#qi-hardware IRC log for Friday, 2011-01-07

xiangfuyizhang: great job : http://en.qi-hardware.com/wiki/Applications00:38
yizhangxiangfu: thanks!00:57
yizhangxiangfu: still a lot of work to do00:58
yizhangxiangfu: try to find icons for our apps00:58
kyakfighting with the whole meaning of openwrt is pointless02:56
kyakit IS a distro for routers02:56
kyakthey make irreversible changes all the time02:57
valhalla_is there any reason why uImage should be copied in a certain order on the SD card to be able to boot it? (I'm trying to boot a custom image made with OE on the nanonote)02:57
kyakxiangfu: btw, your patch for gettext (oh, gettext-full) is still not there... though the bug status was changed to "accepted"02:58
xiangfukyak: yes. I saw your comment, in bug 841302:59
kyakat some point, openwrt will make a really critical change and we will have to branch or to switch to another distro02:59
kyakit will be very painfull02:59
xiangfuvalhalla_: the bootloader hard code to load '/boot/uImage'03:00
valhalla_xiangfu: yes, it was in that position, but it wasn't able to load it03:01
xiangfuvalhalla_: the bootloader only try first partition. and the first partition must ext203:02
valhalla_(it happened yesterday, so I don't have the error message right now, it was just a random idea)03:02
valhalla_xiangfu: check03:02
xiangfuvalhalla_: make sure you using the last version of bootloader. the early u-boot will have some problem on > 8GB sd card03:03
valhalla_how do I check the bootloader version? (it was able to boot a jlime image from the same card, anyway)03:04
xiangfuvalhalla_: do you have openwrt in your nand? it's not easy to check the u-boot version, but you can check the '/etc/VERSION' in openwrt rootfs.03:07
xiangfuvalhalla_: you can update your bootloader to 12-14: http://downloads.qi-hardware.com/software/images/NanoNote/Ben/2010-12-14/openwrt-xburst-qi_lb60-u-boot.bin03:07
valhalla_xiangfu: no, I have jlime (it's the nanowar edition from tuxbrain)03:08
xiangfukyak: since 12-15 , I never have a success on compile config.full_system :(03:08
kyakxiangfu: me too... since that really stupid idea to stub libiconv and gettext03:09
kyaki wonder what will be the next problem after this one with glib2 is fixed03:22
qi-bot[commit] kyak: qstardict: fixed missing dependencies http://qi-hw.com/p/openwrt-packages/0f82e4203:27
kyakzedstar: "Searchable USB flash drives" - very nice!03:46
kyaktalking about this one: http://zedstar.org/blog/2011/01/06/searchable-usb-flash-drives/03:46
kyakzedstar: can you explain more details? what is this search engine? the pdf viewer, is it nupdf? what were you using as httpd? can it search in pdfs only? have you ported the search engine to openwrt?03:49
qi-bot[commit] kyak: put qi openwrt-packages git on top in feed.conf http://qi-hw.com/p/openwrt-xburst/46b044203:57
qi-bot[commit] kyak: Merge branch 'master' of projects.qi-hardware.com:openwrt-xburst http://qi-hw.com/p/openwrt-xburst/ba5459903:57
kyakxiangfu: seems that it can be built now against libiconv and gettext full version (overriden from openwrt-packages).03:58
kyakthe problem is, glib2 has to be overriden too (to have the --with-libiconv=gnu option back)03:58
kyakxiangfu: how do you think, should be override glib2, too? at least temporary03:59
xiangfukyak: yes. I agree04:00
kyakok then04:00
qi-bot[commit] kyak: use full versions of libiconv, gettext; fix glib2 http://qi-hw.com/p/openwrt-packages/f981e8304:02
kyakxiangfu: please test if it works for you, too04:02
xiangfukyak: ok. thanks04:02
xiangfukyak: btw, I also test put qi-package on top of feeds. 1 rm feeds/ tmp/ -rf  2. make package/libiconv/{clean,compile} V=99  3. seems it still build the upstream version(stub)04:03
xiangfukyak: I will test your commit first.04:04
kyakmm, seems you have to make package/symlinks after 1 rm feeds/ tmp/ -rf04:04
xiangfukyak: thanks.04:08
kyaknp.. really hope it works for you, too04:09
