| 00:18.33 | *** join/#edev Ian_Daniher (n=it@69.61.230.246) |
| 00:24.20 | *** join/#edev dijenerate (n=dijenera@72.22.144.246) |
| 02:07.05 | *** join/#edev chouimat (n=dieu@kde/developer/chouinard) |
| 02:13.55 | *** join/#edev Ian_Daniher (n=it@69.61.230.246) |
| 03:12.42 | *** join/#edev notadev (n=notadev@fu/coder/notadev) |
| 03:14.59 | notadev | wow that nail kit looks nice |
| 03:20.13 | RobotGuy | Nail is awesome for developing. Easy to pack up and take with you. |
| 03:23.13 | RobotGuy | If you have a Linux laptop or such you can take your development on the go with Nail. |
| 03:26.11 | notadev | yeah; the hammer kit looks nice too. guess you need a cover for the nail thing tho? |
| 03:26.36 | notadev | i know it's lame ;) |
| 03:27.26 | RobotGuy | Cover for Nail? |
| 03:28.37 | RobotGuy | No, you don't need the case if that is what you mean. |
| 03:28.39 | RobotGuy | I have a Hammer Kit and a couple Nails |
| 03:28.40 | *** join/#edev DrZimmerman (n=theo@77-56-242-58.dclient.hispeed.ch) |
| 03:29.59 | notadev | ah ok cool. hmm need UK shipment, but would be much sweeter to work with than slug |
| 03:30.50 | RobotGuy | Kind of difficult to pack a slug around. You can put Nail in your pocket or even in your laptop case. |
| 03:31.07 | RobotGuy | Portable Hammer development. :) |
| 03:31.10 | solar | it's so much more than a slug |
| 03:31.51 | notadev | yeah portability would be so nice |
| 03:32.26 | RobotGuy | Hammer is so awesome. :) I could be a bit biased though. :) |
| 03:32.53 | notadev | hehe s'ok evangelism is good, in limited doses ;) |
| 03:33.14 | RobotGuy | Everything is 0.1" pin spacing, ready for a solderless breadboard. I'm going to put pin headers on the underside of one of my Nails for breadboard use. |
| 03:33.46 | RobotGuy | I don't say anything that isn't true for me though. :) |
| 03:33.53 | solar | I would avoid that |
| 03:34.03 | RobotGuy | solar: Why? |
| 03:34.06 | notadev | nice idea. hmm bit early for hw bod to be awake |
| 03:34.58 | RobotGuy | I need to get something to eat, but don't have much here right now. :( |
| 03:35.30 | solar | cuz you are going to lose quite a bit of room on the breadboard with the nail in the center of it. What would be better is 2 * 1 x 20@0.1 female headers |
| 03:36.06 | solar | then use jumpers between the breadboard while keeping the ability to shove it back in it's case when you are done prototyping |
| 03:36.27 | notadev | heh |
| 03:36.36 | notadev | thanks fsck he's software |
| 03:37.15 | notadev | always nicer when it's someone else's problem ;) |
| 03:39.54 | MonMotha | screw software :) |
| 03:40.06 | MonMotha | I do hardware |
| 03:40.09 | MonMotha | always nicer when its someone else's problem :) |
| 03:40.16 | notadev | hehe |
| 03:40.44 | MonMotha | I find that the hardware is almost always simpler (in my opinion) than the software |
| 03:41.04 | MonMotha | The software can do "anything". There's only a certain number of ways you can hook up a given set of hardware and have it be useful. |
| 03:41.17 | notadev | true |
| 03:41.25 | MonMotha | and there's inherent scope limiting |
| 03:41.34 | MonMotha | adding gee-whiz hardware features costs $$$ per unit |
| 03:41.55 | MonMotha | sure, I can give you tilt sensitivity. The accelerometer costs $X. |
| 03:42.46 | notadev | i didn't learn to solder properly til I was 25, so not comfy with hw. asm is another matter, especially on ARM |
| 03:42.47 | MonMotha | what's that? You want double the RAM? The extra chip cost $Y and takes up Z board space, resulting in a cost of $W. |
