: well after living overseas for 2.5 years, i doubt i'll travel outside of the country much again |
| 20:24.02 | landley | Probably fly to california and leave Fade the car, though. |
| 20:24.20 | prpplague | landley: why not take a train? |
| 20:24.27 | prpplague | landley: afraid of amtrak? hehe |
| 20:24.43 | landley | It's more even more expensive than driving? |
| 20:24.45 | adyer | lol was in the airport yesterday, I think it was "screener training day" they were so clueless and slow |
| 20:24.48 | landley | Southwest is cheaper than the train. |
| 20:25.31 | prpplague | landley: i don't know about the current cost, but 4 years ago i went to california for just $99 |
| 20:27.44 | prpplague | landley: $135 from austin to san jose |
| 20:27.51 | landley | Southwest is cheap. |
| 20:29.56 | Crofton | they've stoppe dhaving you turn on laptops ... |
| 20:30.00 | prpplague | landley: business class from austin to san jose is $339, coach is $109 |
| 20:30.28 | prpplague | Crofton: except for random searches |
| 20:31.09 | davidc__ | prpplague: dunno |
| 20:31.22 | prpplague | hehe |
| 20:31.44 | landley | prpplague: I generally fly "book rate". |
| 20:32.19 | landley | My general approach to bureaucrats is to slowly, quietly, and politely take up as much of their time as humanly possible. |
| 20:32.29 | landley | They generally figure this out after the first hour. |
| 20:32.39 | landley | "I can wait." |
| 20:42.34 | SpeedEvil | prpplague: was it booting to text? (the laptop) |
| 20:43.01 | prpplague | SpeedEvil: iirc it was redhat7 |
| 20:43.12 | prpplague | SpeedEvil: up to gnome desktop |
| 20:43.41 | SpeedEvil | prpplague: sigh |
| 20:45.02 | prpplague | ~quote scoxq.pk |
| 20:45.42 | SpeedEvil | they say they're delisting - when does it hapen? |
| 20:46.22 | prpplague | SpeedEvil: already happen hence th scoxq.pk rather than scox |
| 20:46.57 | SpeedEvil | ah |
| 20:47.40 | prpplague | what amazes me is that is still trading going on |
| 20:47.53 | prpplague | which means someone is buying the stock |
| 20:48.51 | adyer | is that volume in 1's or 1000's? |
| 20:48.54 | SpeedEvil | how many shares are there? |
| 20:49.06 | SpeedEvil | IOW - what's the total value of the company according to that |
| 20:50.29 | chouimat|work | not much |
| 20:50.39 | CosmicPenguin | 1.93 million |
| 20:50.50 | prpplague | http://finance.yahoo.com/q?s=scoxq.pk |
| 20:51.02 | *** join/#edev RobotGuy (n=RobotGuy@dsl093-038-120.pdx1.dsl.speakeasy.net) |
| 20:51.37 | prpplague | tad lower than 4 years ago, hehe |
| 20:52.43 | prpplague | problem i have is when it switched from scox to scoxq.pk, all of the trading history is gone |
| 20:53.05 | chouimat|work | what .pk mean? |
| 20:53.12 | chouimat|work | or q..pk |
| 20:53.45 | prpplague | chouimat|work: short for "pink" |
| 20:53.58 | chouimat|work | ? |
| 20:54.14 | adyer | http://en.wikipedia.org/wiki/Pink_Sheets |
| 20:54.19 | prpplague | chouimat|work: stock slang for lower value stock |
| 21:25.58 | *** join/#edev lyakh (n=lyakh@p57BD2812.dip0.t-ipconnect.de) |
| 21:28.46 | prpplague | ho ho hum |
| 21:33.20 | *** join/#edev bbradley (n=bbradley@5ac80892.bb.sky.com) |
| 21:41.23 | *** join/#edev samuelr (n=samuelr@dsl093-061-094.pit1.dsl.speakeasy.net) |
| 21:42.35 | prpplague | ~lart samuelr for being a slacker |
| 21:42.35 | purl | runs at samuelr with an origami Swiss Army knife, and inflicts a nasty paper cut for being a slacker |
| 21:51.46 | *** join/#edev roxfan (n=dunno@144.2-200-80.adsl-dyn.isp.belgacom.be) |
| 21:55.35 | Mirell | I just had to do work. |
| 21:55.39 | Mirell | I feel dirty. |
| 21:57.35 | prpplague | hehe |
| 21:57.58 | samuelr | "Huh? Wha...?" |
| 21:58.26 | prpplague | samuelr: just joking with you |