zedstarkyak: thanks....it is something called estraier which can search different things like pdfs, text, word doc etc...this one used pdftotext to be able to look inside the pdf. the search engine comes with own web server process. yeh ported to openwrt although some stuff isnt like nupdf etc04:36
zedstarkyak: i built a bunch of ipks but needs some work to get into an easy installable state as has web page stuff, and config files etc not in ipks04:37
wolfspra1lzedstar: ah yes I just wanted to ask whether we can package this...04:38
zedstarwolfspraul: yeh it should be...not sure how to handle things like nupdf04:40
rohlarsc: git should work again.. please try04:54
wolfsprauladamw_: ok I'm looking at this: http://www.molex.com/pdm_docs/sd/022284040_sd.pdf05:27
adamw_wolfspraul, http://www.molex.com/pdm_docs/sd/022284040_sd.pdf05:27
wolfspraulyeah :-)05:27
adamw_see page 1 firstly05:27
wolfspraul42375-1924 - page 1705:28
adamw_to know "M" and "P" firstly05:28
wolfspraulwhat length do we need?05:28
adamw_I need that  7.6 mm , so M + 2.29mm <= 7.6mm05:29
wolfspraul42375-1924 has a pin length (l) .47/11.94, mating length (m) .19/4.83, pc tail length (p) .19/4.8305:29
adamw_the "P" is actually I don't CARE.05:30
wolfspraulok what is this .19/4.83 - what unit is .19, what unit is 4.8305:30
wolfspraul4.83 mm ?05:30
wolfspraul0.19 inch?05:30
wolfspraulah ok05:30
adamw_if I use "42375-0004" then you can search...it's not that I want!05:31
wolfspraulso we need M + 2.29mm <= 7.6mm - M <= 5.31 mm05:31
wolfsprauland here M is 4.83, so it fits05:31
wolfspraul4.83 < 5.3105:32
adamw_M = 0.23" (seems they are all standard, that's now why Digi-Key and Mouser selling!)05:33
adamw_you will really hard to find a suitable part for our special case.05:35
adamw_wolfspraul, http://dkc3.digikey.com/PDF/TW2010-11/secta.pdf05:36
adamw_wolfspraul, http://tw.mouser.com/Catalog/catalogusd/642/dload/pdf/INTERSECTION.pdf05:36
adamw_i searched both Digi-Key and Mouser, I've not found a suitable one.05:38
wolfspraulok, 0.23'' = 5.8mm05:39
adamw_or 0.24''05:39
wolfspraulthat's only .5mm more than what you need05:39
wolfspraulis this even noticable?05:39
adamw_most catalog show now.05:39
wolfspraulif the header is 0.5mm too high, maybe that's good enough?05:40
wolfspraulright now it's 4mm too high, right?05:40
adamw_yes...too high.05:40
adamw_I can only say I want a total hight of 7.1mm~ 7.6mm from M1 board top side05:41
wolfspraulwhat is the mating length of the current header on m1?05:44
adamw_7.6mm - 7.1mm = 0.5mm, this is the C27's height.05:44
adamw_current mating length on m1 = 5.95mm05:45
adamw_but need to add 2.5mm, so 5.95mm + 2.5mm = 8.45mm05:45
wolfspraulI cannot follow05:46
adamw_calculating the height, you need count mating length + plastic length.05:47
wolfspraulwe found a header with 5.8mm mating length, which is only .5mm more than what you want05:48
wolfspraulit sounds like .5mm may be good enough?05:48
adamw_you forgot to add 2.29 mm05:48
adamw_go to see page1 of "022284040_sd.pdf"05:49
adamw_to feel "M" length + 2.29mm <= 7.6mm (I want)05:49
wolfspraulso m should be <= 5.31 mm05:50
wolfsprauland the one we find available on mouser/digikey has m = 5.8 mm ?05:52
wolfspraul0.23'' = 5.8mm05:52
adamw_5.8mm + 2.29mm = 8.09 mm05:53
adamw_8.09 - 7.6 = 0.5mm05:54
wolfspraulso the different is .5mm05:54
wolfsprauldon't you think that's 'good enough'?05:54
adamw_i know you are saying why 0.5mm I can not accept it.05:54
wolfspraulI'm just wondering how it feels/looks in reality05:55
wolfspraul.5mm - sounds like really small05:55
wolfspraul(in this context)05:55
adamw_but actually is .24/6.09  + 2.29 = 8.38mm, 8.38mm - 7.6 mm = 0.78mm (~ 0.5mm)05:58
wolfspraulok now you make it .24 inch (before was 0.23 inch)05:59
adamw_but please the 7.1 mm is the final i want, if taking C27's height in next run(actually I don't know)05:59
adamw_datasheet says 0.24''05:59
adamw_some other says 0.23''06:00
wolfspraulyou mean 7.1 - 2.29 = 4.81 ?06:01