| 03:43.03 | MonMotha | oh, I can handle software, I just prefer not to |
| 03:43.07 | notadev | heh yeah suits don't like that, especially not if it means any difference in tooling |
| 03:43.08 | *** join/#edev RobotGuy (n=robotguy@dsl093-038-072.pdx1.dsl.speakeasy.net) |
| 03:43.35 | MonMotha | I've written plenty of Linux kernel drivers, and I do know the ARM instruction set well enough to be useful if I have my quickref card. |
| 03:43.58 | MonMotha | I know 8051 fairly well (I always outoptimize the compiler), and I'm decent on AVR (I often can outoptimize the compiler if I want to). |
| 03:44.12 | MonMotha | but those are much simpler archs. 2 stage pipeline at most. |
| 03:44.29 | MonMotha | the people who can optimize for things like Netburst scare me. |
| 03:45.02 | MonMotha | 30+ stage pipeline, and they can write an inner loop that'll keep it full. The people who know the branch prediction well enough to redo their conditionals REALLY scare me. |
| 03:45.31 | MonMotha | I played with TI C6000 at one point. Got OK at it. |
| 03:45.38 | notadev | have to admit I'm outta date; ARM3 |
| 03:45.41 | MonMotha | but hardware...that's where it's at :) |
| 03:45.50 | notadev | hehe |
| 03:46.21 | notadev | well distinction was originally a marketing one. |
| 03:47.05 | MonMotha | Layout is almost more an art than a science. Yes, there's the theory behind it like all the transmission line stuff, and that's important, but coming up with routes that meet all the necessary criteria and fit in the board space allotted on the minimum spec PCB; that's art. |
| 03:47.28 | notadev | I'd argue sw is too |
| 03:47.34 | MonMotha | at times |
| 03:47.48 | MonMotha | algorithms are science, but nice APIs are something of an art |
| 03:48.04 | notadev | hmm i see that as craft vs art |
| 03:48.12 | notadev | can't have art without craft |
| 03:48.47 | notadev | well you can, but it's not what i'd call art ;) |
| 03:49.42 | MonMotha | Well, the science is what you pay people to have the degree in. The "art" or "craft" is what you pay people for experience with. |
| 03:50.45 | MonMotha | obviously you can trade those around, some, and after a while, the science changes sufficiently from when a degree was awarded that experience also carries knowledge of the modern "science" (in this case, computer science or EE), but you get the idea |
| 03:50.49 | notadev | hmm s'pose, though I'd see the science there as part of the craft; guess it's a more generic thing in that sense |
| 03:51.07 | notadev | yeah :) |
| 03:51.27 | MonMotha | what's a bummer for many CS people is that real-world software has remarkably little computer science in it |
| 03:51.40 | MonMotha | most real-world software is all interface, "business logic" and a butt-ton of glue |
| 03:52.13 | notadev | yeah that's true, though I think understanding say Context Free Grammars (CS101) would do many "developers" the world of good |
| 03:52.34 | MonMotha | I've forgotten so much of my CS :) |
| 03:52.40 | MonMotha | I only took a couple real CS classes |
| 03:52.44 | notadev | heh Chomsky man |
| 03:52.51 | MonMotha | I rarely use it |
| 03:53.03 | MonMotha | most of the stuff I do is all glue: make a device talk a standard API |
| 03:53.17 | notadev | yeah I took a year of EEng but I never use it |
| 03:53.24 | MonMotha | or make some highly embedded device do somehting, which rarely involves anything other than reacting to events and shoveling some data around |
| 03:53.36 | *** join/#edev chouimat (n=dieu@kde/developer/chouinard) |