| 22:00.28 | Mirell | Man, I just begin to zone out at 4:00 |
| 22:01.07 | *** join/#edev CosmicPenguin (n=nobody@163.181.251.103) |
| 22:01.08 | *** mode/#edev [+o CosmicPenguin] by ChanServ |
| 22:01.24 | hapchap | prpplague: have i ever mentioned how much i hate dbus? |
| 22:01.36 | prpplague | hapchap: earlier this morning |
| 22:01.46 | hapchap | prpplague: ahhh... good then... |
| 22:03.57 | *** join/#edev samuelr (n=samuelr@dsl093-061-094.pit1.dsl.speakeasy.net) |
| 22:04.57 | samuelr | prpplague: Just joking? Just JOKING? You lart'ed me clear off the channel! Made my computer reboot, and caused my hair to turn pure white in fear! |
| 22:05.08 | prpplague | hehe |
| 22:05.19 | *** join/#edev dijenerate (n=dijenera@72.51.111.182) |
| 22:05.33 | samuelr | You must pledge to always use this awesome power for good, and never for evil... unless you get bored. |
| 22:05.43 | prpplague | indeed |
| 22:06.22 | chouimat|work | hapchap: dbus is evil ... like all the ipc created in the last 30 years |
| 22:11.09 | *** join/#edev landley (n=landley@adsl-76-197-33-46.dsl.austtx.sbcglobal.net) |
| 22:14.10 | landley | prpplague: not lower value, specifically "not traded on a major exchange". |
| 22:14.14 | landley | Also know as OTC (over the counter) stocks. |
| 22:14.28 | hapchap | chouimat|away: hehe... i agree! |
| 22:14.41 | landley | It could be a small company that's new and local, or it could be a company that got itself delisted. |
| 22:14.52 | prpplague | landley: yea i mispoke |
| 22:15.30 | *** join/#edev RobotGuy (n=RobotGuy@dsl093-038-120.pdx1.dsl.speakeasy.net) |
| 22:17.00 | RobotGuy | Hammer Time is anytime. |
| 22:17.18 | landley | chouimat|work: there's a reason the hammer documentation I'm writing is in a "thyme.html" file... |
| 22:17.33 | chouimat|work | ;) |
| 22:17.40 | RobotGuy | Hammer Cooks! |
| 22:18.15 | chouimat|work | http://en.wikipedia.org/wiki/Hammertime |
| 22:18.27 | RobotGuy | I want to eventually put a hammer on my wheel chair with a camera, and small keyboard with display. |
| 22:19.23 | prpplague | RobotGuy: what size display? |
| 22:22.43 | RobotGuy | I'd really like to use http://www.sparkfun.com/commerce/product_info.php?products_id=8335 but will probably only be able to use something like http://www.sparkfun.com/commerce/product_info.php?products_id=8490 |
| 22:24.58 | prpplague | RobotGuy: actually neither are support on the hammer |
| 22:25.36 | RobotGuy | I'd have to have a separate display controller then. I have to be able to see crisp color graphics. |
| 22:26.16 | RobotGuy | It has to show nice images from the camera. |
| 22:27.08 | prpplague | RobotGuy: you may want to wait for the sledge hammer to do taht |
| 22:27.35 | RobotGuy | Camera and a strong LED flashlight would be on a pan/tilt on a short pole mounted to my headrest. |
| 22:28.26 | RobotGuy | This is not something I can work on right now, but IS something I DEFINITELY want to do. It's a safey add-on for my wheel chair and will also be a lighting controller for lights I will add. |
| 22:28.47 | prpplague | RobotGuy: yea, probably wait till the sledge hammer comes out |
| 22:29.11 | prpplague | RobotGuy: hehe, keep up the good work on the hammer dev and i'm sure you'll be on the beta for the sledgehammer |
| 22:29.16 | RobotGuy | prpplague, Waiting won't be a problem. |
| 22:29.30 | RobotGuy | prpplague, I am doing my best to help. |
| 22:29.48 | landley | Do any of the things like http://www.usbgear.com/computer_cable_details.cfm?sku=USBG-SVGA2&cats=169&catid=125%2C106%2C169 work on Linux yet? |
| 22:30.35 | prpplague | landley: i'm not sure |
| 22:31.00 | landley | There are 2 or three different manufacturers of 'em, but they never claim to have linux drivers... |
| 22:31.09 | cajaka | One usb 2 vga works with the NSLU2 aka nslug |
