#qi-hardware IRC log for Friday, 2011-08-19

wpwrakmaybe they'll open source it ? then someone can port it to the freerunner ;-)00:07
wolfspraulare there parts not open sourced yet?00:08
wolfspraulyeah maybe they'll do a code dump in the end, and a few pieces can be rescued to OpenWrt and other healthy projects00:09
kristianpaulall gui is closed source so far i now.. but people said it uses OE00:09
wpwrak(open) uh, no idea really. just assumed :)00:09
wolfspraulyes I think so (OE), or they forked or something00:09
wolfspraulI was involved in the very beginnings of this project, when it was still called Maguffin in 2003/2004. Sad to see it end like this, but I cannot say that I lost much time over it the last few years :-)00:11
wolfspraulIn fact somewhere around 2006, I refused to continue working on it, causing a nice little run-in with my boss.00:12
wolfsprauldevelopment was so slow that we fell behind our own hardware components! :-)00:13
wolfspraulthe first few years everything was hardcoded to one particular screen resolution. problem was - bad sourcing feedback - that screen resolution was unusual (forgot which one, something high-end at the time it was chosen).00:14
wolfspraulthen it became harder and harder to source that screen, crazy hard in the end.00:14
wolfspraulfinally it was not sourcable anymore and the entire software stack came down and had to be redone with another resolution (or supporting multiple).00:15
wolfspraulall this took years00:15
wolfspraulso many mistakes we understand so well now. Like - be super careful choosing the 'latest and greatest' chip or component.00:16
wpwrak"do not hard-code the resolution" ;-)  (well, i sometimes do that too :)00:16
wolfspraulHP still plans to "explore options to optimize the value of WebOS software going forward."00:24
wolfspraulRead more: http://news.cnet.com/8301-17938_105-20094249-1/what-hps-pc-spin-off-plans-mean-for-you/#ixzz1VQi49CXm00:24
wolfspraul(I don't like how copy/paste creates this auto-url...)00:24
wolfspraulhave to clean it through a text editor00:24
wolfspraul"optimize the value", let's see what that means in the end00:25
wolfspraulmaybe code dump, yeah. I don't think anybody has the guts to make devices for it anymore.00:25
wpwrakprobably just management blabla00:25
wolfspraulyes they need some lessons in clear speaking00:26
wpwrakweak lie-ability :)00:35
DocScrutinizerI wonder what's going on at Samsung and with Raster's Linux01:55
wpwrakyeah, the most secret project in existence01:56
DocScrutinizerthere's been a 2.3 pages press release with a bit of block diagrams with all the closed components colored blue01:57
DocScrutinizerbeen quite a while since01:57
DocScrutinizerand Raster notably stopped spreading teaser rants across the channels every now and then, since also quite a while01:58
wolfspraulSamsung is one of the few companies where I like their strategy (amazon being another one)02:01
wolfspraulmaybe not 'like' but 'understand'02:02
wolfspraulSamsung's strategy is "we make devices with _all_ software we can get"02:02
wolfspraulthat works02:02
wolfspraulalso very unique, afaik nobody else has that strategy02:02
wpwrakyou get a big of that with oscilloscopes. a tek or agilent will have the basics plus a number of expensive extension packages. some of the asian competitors include these features in the base version.02:05
DocScrutinizerprobably it never occurred to the asian folks you could also *sell* software rather than just *copy* it02:08
DocScrutinizerseems to me the concept of copying is much closer to the asian mentality than the concept of selling IP02:10
DocScrutinizerbut then I don't know a shit about asian mentality at large02:11
wpwraki guess it's materialism at work. you sell material goods. you copy the immaterial ones. makes perfect sense :)02:13
DocScrutinizeryup, that's what I meant I think02:13
DocScrutinizerimitation is a compliment in Asia afaik. So how could you ask money for copies of whatever02:14
wpwrakeven better :)02:15
wpwraki mean, non-corporate westeners think very much alike. if "file shearing" is so easy, how could it be wrong ?02:16
DocScrutinizeroooh I think each westerner knows it's basically wrong. Just nobody cares anymore02:17