| 03:53.42 | *** join/#edev RobotGuy (n=robotguy@dsl093-038-072.pdx1.dsl.speakeasy.net) |
| 03:54.12 | notadev | heh |
| 03:54.33 | MonMotha | I've crashed the Linux kernel plenty of times, though :) |
| 03:55.45 | MonMotha | most common error seems to be summarized by "*(void *)0;" though :) |
| 03:56.17 | MonMotha | who knew that dereferencing NULL doesn't usually work right :) |
| 03:56.29 | notadev | hehe |
| 05:03.21 | *** join/#edev mbuf (n=user@61.16.248.242) |
| 06:40.26 | *** join/#edev likewise (n=likewise@82-171-51-231.ip.telfort.nl) |
| 07:10.57 | HcE | Chuck Norris can dereference a NULL pointer ;) |
| 07:13.12 | *** join/#edev [AD]Turbo (n=turbo@host45-243-dynamic.12-79-r.retail.telecomitalia.it) |
| 07:14.57 | [AD]Turbo | yo |
| 07:20.42 | *** join/#edev chouimat (n=dieu@r2351064.cidc.net) |
| 07:36.08 | *** join/#edev hw_ (n=hw@p578b3905.dip0.t-ipconnect.de) |
| 07:45.35 | *** join/#edev magnetron (n=magnetro@unaffiliated/magnetron) |
| 07:56.18 | RobotGuy | Chuck Norris can do pretty much anything he wants. ;) |
| 08:10.43 | *** join/#edev lyakh (n=lyakh@p57BD2629.dip0.t-ipconnect.de) |
| 08:13.52 | *** join/#edev chouimat (n=dieu@kde/developer/chouinard) |
| 08:42.12 | *** join/#edev harpal (n=Harpal@122.169.108.195) |
| 10:10.51 | *** join/#edev chouimat (n=dieu@kde/developer/chouinard) |
| 10:39.58 | *** join/#edev g1powermac_PB (n=g1powerm@unaffiliated/g1powermac) |
| 10:48.16 | *** join/#edev g1powermac (n=g1powerm@unaffiliated/g1powermac) |
| 11:11.04 | *** join/#edev g1powermac_PB (n=g1powerm@unaffiliated/g1powermac) |
| 11:42.19 | *** join/#edev dijenerate (n=dijenera@72.22.144.246) |
| 11:47.19 | *** join/#edev g1powermac_PB (n=g1powerm@unaffiliated/g1powermac) |
| 13:03.49 | *** join/#edev flavioribeiro (n=flaviori@201.70.21.178) |
| 13:10.52 | *** join/#edev harpal_ (n=Harpal@122.169.108.195) |
| 13:25.00 | *** join/#edev prpplague (n=dave@mail.americanmicrosystems.com) |
| 13:47.32 | *** join/#edev GPSFan (n=kenm@12.10.255.246) |
| 13:47.32 | *** mode/#edev [+o GPSFan] by ChanServ |
| 13:50.33 | prpplague | ho ho ho, merry freakin friday |
| 13:50.43 | *** join/#edev Chocobo (n=swinchen@strongbad.eece.maine.edu) |
| 13:50.50 | RobotGuy | Greetings Program |
| 13:51.22 | Chocobo | Does anyone know the format of -O binary when using objcopy? |
| 13:51.57 | *** join/#edev [AD]Turbo (n=turbo@host45-243-dynamic.12-79-r.retail.telecomitalia.it) |
| 13:52.01 | HcE | binary code? |
| 13:53.28 | Chocobo | So no address information is encoded in it? |
| 13:53.37 | Chocobo | I guess that makes sense really |
| 13:53.51 | HcE | do not think so |
| 13:54.05 | HcE | IIRC -O binary produces an output file with machine code |
| 13:55.40 | Chocobo | Ahhhh! I see why looking at the machine code is so confusing... I is amazing how different things look when you aren't paying attention to endian. |
| 13:59.02 | *** join/#edev chouimat|work (n=dieu@kde/developer/chouinard) |
| 14:42.20 | RobotGuy | prpplague: I am on the road to being able to build stuff for Hammer/Nail again. |
| 14:43.23 | RobotGuy | can't believe he's been up all night. |
| 15:04.53 | *** join/#edev Ian_Daniher (n=it@204.10.219.119) |
| 15:06.18 | prpplague | RobotGuy: good deal |
| 15:07.28 | RobotGuy | prpplague: I think so! I am getting back in the swing of Gentoo Embedded now, thanks to help from solar and torindel. |
| 15:20.02 | gandhijee | hey any of you guys mess around with CANbus any? |
| 15:21.55 | MothaWork | I've used it once |
| 15:22.03 | MothaWork | can't say I would do it again voluntarily |
| 15:29.23 | gandhijee | what did you use it for? |