| 22:31.15 | cajaka | Let me look |
| 22:31.25 | RobotGuy | Hmmmm, interesting. |
| 22:31.42 | RobotGuy | prpplague, Anything new on buildroot? |
| 22:32.37 | prpplague | landley: http://www.qbik.ch/usb/devices/showdev.php?id=3405 |
| 22:32.46 | prpplague | RobotGuy: no why? don't tell me you are still having troubles |
| 22:32.54 | prpplague | RobotGuy: i thought you got yours to build |
| 22:33.05 | cajaka | Here is the nslug page for them http://www.nslu2-linux.org/wiki/HowTo/AddVGAAdapter |
| 22:33.45 | RobotGuy | prpplague, Yes, still. :( etc is still not getting populated properly. |
| 22:34.11 | prpplague | RobotGuy: you did a make dirclean and a make? |
| 22:34.33 | RobotGuy | prpplague, I will try removing root and that .root and try again. Yes, I did make dirclean and then make. |
| 22:35.28 | prpplague | RobotGuy: remove root and the .root , also remove the toolchain_arm/gcc/.libs_installed |
| 22:35.38 | landley | prpplague: the 2005 date on that makes me a touch nervous. |
| 22:35.49 | prpplague | indeed |
| 22:36.04 | landley | Yeah, the linux-usb link it points to is down. |
| 22:36.07 | landley | Some ad page now. |
| 22:36.21 | prpplague | just wish they worked with usb1.1 |
| 22:37.25 | landley | http://linux.die.net/man/4/sisusb |
| 22:37.31 | landley | prpplague: nothing's going to work with usb1. |
| 22:37.33 | landley | It's too slow. |
| 22:37.38 | prpplague | landley: yea i know |
| 22:37.38 | landley | Does nail only provide usb1? |
| 22:37.53 | prpplague | landley: yea the s3c2410 is only 1.1 |
| 22:37.57 | landley | Ouch. |
| 22:38.02 | landley | Ok, that's a bit red warning to put in the docs. |
| 22:38.15 | landley | "You won't be able to plug anything faster than a mouse into the downstream USB port, it's really there for show". |
| 22:38.18 | RobotGuy | drwxr-xr-x 7 linuxguy staff 816 2008-01-07 18:28 fakeroot-1.7.1-host |
| 22:38.18 | RobotGuy | drwxr-xr-x 24 linuxguy staff 1848 2008-01-06 19:39 gcc-4.1.2 |
| 22:38.18 | RobotGuy | drwxr-xr-x 9 linuxguy staff 568 2008-01-07 18:25 gcc-4.1.2-final |
| 22:38.18 | RobotGuy | drwxr-xr-x 8 linuxguy staff 440 2008-01-06 19:43 gcc-4.1.2-initial |
| 22:38.25 | RobotGuy | prpplague, Which GCC? |
| 22:38.29 | prpplague | landley: thats not totally correct |
| 22:38.43 | prpplague | RobotGuy: gcc-4.1.2.final |
| 22:38.45 | landley | I know, but I'm trying to think what the last USB1 device I bought was. |
| 22:38.49 | RobotGuy | prpplague, OK |
| 22:38.53 | SpeedEvil | landley: mouse |
| 22:38.56 | landley | Yeah, pretty much. |
| 22:39.10 | landley | 100baseT is faster than USB 1. |
| 22:39.27 | prpplague | landley: think you are confusing usb2 highspeed with usb in general |
| 22:39.38 | SpeedEvil | but 100baset on usb1 is good |
| 22:39.38 | adyer | you know that a lot of 'usb 2.0' device are still at 12mbit/s not 480 mbit/sec, right? |
| 22:39.38 | SpeedEvil | as you don't saturate the bus |
| 22:39.39 | RobotGuy | prpplague, Building now. |
| 22:39.45 | SpeedEvil | though with switched networks, that's not too bad. |
| 22:40.05 | landley | Possibly I'm holding it against USB1 in general. |
| 22:40.11 | SpeedEvil | I've got some 1.5Mhz |
| 22:40.19 | SpeedEvil | mbps |
| 22:40.20 | prpplague | landley: just because its usb2 doesn't mean that it is a high speed device |
| 22:41.03 | prpplague | landley: usb2 can operate at 2mb,12mb and 480mb |
| 22:41.15 | landley | usb1.1 is 12 mbps, slightly faster than 10baseT. |
| 22:41.33 | landley | This means that sending a cd-rom image through it can be expected to take about 10 minutes. |
| 22:41.34 | prpplague | landley: correct, usb1.1 is limited to 12mbps |
| 22:42.05 | prpplague | landley: correct, but just because a usb device says it's a usb2 device doesn't mean it won't work on a usb1.1 controller |