DocScrutinizeryet you can hear all sorts of the weirdest excuses why they do it _nevertheless_02:18
DocScrutinizermost popular one "I wouldn't buy it anyway"02:19
wpwrakwolfspraul: btw, how are the bens selling these days ? constant trickle ? or some changes ?03:06
wolfspraulI think my rough plan is like this - first really finally get my second product, m1, into stock03:07
wolfspraulit's crazy I say this since January but well, it's still not done03:07
wolfspraulafter it is in stock, back to marketing03:07
wolfspraulgreat shop03:07
wolfspraulnice and clean and good looking03:07
wpwrakthings always take longer than you think ;-)03:07
wolfsprauleasy shipping options03:08
wolfspraulwant to work on backend too, openerp03:08
wolfsprauland come out with improved openwrt images or other ben software03:08
wolfspraulthe Linux 3.0 bot messages were annoying but also encouraging :-)03:08
wolfspraulI think we can get a lot of music software to the ben03:08
wolfspraulnot just music player, also more things like tracker, all the way down to metronome :-)03:09
wolfspraulI will just continue with the Ben, basically.03:09
wpwrakrosegarden ;-))03:09
wolfspraulthen have to find partners who see the technology and potential and want to turn the tech into a different kind of product, or just move it forward to Ya.03:10
wpwrakyup, the ya should be our real platform. the ben is still a weak basis. aging and not fully open.03:11
wolfspraultotally agree03:15
hamneggaIs there a special kind of solder to use for electronics, especially 2.4GHz hardware components.  I have some Oatey Safe Flo Lilver Lead Free solder, and another from archer that's rosin core solder.  Would either work, or should  I use neither.  I was thinking the lead free, but it's typically used for plumbing/heat-hot water piping.05:40
hamneggathe rosin core is 60/40 whatever that means, and the silver/lead-free doesn't say05:41
wpwrakdon't use plumbing solder. it has the wrong chemistry for electronics.05:48
wpwrak60/40 is a common solder that works well. lead-free solder is usually a bit harder to work with.05:51
wpwrakif you do SMT components, you'll also have to add extra flux. there RA/RMA (rosin), water-soluble, and no-clean. i find the water-soluble fix the most convenient to work with. note that "water-soluble" doesn't mean it will come off easily. you still have to do a good amount of scrubbing and rinsing.05:52
hamneggaoh, cool, actually then I don't have to resolder the little brown thing I attached to my GPU when it broke off.  I thought led was a bad conductor.  The card has been working up to par, but wasn't sure what the little brown rectangle was and thought I might have lost a shader core or something in the process.  06:42
hamnegga3 in the morning though, I gotta go to bed and get up for work in 4 hours so, I'll talk to u tomorrow about soldering up some extra custom wifi dishes.  I might actually go at it again.  Would additional uhf antennas on the back of a direct tv dish benefit at all for 2.5 GHz, even if I kept the male and female respective, or would it just add more length and dBM and dBi loss06:44
hamneggaI thought the more silver would be better for 2.4GHz/wifi, because that's what all their components use, like sma, tnc, and bnc regarding microwaves.06:46
hamneggayou need to be an expert I guess and each connector nevermind noobish soldering job is bound to lose at least 1-5dBi06:47
xiangfutry to boot linux 3.1 on nanonote:08:38
xiangfu    0.000000] Linux version 3.1.0-rc2+ (xiangfu@macbook) (gcc version 4.5.2 (Linaro GCC 4.5-2011.02-0) ) #1 PREEMPT Fri Aug 19 16:29:03 CST 201108:38
kyakxiangfu: does it boot?08:39
xiangfuyes. 08:39
xiangfuworks fine. 08:39
xiangfuone little thing I forget to change to load '/etc/preinit'08:39
xiangfunow I am try to move those patches to openwrt-xburst and try it inside openwrt.08:40
kyakgreat :) congratulations!08:40
xiangfukyak, we should think about how to merge the work from qi-kernel.git to openwrt-xburst.git patches. :)08:40
kyakis 3.1 avialable in qi-kernel.git?08:41
kyakxiangfu: i think the main problem is that openwrt trunk is still using 2.6 for xburst08:41