| 15:29.43 | chouimat|work | prpplague: http://failblog.files.wordpress.com/2008/09/fail-owned-clean-energy-fail.jpg |
| 15:29.52 | gandhijee | and how does the license for the protocol work? |
| 15:30.17 | *** join/#edev linac (n=lin@121.41.56.214) |
| 15:30.29 | MothaWork | gandhijee: when I used it, it was just a "classroom exercise" situation, though we ended up with about 15 things on the ubs |
| 15:30.30 | MothaWork | *bus |
| 15:30.43 | MothaWork | as far as licensing, I wasn't aware there was any, but that doesn't mean there isn't |
| 15:30.51 | gandhijee | eh |
| 15:30.55 | chouimat|work | doesn't want to remember CANOpen |
| 15:30.58 | MothaWork | LIN is becoming more popular these days as it's simpler and just needs a UART instead of a dedicated CAN controller |
| 15:31.12 | gandhijee | you happen to know any resources for the hardware? |
| 15:31.26 | MothaWork | no. Just find a suitable micro with a built in CAN controller if you can |
| 15:31.30 | MothaWork | there aren't a bunch of standalone ones |
| 15:31.43 | gandhijee | apparently one of our customers wants a canbus controller and we need to have it on our board |
| 15:32.10 | gandhijee | are there any that can be plugged in to the PCI |
| 15:33.45 | chouimat|work | I know a few usb based one |
| 15:34.21 | chouimat|work | http://www.softing.com/home/en/industrial-automation/products/can-bus/interface-cards/can/pci-1.php |
| 15:37.13 | MothaWork | "board" with a PCI slot? That's a hell of a "board" |
| 15:39.35 | gandhijee | yeah |
| 15:39.57 | gandhijee | ATOM + 945GSE + ICH7 |
| 15:40.04 | MothaWork | so...it's a PC :) |
| 15:40.44 | gandhijee | yeah |
| 15:41.02 | gandhijee | apparently its gonna be shoved inside a bus |
| 15:41.25 | gandhijee | 11 comm ports as well |
| 15:41.45 | gandhijee | chouimat|work: thanks man |
| 15:43.55 | ds2 | morning |
| 16:05.22 | *** join/#edev harpal (n=harpal@122.169.108.195) |
| 16:16.52 | *** join/#edev CosmicPenguin (n=nobody@host-245-157.pubnet.pdx.edu) |
| 16:16.52 | *** mode/#edev [+o CosmicPenguin] by ChanServ |
| 16:26.55 | *** join/#edev lyakh (n=lyakh@p57BD2629.dip0.t-ipconnect.de) |
| 16:36.11 | *** part/#edev adye2 (n=adyer@adsl-69-223-13-78.dsl.chcgil.ameritech.net) |
| 16:59.06 | *** join/#edev flavioribeiro (n=flaviori@201.70.21.178) |
| 17:17.39 | *** join/#edev bbradley (n=bbradley@78-105-167-16.zone3.bethere.co.uk) |
| 17:29.15 | *** join/#edev EI5GTB__ (n=Paul@78.16.65.25) |
| 17:31.07 | *** join/#edev DrZimmerman (n=theo@77-56-242-58.dclient.hispeed.ch) |
| 17:31.35 | *** join/#edev flavioribeiro (n=flaviori@201.70.21.178) |
| 17:45.27 | *** join/#edev s (n=flaviori@201.70.21.178) |
| 18:02.49 | *** join/#edev Ian_Daniher (n=it@204.10.219.119) |
| 18:08.45 | *** join/#edev flavioribeiro (n=flaviori@201.70.21.178) |
| 18:22.27 | *** join/#edev s (n=flaviori@201.70.21.178) |
| 18:23.12 | *** join/#edev s (n=flaviori@201.70.21.178) |
| 18:36.52 | *** join/#edev s (n=flaviori@201.70.21.178) |
| 18:59.08 | ds2 | firefox plus googlemaps is such a big battery blackhole :( |
| 19:22.40 | *** join/#edev bbradley_ (n=bbradley@78-105-167-16.zone3.bethere.co.uk) |
| 19:46.48 | *** join/#edev notadev (n=notadev@fu/coder/notadev) |
| 20:15.33 | MothaWork | prpplague: what is your experience of average serial port latency to/from userspace these days? |
| 20:16.15 | MothaWork | setserial has a "low latency" setting, but it looks like an old wart |
| 20:26.16 | *** join/#edev notadev (n=notadev@fu/coder/notadev) |
| 20:34.12 | prpplague | MothaWork: i've not had any problems with latency |