| 22:42.15 | RobotGuy | I have an AVR question, if I may.. How can I tell if http://www.sparkfun.com/commerce/product_info.php?products_id=12 will work under Linux? |
| 22:42.39 | landley | So a video adapter is out, considering that updating the screen once at 1024x768x24bit is over 2 megabytes. |
| 22:42.52 | prpplague | RobotGuy: if your avr is just a 3.3v device, jsut use the flyswatter |
| 22:43.16 | adyer | landley depends if the vga adapter has a local frame store or not |
| 22:43.16 | landley | Many devices will downshift, sure. |
| 22:43.21 | RobotGuy | prpplague, Ah, OK. I want to use Atmega168 and Atmega128 |
| 22:43.23 | prpplague | landley: correct, in the case of the vga adapter, the 12mbps limitation would be a problem |
| 22:43.38 | prpplague | RobotGuy: if they are 3.3v parts you can use the flyswatter |
| 22:43.45 | landley | I've had print jobs blow up to 400 mb for ~30 pages. |
| 22:43.49 | RobotGuy | prpplague, OK, I will check, thanks. |
| 22:44.03 | prpplague | landley: indeed |
| 22:44.08 | landley | Doing that through USB1.1 it printed about a page a minute. |
| 22:44.31 | landley | The printer was capable of at least 10x that. |
| 22:44.49 | landley | It's definitely something to mention in the documentation. |
| 22:45.05 | RobotGuy | prpplague, etc is not being populated correctly. |
| 22:45.17 | prpplague | landley: you have to remember, the s3c2410 and similiar SoC's were designed to be used with mobil devices |
| 22:45.23 | landley | Sure. |
| 22:45.28 | landley | Understood, it just needs to be documented. |
| 22:45.35 | landley | It's a bit disconcerting to have this sprung on you without warning. |
| 22:45.46 | prpplague | RobotGuy: got time to start a fresh buildroot? |
| 22:45.55 | landley | The reaction I just had probably wasn't unique to me. |
| 22:46.05 | landley | just be up front about it. |
| 22:46.23 | landley | USB 1.1 is lower power, smaller circuitry. Check. |
| 22:46.40 | prpplague | landley: yea i just kinda assumed you realized that |
| 22:46.48 | landley | Nope. |
| 22:46.53 | prpplague | landley: virtually none of the mobile phones or other pda's have usb2 |
| 22:47.03 | RobotGuy | prpplague, Plenty of time. Meeting pickup is not for 3 hours |
| 22:47.09 | prpplague | landley: its only now showing up in some the cutting edge devices |
| 22:47.18 | prpplague | RobotGuy: delete your buildroot dir |
| 22:48.04 | RobotGuy | prpplague, Done. Got a new CD image for me to try? |
| 22:48.12 | landley | It just means that the 500mb external drive I could plug into this thing would be accessed more slowly than an ATA/33 system from 1998. |
| 22:48.16 | prpplague | RobotGuy: yea one sec |
| 22:48.36 | prpplague | landley: indeed |
| 22:49.03 | prpplague | landley: if the nail looks promising, there is no reason we couldn't shove a usb2 controller inside |
| 22:49.08 | landley | Same for plugging in a USB flash key. Enabling swap on an external USB drive would thus be a Bad Idea (tm). |
| 22:49.18 | landley | Hmmm... |
| 22:49.27 | landley | There's an "upstream" controller and a "downstream" controller. |
| 22:49.34 | landley | I wonder... |
| 22:50.02 | landley | Having the external USB be 2.0 has some distinct advantages. |
| 22:50.32 | landley | USB video adapter, USB drive you could build from or swap to rapidly, 100baseT network adapter you could serve web pages to. |
| 22:50.42 | SpeedEvil | ;3~54g |
| 22:50.44 | SpeedEvil | 54g |
| 22:50.53 | landley | SpeedEvil: sure. |
| 22:51.07 | adyer | almost all external usb 2.0 controllers require PCI |
| 22:51.21 | prpplague | adyer: not true |
| 22:51.26 | landley | Having the _upstream_ adapter be USB2 is nice, but if the device only has 16MB built in storage I'm not sure it's as much of an improvement. |