kyakthe patches would be too big08:41
xiangfuI saw the 'jz-3.0' branch. 08:41
kyaksomeone must transfer this work to openwrt directly08:42
xiangfualmost same as 2.6 patches in upstream.08:42
xiangfuI direct work on latest upstream 3.1-rc2 08:42
xiangfuI mean I direct move 2.6.37 patches on top of 3.1-rc2. 08:43
xiangfukyak, yes. 08:43
xiangfueven more sync with linux upstream.08:43
kyakright now it's 2.6.37 in openwrt trunk08:43
xiangfukyak, today I forward all u-boot and linux patches on top of last upstream :)08:43
kyaki remember lars told several weeks ago he was going to update it to 2.6.39. I think it's time to update to 3.1 :)08:44
xiangfukyak, by the way I think I will release a test/debug version recently. 08:44
kyakxiangfu: hope your patches will make it into openwrt...08:44
xiangfukyak, I saw there are already some target have linux 3.108:45
kyakactually, even malta has it08:45
xiangfukyak, let me check now if there is 3.1 in openwrt recently.08:46
xiangfukyak, our last rebase have 3.008:46
kyakyeah, but not for xburst08:47
xiangfuyes not for xburst.08:48
xiangfuI mean "no, not for xburst"08:48
xiangfuChinese and English grammar ;D08:50
kyakbtw, "yes, not for.." sounds perfectly find in Russian :)08:51
kyakyou can even say "yes, no, maybe" and this still make sense08:51
xiangfukyak, I think I will release : http://fidelio.qi-hardware.com/~xiangfu/compile-log/openwrt-xburst.full_system-08172011-0556/ as the next release anyway.08:51
xiangfuafter release , rebase immediately on last openwrt upstream 08:52
kyakxiangfu: i see it has python?08:52
kyaki mean, the python is there as module08:52
kyakbut it will segfault?08:53
xiangfukyak, no. only in package. not include by default08:53
kyakthis sounds right08:53
kyakyeah, probably if we rebase now, especially start using 3.1 we will run into mode troubles :)08:54
kyaki will flash the 08172011 and give it a try08:54
kyakthe only really great thing i will be missing is busybox patch to enable reverse history search08:55
kyakprobably we can include this patch into openwrt-xburst.git?08:56
xiangfukyak, you can use the last reflash_m1.sh as reflash_ben.sh -d openwrt-xburst.full_system-08172011-0556 for download 'dailybuild'08:58
xiangfu-d is for dailybuild08:58
xiangfuyou maybe already know that :) 08:58
xiangfukyak, it will download the image in currect folder.09:01
xiangfukyak, Ctrl + R, yes  sure why not .09:03
kyakxiangfu: yep, i remember the -d flag :)09:31
kyakxiangfu: the "killall gmenu2x" thing in ben-nanonote rc script. Do you think it is right?09:33
kyaktaking into account that gmenu2x is started by init09:33
kyakkillall gmenu2x would make it restart immediately09:34
kyakwhich could lead to some strange effects, because init itself is being killed09:34
xiangfukyak, it start the shell not gmenu2x I think. inside the shell there is trap "" HUP which is for ignore the killall gmenu2x. 09:43
xiangfumaybe we should killall gmenu2x.bin ?09:45
kyakhm, why killall gmenu2x theN?09:45
DocScrutinizer~seen roh09:46
kyakxiangfu: just trying to understand, why kill it at all? Isn't it done automatically when init is being killed?09:46
xiangfukyak, just for fast, kill gmenu2x.bin first. then the screen goto blank. shows the poweroff is running.09:47
Aylamaybe because of one small bug of SDL09:47
xiangfuAyla, gmenu2x SDL bug?09:48
AylaI was working on it yesterday09:48
Aylalet me build a test example09:48
Aylaah, I have one online09:48
Aylaafter the SDL_Init(), gmenu2x has 3 threads running09:49
Aylawhile the Timer subsystem of SDL creates only one (I checked it, only one SDL_Thread object is created)09:49
Aylaso there should be two threads09:49
Aylaanyway. After SDL_Quit(), the unique thread created by the timer subsystem is exited; so two threads are left, while there should be only one09:50
Aylanow it begins to be very problematic: after the exec call to /bin/vi, the 'ghost' thread is still running09:51
kyakdoes it matter if we are going to shutdown anyway?09:53
AylaI didn't get that right :)09:53
Aylabut could you please confirm you also have that bug on the nanonote?09:54
kyaki could try, but in several hours. Should i look at the number of processes spawned by gmenu2x? Or how do i see the threads?09:57