| 20:34.26 | prpplague | MothaWork: the only issue i've seen is when you use a fairly large buffer |
| 20:34.49 | prpplague | MothaWork: or have the min bytes flush setting fairly high |
| 20:35.06 | prpplague | MothaWork: for low latency stuff, you should have the uart set for 1 byte flush |
| 20:35.27 | MothaWork | where is that setting? |
| 20:35.45 | MothaWork | we have a userspace app that needs to effectively do CSMA on a serial line, which means low latency is a requirement |
| 20:44.44 | prpplague | MonMotha: sorry phone, one sec |
| 20:45.17 | prpplague | MonMotha: if you check the man page for the stty command, there is an option for "min" |
| 20:45.39 | prpplague | MonMotha: you can check the current setting by doing: stty -a -F /dev/ttyS0 |
| 20:46.19 | prpplague | MonMotha: some applications will set this to 5 or higher |
| 20:46.41 | MothaWork | well, our app just leaves it at whatever the kernel sets it to initially |
| 20:47.01 | MothaWork | which appears to be 1 |
| 20:48.53 | MothaWork | from what I've read, there's some delayed processing in the kernel that results in 5-10ms of latency, but I don't see any mentions of it on anything newer than 2.4 |
| 20:49.01 | MothaWork | I don't know if that has changed or just nobody's mentioned it |
| 20:50.52 | prpplague | MothaWork: yea, doing that kind of stuff isn't impossible, but probably takes some tweaking |
| 20:51.09 | prpplague | MothaWork: does your device have FIQ that can be associated with the uart? |
| 20:51.59 | MothaWork | FIQ is only associated with the external FIQ pin |
| 20:52.10 | MothaWork | there is of course the internal USART IRQ |
| 20:52.11 | prpplague | MonMotha: ahh |
| 20:52.34 | MothaWork | my experience has been that FIQ doesn't work right anyway |
| 20:52.39 | prpplague | MothaWork: your int controller have the capability to reorder the priorities? |
| 20:52.58 | MothaWork | prpplague: maybe, but this isn't interrupt latency |
| 20:53.25 | MothaWork | from what I read, this is the kernel delaying handing off the data to the line discipline (which then allows userspace to wake up on the blocked read call) |
| 20:53.41 | MothaWork | basically, it delays that handoff until the next tick, rather than doing it from the IRQ |
| 20:54.00 | prpplague | MothaWork: ahh, i follow you |
| 20:54.33 | MothaWork | I almost wonder if this is doable from userspace at all |
| 20:55.12 | prpplague | MothaWork: i'm wondering if it might better to use a atmega48 via spi to do the uart stuff, and add the csma stuff there |
| 20:55.20 | MothaWork | yeah, well, that's not an option at this point |
| 20:55.28 | MothaWork | the hardware has been shipped |
| 20:55.35 | prpplague | ahh |
| 20:56.13 | MothaWork | bus utilization is low, though, so I don't think CSMA failures are the only thing resulting in our issues |
| 20:56.30 | prpplague | MonMotha: using a stock uart driver? |
| 20:56.40 | MothaWork | yeah, it's the atmel_serial driver in the kernel |
| 20:57.00 | prpplague | MonMotha: avr32? |
| 20:57.05 | MothaWork | AT91 ARM |
| 20:57.09 | prpplague | ahh |
| 20:57.17 | MothaWork | silly, you seem to think we have more than one project :) |
| 20:57.27 | *** join/#edev g1powermac_PB (n=g1powerm@unaffiliated/g1powermac) |
| 20:57.32 | prpplague | MothaWork: hehe, been a long week |
| 20:57.39 | MothaWork | I'd love to play with AVR32, but the opportunity has not yet come up |
| 20:58.00 | prpplague | MothaWork: definetly learned something this week |
| 20:58.21 | prpplague | MonMotha: don't use intersil parts if you need quality stuff for higher speed SoC's |