| 22:51.37 | RobotGuy | Vinculum does not need PCI. I don't know how good it is though. |
| 22:51.45 | landley | On the other hand, USB 1.1 is something you can certainly _live_ with. |
| 22:51.47 | adyer | i said almost |
| 22:51.51 | prpplague | hehe |
| 22:51.51 | SpeedEvil | Is it USB2, or USB2 compatible? |
| 22:52.20 | SpeedEvil | <PROTECTED> |
| 22:52.27 | SpeedEvil | full is 12? |
| 22:52.33 | SpeedEvil | (vinculum) |
| 22:52.36 | adyer | yes, high is 480 |
| 22:53.10 | adyer | god i hate their terminology... |
| 22:54.25 | SpeedEvil | I suppose in principle you could hang something off the memory bus. |
| 22:54.31 | SpeedEvil | But that's probably a mistake. |
| 22:54.31 | prpplague | RobotGuy: http://www.elinux.org/images/b/b6/Buildroot-01082008.tar.gz |
| 22:54.46 | RobotGuy | I very much want to add WiFi G to Hammer for W.A.L.T.E.R. |
| 22:54.52 | SpeedEvil | you can. |
| 22:54.59 | adyer | something like nxp ixp1761 would work |
| 22:55.02 | SpeedEvil | just not at full rate. |
| 22:55.26 | prpplague | adyer: yea, i have protoboard with the nxp ixp1761 connected to the s3c2410 |
| 22:56.07 | prpplague | RobotGuy: download that |
| 22:56.10 | RobotGuy | I am thinking more along the lines of a Lantronix module. |
| 22:56.17 | RobotGuy | prpplague, Am dling. |
| 22:56.45 | landley | prpplague: you mentioned that internally, the nail has a 4 port hub? |
| 22:56.52 | prpplague | landley: correct |
| 22:57.09 | prpplague | landley: there are two ports unused |
| 22:57.14 | landley | Is the "outbound" usb part of that? |
| 22:57.39 | landley | The female port, I guess. |
| 22:57.46 | prpplague | landley: slave and host |
| 22:57.56 | landley | oh, great. |
| 22:57.57 | prpplague | landley: the slave port on the hammer is connected to the hub |
| 22:58.20 | landley | So the female port is the one that we're the host of, and the male port is the one we're the slave of? |
| 22:58.21 | prpplague | landley: the host port for the hammer is the host port on the opposit end |
| 22:58.27 | RobotGuy | prpplague, Copied Hammer-config to .config, make oldconfig, make |
| 22:58.30 | landley | Or is the male port the one that plugs into a host and the female one that plugs into a slave? |
| 22:58.35 | prpplague | RobotGuy: yep |
| 22:58.51 | RobotGuy | prpplague, I don't need to look it up anymore. :) |
| 22:59.01 | landley | prpplague: the host port for the hammer is the host port on the opposite end? |
| 22:59.12 | prpplague | landley: correct |
| 22:59.26 | landley | Does that mean the host port for the hammer is the host port on the laptop, or the host port for the hammer is on the opposite end from plugging into the laptop? |
| 22:59.27 | *** join/#edev Dr{Wh0} (n=mathewss@dev.null.nutech.com) |
| 22:59.40 | prpplague | landley: lets look at the layout on the page - http://www.elinux.org/images/b/b6/Buildroot-01082008.tar.gz |
| 22:59.49 | RobotGuy | prpplague, So USB devices plug into the device port on Hammer? |
| 22:59.49 | prpplague | landley: http://www.elinux.org/Nail_Board |
| 22:59.52 | landley | That's a tarball. |
| 23:00.00 | prpplague | landley: yea sorry |
| 23:00.16 | SpeedEvil | technically both are host ports. one is a device port too. The kernel configures one as a host. |
| 23:00.19 | prpplague | landley: see J1 on the layout? |
| 23:00.31 | landley | Yeah. That's the "pen cap" end with the female port? |
| 23:00.51 | prpplague | landley: scroll down some, there is a layout picture |
| 23:01.00 | landley | I see the layout picture. |
| 23:01.04 | landley | J1 is a rectangle, j3 is a rectangle. |
| 23:01.09 | landley | I have no idea which end plugs into the laptop. |
| 23:01.18 | prpplague | landley: ok J1 is the one that connects to the laptop |