Aylakyak: htop10:10
kyaki thought that :)10:10
qi-bot[commit] Xiangfu Liu: nanonote: reflash_ben.sh update option -t for download testing version images (master) http://qi-hw.com/p/openwrt-packages/553830311:30
xiangfukyak, ^  just add another option for testing images. : http://downloads.qi-hardware.com/software/images/NanoNote/Ben/testing/11:32
rjeffriesDoes Tuxbrain visit this channel these days? If so, he may find this link interesting:12:51
rjeffriesZigduino with Contiki OS should provide an interesting end point for Ben w/ATben to reach out and touch over 802.15.4 and (eventually) 6LoWPAN12:53
kyakxiangfu: ah yeah, that's good thing. Are nightly images copied to testing/ automatically or by hand?13:07
viricquestion, dear hw people13:38
viricDo you know of any kind of usb-connected lamp that I could switch on and off in linux?13:39
virichttp://www.linuxjournal.com/article/7353 this looks fine13:40
virichttp://www.delcomproducts.com/products_usblmp.asp very expensive13:44
rjeffrieswhat is your use case viric13:48
viricI have some computers away. And I'd like of people to be 'notified' that some processing is ready13:50
virics/of //13:50
rjeffriesah I see. cool idea13:53
virichm nothing new, I imagine :)13:58
kristianpaulring ring :)13:59
virickristianpaul: ring ring?14:10
kristianpaulviric: lol. i mean you can use a electric bell, nv..14:13
virickristianpaul: do you know of a usb belL? :)14:14
kristianpaulafaik i dont ;)14:15
viricthose I gave the link cost $9914:15
viricthat's mad14:15
viricit's a usb led.14:15
larscnothing compared to the $500 ethernet cable14:17
xiangfukyak, Are nightly images copied to testing/, no.  only the image like '08172011' it's for release but have a little problem. :( that is the plan, hope we don't have image like '08172011' any more, mean don't have testing image anymore :D14:18
viriclarsc: there? Impressive :)14:24
rjeffriesviric maybe you need to kludge together a little arduino-ish solution14:50
viricrjeffries: to get until the $99? :)15:23
qi-bot[commit] kyak: reflash_ben.sh: small fixup in help output (master) http://qi-hw.com/p/openwrt-packages/b3492d315:24
rjeffriesif you wanted to you could use your BEN and a 15:27
rjeffriesthe next nanonote if there is a next nanonote needs among other things more i/i e.g. i2c, some gpios that can be used without using the 8:10 (microSD) port yada yada yada. maybe even USB host or OTG15:28
viricno no15:35
viricI want the solution to be *cheap*15:35
viricand zero effort. using the linux LEDs interface :)15:35
larscand now!15:40
viricI can't believe noone thought of a usb light controlled by software :)15:45
wpwrak_viric: you could use a C8051F326 plus a LED. maybe add a transistor. not too expensive.15:48
wpwrak_viric: thre a USB stack for it in the f32xbase project :)15:49
wpwrak_viric: including in-circuit programmer circuit for the Ben, etc.15:49
wpwrak_viric: afaik, the c8051f326 is the cheapest/simplest readily available solution for this sort of thing. (and without giving your money to evil FTDI)15:50
wpwrak_ah, wait. there's cheaper: an attiny with v-usb. some of them also don't need a crystal.15:51
wpwrak_viric: also some of the USB PICs can operate without crystal, but i don't know if they have a better price point for what you need.15:52
larsceven simpler ftdi + led15:53
wpwrak_rjeffries: using the Ben is also a nice idea ;-)15:53
wpwrak_larsc: ftdi is evil. that's like buying from apple :)15:54
larscwpwrak_: why?15:54
wpwrak_larsc: they don't document their chips well enough for proper free software drivers to be written. there's an amazing amount of reverse engineering and guesswork in these drivers, and if you have a recent chips, it's hit or miss whether things work or not.15:55
wpwrak_larsc: ironically, they seem to hate Open Source15:55
viricI have not had troubles with ftdi chips15:56
wpwrak_larsc: what they want you to use is their closed source library. they offer it for linux as well, but ...15:56
viric(maybe thanks to the efforts of reverse engineers :)15:56
viricI'd prefer a light brighter than a LED though15:57