| 20:59.04 | MothaWork | ok |
| 20:59.09 | MothaWork | still having issues with those power supplies? |
| 20:59.17 | prpplague | MonMotha: aparently the EEdiot changed the part without telling anyone , and it looks like he never even looked at the ramp up times on the scope |
| 20:59.42 | MothaWork | I see |
| 20:59.45 | MothaWork | CPU lacks a BOD? |
| 21:00.09 | prpplague | MothaWork: we reworked the 1.8 using a ltc3406 , the 1.3 keep alive with a TI ldo, and the 1.3 core with another ltc3406 variable |
| 21:01.13 | prpplague | MonMotha: has 3.3v, 1.8v , 1.3 core, 1.3 keepalive, all that have to come up in sequence |
| 21:02.58 | prpplague | the 1.8v and 1.3 core ramp up times were so slow on the intersil parts that the 1.3 was up before they had finished ramping up |
| 21:04.04 | prpplague | MothaWork: apparently the EEdiot out right lied about having done the tests |
| 21:05.17 | chouimat|work | ouch |
| 21:06.11 | MothaWork | wow |
| 21:06.17 | MothaWork | power sequencing is often a complicated beast |
| 21:06.27 | MothaWork | I got lucky in that neither of my current CPUs required explicit sequencing |
| 21:06.49 | *** join/#edev roxfan2 (n=dunno@56.60-136-217.adsl-dyn.isp.belgacom.be) |
| 21:08.02 | prpplague | MothaWork: normally it isn't as a big of a deal, but we are using a nand flash booting feature that requires the 1.8v memory be power before the 1.3 keepalive, since the 1.3 keepalive power on startes a hardwired process in the SoC to read from the nand flash |
| 21:08.29 | prpplague | MonMotha: so if the 1.8v is still ramping up with the 1.3 keep alive comes up, the nand flash isn't there, so the system halts |
| 21:08.45 | prpplague | s/with/when |
| 21:10.34 | MothaWork | prpplague: ah, that would be a problem |
| 21:11.02 | MothaWork | I do have to give Atmel credit for their power-on stuff. They're usually very flexible. |
| 21:11.06 | MothaWork | AVRs even have a built in BOD |
| 21:13.09 | prpplague | yea, problem here is we have alot of different rails coming up |
| 21:13.25 | MothaWork | sounds lik eit |
| 21:13.33 | MothaWork | ever had the pleasure of dealing with FPGAs? |
| 21:13.46 | MothaWork | some of those power sequencing requirements are so strict that special PMICs get made just for certain FPGAs |
| 21:13.49 | prpplague | yea we use some on a couple legacy projects |
| 21:15.25 | prpplague | solar: you gonna be online this evening, i need to give you some updates |
| 21:24.02 | *** join/#edev notadev (n=notadev@fu/coder/notadev) |
| 21:36.26 | solar | prpplague: yes and I'll probably be sober |
| 21:48.14 | prpplague | solar: i've got to run some errands, be back online in about 2-3 hours |
| 21:49.35 | *** join/#edev notadev (n=notadev@fu/coder/notadev) |
| 21:54.06 | solar | ok |
| 22:07.58 | *** join/#edev frikker (n=blaine@107.135.68.216.DED-DSL.fuse.net) |
| 22:14.01 | *** join/#edev notadev (n=notadev@fu/coder/notadev) |
| 22:23.25 | *** join/#edev notadev (n=notadev@fu/coder/notadev) |
| 22:32.49 | *** join/#edev notadev (n=notadev@fu/coder/notadev) |
| 22:44.38 | *** join/#edev notadev (n=notadev@fu/coder/notadev) |
| 22:53.34 | *** join/#edev notadev (n=notadev@fu/coder/notadev) |
| 22:59.44 | *** join/#edev notadev (n=notadev@fu/coder/notadev) |
| 23:17.51 | *** join/#edev notadev (n=notadev@fu/coder/notadev) |
| 23:21.39 | *** join/#edev GPSFan_ (n=kenm@64.92.145.112) |
| 23:21.39 | *** mode/#edev [+o GPSFan_] by ChanServ |
| 23:23.39 | *** join/#edev Madsy2 (n=madsy@ti500710a342-0803.bb.online.no) |