| 23:01.24 | landley | Ok. |
| 23:01.34 | prpplague | landley: that is considered the slave port |
| 23:01.52 | landley | The one that plugs into the laptop is the slave port. |
| 23:01.54 | prpplague | landley: J1 is connected to the usb 4 port hub which has 4 host ports |
| 23:01.59 | landley | Ok, "this is the port we are the slave of". |
| 23:02.00 | prpplague | landley: correct |
| 23:02.06 | landley | Not "this is the port we connect to a slave through". |
| 23:02.12 | prpplague | landley: correct |
| 23:02.15 | landley | (This terminology is inherently vague.) |
| 23:02.35 | prpplague | landley: indeed |
| 23:02.40 | landley | I think I'm calling them the "laptop" and "peripheral" ports. |
| 23:03.10 | Mirell | I should theoritically be able to escape here shortly. |
| 23:03.10 | prpplague | landley: that isn't very accurate, because there are multiple hosts and multiple slaves |
| 23:03.13 | landley | Must go whip up a fresh batch... |
| 23:03.23 | landley | Mirell: congrats. |
| 23:03.37 | landley | prpplague: per nail board? |
| 23:03.41 | prpplague | landley: correct |
| 23:04.04 | prpplague | landley: lets start with just a laptop and a 4 port hub, no nail board involved |
| 23:04.16 | landley | ok. |
| 23:04.16 | prpplague | landley: your laptop has a host port |
| 23:04.25 | prpplague | landley: the 4 port hub has one slave port |
| 23:04.38 | landley | My original question was "do both ends of the thing have to be USB2 if you go to USB2, and how much difference in cost/power consumption is it, anyway?" |
| 23:04.43 | prpplague | landley: the slave port on the hub connects to the host port on your laptop |
| 23:04.46 | landley | The hub is purely internal to the nail board. |
| 23:04.52 | SpeedEvil | landley: yes - to get the speed |
| 23:05.04 | SpeedEvil | landley: the ports are electrically downwards compatible. |
| 23:05.08 | prpplague | SpeedEvil: no |
| 23:05.15 | SpeedEvil | landley: but some USB2 devices are only USB2 |
| 23:05.23 | SpeedEvil | landley: they won't work in 'full' speed mode. |
| 23:05.41 | prpplague | SpeedEvil: he's talking about both ends of the nail board |
| 23:05.42 | RobotGuy | I'll bbiab. |
| 23:05.45 | SpeedEvil | landley: some dvb-t sticks, some cameras, and some wifi devices are examples of this. |
| 23:05.45 | landley | You mean they only work in full speed mode? |
| 23:06.08 | SpeedEvil | landley: some devices, yes - if you are not talking of the nail board. |
| 23:06.15 | landley | ok, backing up. |
| 23:06.19 | landley | 2 separate questions here. |
| 23:06.55 | landley | 1) I'm trying to come up with unambiguous terminology to use in the documentation to refer to the two different USB ports the nail board provides to the outside world. |
| 23:07.07 | landley | The one you plug into your laptop, and the other one at the other end of the device. |
| 23:07.15 | SpeedEvil | host and device is popular |
| 23:07.24 | landley | This thing is both a host and a device. |
| 23:07.25 | flyback- | http://www.microsoft.com/technet/security/bulletin/ms08-001.mspx <-- all you windows users.... here we go again!!! YOU GO TO HELL AND YOU DIE! |
| 23:07.30 | SpeedEvil | you plug a device (a printer, or ...) into a host |
| 23:07.36 | landley | Are we talking about "where we plug into a host" or "where we are a host". |
| 23:07.57 | SpeedEvil | port 0 and port 1 |
| 23:08.28 | SpeedEvil | one is host-only, one is host/device. (though set as device in the kernel distributed with the hammer. |
| 23:08.31 | landley | Especially people who are new to embedded development, "the host port" can trivially be mistaken for "where we plug into a host, I.E. the laptop". |
| 23:08.36 | prpplague | SpeedEvil: hang on you are confusing him, because we are talking about the nail, not usb in general |