wpwrak_viric: if you just use serial, then you're fine :) also the serial protocol engine (for jtag and such) in some of their pricier chips seems to be well-understood. but when you get int bit-banging, ...15:57
viric500mA at 5V give for 2.5W of light :)15:57
viricwpwrak_: I only used serial.15:57
wpwrak_multiple LEDs ? high-power LEDs ?15:57
wpwrak_or use a relay and connect some multi-kW halogen ;-)15:58
viricI'm really not that interest to build the solution myself now :)15:58
wpwrak_the circuit would be quite simple15:58
viricSo, I either find something built that works on linux, plug and "echo 1 > led", or I'll go without.15:59
viricI know I know15:59
rjeffriesviric you remind me of me15:59
viricFor $99 I could buy a sheevaplug; it has two leds with proper kernel interface :)15:59
viricrjeffries: I'm simply proud of having better things to do ;)16:00
rjeffriespoint well taken16:00
kristianpaulyes pic pic !!16:32
kristianpaulviric: you can build a pinguino board (based on a pic18f2550) for less than 10usd in my countri16:33
kristianpaulactually there is python/c++/java code for usb comuncation but no linux support...16:33
kristianpaulwell, it can emulate a serial device and cheat by seriall;)16:34
wpwrak_a serial cheater ;-)16:34
kristianpaulviric: if you are near  spain in think a guy sell then16:35
kristianpauli actually tried with python but threads we're anoying me at than time..16:36
kristianpaulbulk mode i think..16:36
wpwrak_kristianpaul: cheat = estafar (also "poner los cuernos"); chat = charlar16:36
kristianpaulle puse los cuernos al puerto serial, jajaja16:37
wpwrak_lo hizo con USB ;-)16:37
kristianpaulviric: where are you located?16:38
qi-botThe build was successfull, see images here: http://fidelio.qi-hardware.com/~xiangfu/compile-log/openwrt-xburst.full_system-08182011-1248/16:38
erikkugelHey guys, I need to update a package from a feed (GNU Pem) -  a new version came out which has special Nanonote-friendly tweaks. How would I go about something like this? What's the procedure to update a package so that future pulls use a new version of it? I thought I'll ask around before emailing Xiangfu directly...16:48
erikkugellet me know if the question is not clearly worded...16:48
mthyou mean automatically fetching the latest version, or doing a manual version upgrade?16:52
erikkugelI mean doing a version upgrade. I ported PEM a month ago at version 0.7.8, but the developer added some new features especially for the Nanonote in 0.7.9. What I would like to ideally do is to edit/replace the current Makefile with a new Makefile on the servers from which the feeds are pulled.17:06
rjeffrieswpwrak kristianpaul No hablo Espaniol. LOL17:31
Artyomhi kristianpaul19:45
kristianpaulhello Artyom 20:02
ArtyomHow is your success with namuru+osgps? ;)20:02
kristianpaulnamuru accumualtors are operationl now20:02
kristianpaulafaik i dint measure its opetation in depth yet20:03
Artyomand what are your plans? What is your next step? 20:05
kristianpauli have a concern now, because i have to fire by software the accum interrupt  signal...20:05
kristianpaulplans, sure, i'm goint to improve some test tool to support now right PRN codes,20:05
kristianpaulalso collect that accumulator data and plot it20:06
kristianpaulthen guess a threshold from wich in can improve the acquisition loop20:06
ArtyomI would suggest the following easy test:20:07
ArtyomChoose a settelite that is definitly visible. Set it's prn number. Set code_reference_frequency to default value (2.046MHz). Set carrier_frequency to zero doppler. Then pass through all delays and record values in memory. After that set next doppler frequency. And repeat. This way pass through all doppler frequencies.20:10
ArtyomDoppler frequency step can be 1000 Hz20:10
ArtyomDelay step can be 1 or two half-chips20:11
ArtyomFinally move all values from memory to the file and plot the result20:11
ArtyomYou should see one peak in the plot20:12
ArtyomThis can help you to check what is your noise floor and what is your peak maximum.20:12
kristianpaulwhat registers have effect in doppler freq? carrier nco?20:12
Artyomyes only carrier_nco20:14
Artyomof coarse doppler affects also code_nco but this affect is negligble20:14