| 23:08.58 | landley | Oh I've got it straight in my head now, I just can't _document_ it using this terminology. |
| 23:09.02 | landley | Because people reading it will be confused. |
| 23:09.09 | SpeedEvil | Sorry - I was talking of my understanding of the nail. /me bes quiet. |
| 23:09.17 | adyer | i would read chapters 3 & 4 in the usb spec. |
| 23:09.19 | landley | One side effect of _me_ being easily confused is I tend to want to write very, very clear documentation. |
| 23:09.35 | prpplague | landley: np, thats why you got an eval kit |
| 23:09.42 | landley | adyer: see "Hobbyists", "new to embedded development", "dabbling". |
| 23:09.51 | landley | "Go read the USB spec" isn't helpful from a documentation perspective. |
| 23:09.56 | landley | Not user friendly. |
| 23:10.17 | prpplague | landley: the "device" port is what you will connect to your laptop's host port |
| 23:10.18 | CosmicPenguin | also, its bad for your health |
| 23:10.28 | SpeedEvil | 'into port A you can plug any USB 1.1 device, like a mouse, ... Port B pretends to be a device (or several), you plug this into a laptop or computer |
| 23:10.45 | adyer | i understand, but if you want to write docs, YOU need to understand it. I'm not saying to point a user to it |
| 23:11.57 | prpplague | landley: the host port on the nail is what you would connect other slave devices such as card readers and keyboards |
| 23:12.29 | landley | prpplague: sure, but that's not the problem. |
| 23:12.54 | prpplague | landley: what is the problem then? hehe |
| 23:12.57 | landley | The problem here is that the nail board talks to its host through the slave port, and we talks to its slave through the host port. |
| 23:13.14 | landley | There is no way to document that in such a way that people will keep it straight in their heads on the first go. |
| 23:13.24 | adyer | perhaps 'upstream hub port' might be a better term |
| 23:13.38 | landley | I was using upstream and downstream, but that's slightly arbitrary. |
| 23:13.48 | SpeedEvil | Is there a physical difference? |
| 23:13.55 | landley | Laptop and peripheral are slightly less arbitrary, but also slightly less accurate. |
| 23:13.56 | SpeedEvil | 'yellow and blue' port for example |
| 23:14.05 | landley | SpeedEvil: one's a male usb connector, one's a female usb connector. |
| 23:14.16 | landley | That people will keep straight, but it's also politically incorrect. |
| 23:14.27 | adyer | speedevil hubs do different things on their upstream and downstream ports |
| 23:14.39 | landley | The hub is an implementation detail. |
| 23:14.55 | landley | I'm talking about the external USB plugs the nail board presents to the outside world when the case is closed. |
| 23:15.05 | prpplague | landley: let me finish the functional layout drawing and post it on the wiki, then we can start labeling the blocks |
| 23:15.10 | landley | ok |
| 23:15.34 | hapchap | prpplague: did you tell me that this bluetooth cradle should auto-sense baud? |
| 23:16.21 | SpeedEvil | 'You will notice two different USB connectors. Take a mouse and plug it in. It fits only one connector. To this connector you can connect any USB 1.1 device that linux supports and that you have drivers for. The other port however does the conjouring trick of pretending to be a device. You enable this by loading the right driver (list), and plugging it into your laptop or computer as you would a mouse or memory stick. It's then detected as a memory |
| 23:16.28 | prpplague | hapchap: no, its hardcoded to 9600 |
| 23:16.51 | prpplague | hapchap: on the bluetooth side |
| 23:17.10 | landley | SpeedEvil: http://landley.net/thyme.html |
| 23:17.15 | landley | Already past that point. :) |