kristianpaulbtw, where you the that 2.046MHz value?20:15
ArtyomThis is because code_nco works with double chip rate (2*1.023 MHz) as I remember20:16
kristianpauli wasnt aware of that i'll recheck my notes :)20:17
kristianpaulArtyom: so how is work from your side?20:20
kristianpaulbtw are you reading some RTC or similar for keep measurememnt timing constant?20:20
kristianpaulor doest matter yet to care about it..20:21
Artyomwhat do you mean by "timing constant" and "RTC"?20:21
kristianpaullet me explain20:22
qi-bot[commit] kyak: busybox: backport reverse history search patch (master) http://qi-hw.com/p/openwrt-xburst/3368b5e20:23
qi-bot[commit] kyak: config.full_system: enable busybox reverse history search (master) http://qi-hw.com/p/openwrt-packages/68f277920:25
kristianpauli modifify namuru to clear after every clock cycle status, new_data and ch0_prn_key internals registers20:26
kristianpaulalso in software i have to enable ch0 and set clear status flag20:28
kristianpaulmy point is, i still dont like the idea that the accum int signal is fired by software it self20:28
kristianpauland not from know constant timing source20:28
ArtyomI've spent a lot of time on playing with hardware... But with little success. I couldn't achive stable tracking. I recorded correlator outputs and then plot the result in scilab. The best result that I saw - is couple of bits (like in the picture that you showed me). At the same time the same code for tracking works fine on PC (program that uses soft-correlator from osgps).20:30
Artyomso you always have new_data=0?20:32
kristianpaulwell i changed namuru code a bit from last time we talk20:36
kristianpaullike adding  a exclusive enable signal for channels20:37
kristianpauland the prn key clear i pointed you earlier20:40
Artyomyou clear status_read register by writing to special memory address?20:40
Artyom(that's funny - I've done the same thing in my design ;) )20:45
kristianpauldont you ask you self if that namuru code really worked as it was published?20:46
kristianpaulanyway this is getting better :)20:47
ArtyomThis question raised so many times during last two weeks. Before I gave up today to fire tracking loop - I tried everything: switched prompt-early-late... And I and Q... My last idea is to rework everything... Starting from pc-program that models hardware and finishing with reworking all namuru-code. And making exhausting testing on every stage.20:52
kristianpaulhumm.. and and you graphing this?20:59
kristianpaulmaybe it looks like this  http://www.colorado.edu/geography/gcraft/notes/gps/gif/halfcorr.gif21:00
Artyomsorry... didn't understand questions21:00
kristianpaulnah, is me it used to happen :p21:01
kristianpaulyou said having problems with tracking loop,21:01
kristianpaulso i wonder you can make a plot of it and compare with your expected signal some how...21:02
Artyomyes, I can plot it. And I made a lot of plots. But it's difficult to debug tracking loop in hardware. Because each time it starts to work new data is comming. I cannot repeat the same data several times :(21:05
kristianpaulArtyom: btw at #ephel channel there is this guy called lemay21:16
ArtyomI've seen your conversation in the history :) He is experienced in GPS ;)21:17
kristianpaulyeah :)21:17
Artyomtime to sleep. gn ;)21:26
kristianpaulbye !21:26
kristianpaulokay lets move some stuff to rtems...  no time to deal with gettng linux to boot again..22:48
zedstarsome ben wpan pimpage http://www.vimeo.com/2792400423:10
qi-bot[commit] kyak: add vitetris (master) http://qi-hw.com/p/gmenu2x/e98e68a23:17
qi-bot[commit] kyak: config.full_system: include vitetris (master) http://qi-hw.com/p/openwrt-packages/0c3238123:19
qi-bot[commit] kyak: gmenu2x: update to the latest git (include vitetris) (master) http://qi-hw.com/p/openwrt-packages/d4c8f4d23:19
kyakzedstar: nice :) i couldn't help noticing that you are living 40 years behind though :)23:23
zedstarkyak: how u mean?23:24
kyak2 Jan 1970 :)23:26
zedstarkyak: lol23:27
kristianpaulnice music :)23:37
kristianpaulwow you had whole army there zedstar :)23:38
zedstarkristianpaul: yeh :)23:41
--- Sat Aug 20 201100:00

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