| 23:17.19 | landley | I just need names for 'em. |
| 23:17.27 | SpeedEvil | Innie and outie? |
| 23:17.35 | landley | I can have a paragraph clarifying that I'm calling them "larry" and "charles". |
| 23:17.52 | landley | The problem is, I don't want people to have to refer back to that pargraph multiple times while reading later paragraphs. |
| 23:17.52 | CosmicPenguin | is it plugged into your larry port? |
| 23:18.59 | adyer | a tree diagram might be a good way to show what you're doing |
| 23:18.59 | prpplague | hehe |
| 23:19.43 | SpeedEvil | Device and hub port? |
| 23:19.50 | SpeedEvil | After all, it is a hub - a root hub |
| 23:21.31 | SpeedEvil | Does it matter if they get the names confused? |
| 23:21.49 | SpeedEvil | all they need to work out is that into one you plug a mouse, and the otehr can pretend to be a mouse. |
| 23:27.55 | landley | SpeedEvil: people know what a hub is from the networking world. |
| 23:28.04 | landley | Nail has exactly 2 ports. That ain't a hub. |
| 23:28.15 | landley | The internal hub is, as I said, an implementation detail. |
| 23:28.48 | landley | SpeedEvil: they have to be able to keep them straight while reading the documentation. |
| 23:32.12 | SpeedEvil | If I was writing it, I'd probably be using constructions that reminded them of the function of the port each time I used the name. |
| 23:33.39 | landley | SpeedEvil: hence "laptop" and "device". |
| 23:33.52 | landley | Or male and female. |
| 23:34.10 | SpeedEvil | 'Load the module gadget_mass with the options 'fkfkf' and the device port will behave like a mass storage device, sharing the NAND flash of the hammer' |
| 23:34.10 | landley | Upstream and downstream don't do this, but are also unlikely to confuse people as to which is which. |
| 23:34.18 | landley | Although that could be a US cultural thing and not translate well overseas. |
| 23:34.42 | SpeedEvil | upstream and downstream are unclear. |
| 23:34.43 | landley | Except that the it won't. That's the other port. :) |
| 23:34.47 | SpeedEvil | which way does USB 'flow' |
| 23:35.05 | landley | SpeedEvil: people are used to "upload" and "download". But yeah, not my first choice. |
| 23:35.18 | adyer | in USB the root host is pretty much all powerful |
| 23:35.50 | SpeedEvil | I think your best plan is a big diagram at the top, with a laptop, with a mouse plugged into it, and a label 'host' 'device'. |
| 23:36.02 | landley | adyer: the two USB ports are logically disconnected. |
| 23:36.09 | landley | (This may not be how it's implemented, but it's how it behaves.) |
| 23:36.27 | SpeedEvil | landley: they are pretty much completely seperate apart from clocks. |
| 23:36.31 | landley | The one that plugs into the laptop and the one that plugs into an external device are like two ethernet cards in different PCI slots. |
| 23:36.32 | SpeedEvil | (in implementations) |
| 23:36.53 | landley | So saying "the host is all powerful"... the nail board hosts one. The laptop hosts the other. |
| 23:37.03 | landley | Each port is connected to a host. |
| 23:37.19 | adyer | I was answering speedevil's question about usb flow. |
| 23:37.43 | SpeedEvil | adyer: yes - that doesn't clarify which is upstream or downstream though |
| 23:38.04 | SpeedEvil | adyer: I understand how this works - I was trying and failing to find good terminology. |
| 23:38.51 | adyer | speedevil: the usb root hub is upstream of all |
| 23:39.01 | landley | Vote away on what you prefer, I'll be off banging on sed in toybox some more. |
| 23:40.11 | SpeedEvil | good luck. |
| 23:41.09 | prpplague | landley: i'll have the block diagram up in the morning |
| 23:41.18 | prpplague | landley: they are closing the office, so i gotta get out of here |
| 23:41.19 | prpplague | bbiab |