wolfspraulso you want M <= 4.81 ?06:01
wpwrakthe dark side of math: making addition confusing :)06:02
wolfspraulat least we get some practical experience in mating length calculations now, who knows when this may come in handy... :-)06:02
adamw_i can only say I want 7.1mm <=height <= 7.6mm from my exactly jtag/seral and m1 now. :)06:03
wpwrakkyak: xiangfu once mentioned  http://www.debwrt.net/trac/wiki  maybe this could be a more suitable candidate in the longer term ?06:04
adamw_wolfspraul, I have another plan which can be done in smt vendor.06:04
wpwrakwolfspraul: the best lessons are taught by mistakes :)06:05
adamw_wpwrak, the mistakes we created already and knew . :) need to clean it. :)06:06
adamw_wolfspraul, the package is on the way ST INGBERT DE.06:09
rohwolfspraul: do the cases fit on your boards?06:09
wolfspraulI have no boards here.06:12
wolfsprauladamw_: but I just got the box from fedex with the jtag-serial adapters - nice!06:12
wolfspraulwpwrak: btw, did you get your nanos?06:12
adamw_wolfspraul, ok, good, firstly I plug a very small metal into P1 on jtag/serial to measure..the depth. I got 2.5mm.06:14
adamw_wolfspraul, do you think you can go to somewhere to hook jtag/serial in m1?06:15
wolfspraulbut I will send the boards to the people who have the boards06:16
wolfspraulof course they will first run into the height problem, have to clip their headers (all of which is very easy of course)06:16
wpwrakwolfspraul: they're still sitting at customs. sat there all day long yesterday.06:18
wolfsprauladamw_: I fully understand the C27 problem now :-)06:20
adamw_wolfspraul, also to feel U1 on jtag/serial.06:20
adamw_U1's body edge.06:21
adamw_the C27's height is around 0.5mm. :)06:22
wolfsprauladamw_: U1 is too close to the jtag header?06:26
adamw_just its edge interference. Yes, too close!06:27
kristianpaul"Developed an adaptive integration routine in Lua--a language I had not heard of until I got the Nanonote. Lua is quite fast and small" nice to read this06:27
adamw_wolfspraul, http://en.qi-hardware.com/w/images/5/56/M1_rc2_J5_header_cut_shorter_pod_U1_collide_J6_headers_outlet_on_m1.png06:28
adamw_that's why I took this picture.06:28
adamw_thus actually jtag connector can be contacted very well if C27 & U1 move away a bit.06:29
wolfspraulyeah, got it06:29
rohwolfspraul: i see.. because i have only tested them with the rc1... dont have a real board atm either06:29
lekernelit can be contacted very well even if those components aren't moved... so don't worry06:30
lekernelactually I didn't even notice that before you looked at it, even though I used the jtag pod many times06:30
adamw_well...but fig. 13 shown is not big deal, because jtag connector contacted well to work enough!06:30
lekernelthere are more important issues imo06:31
adamw_lekernel, hi yes, actually the C27 and U1 are not big deal06:31
adamw_just  I recorded first then we improved next time. :)06:31
rohwhat about some ribbon cables with press-on connectors?06:31
lekernelgood to hear you say that :)06:31
rohnot long.. just like 5 or 10cm?06:31
adamw_lekernel, but we just need to determine the J5's height and pick a suitable part.06:33
lekernelyeah, exactly, and just cut it on the existing boards... please don't delay anything because of this trivial problem06:33
adamw_well...we need to think this carefully...I don't want to rework news coming from everywhere later.06:34
wpwrakroh: do you also have press-on connectors for single-line headers ?06:34
adamw_trimming or cut for me is the 'rework'. :)06:34
wolfspraulroh: ahh good point, I send you one jtag-serial too, just in case06:35
rohwpwrak: nope.. but who cares.. use a 2row06:35
wpwrakroh: with the jtag board floating freely, i would be a bit worried about it accidently making contact with the main device. but maybe that's acceptable. (i was scared about the same issue with the openmoko debug board)06:36
rohwpwrak: for the 4pin 1row serial i would use something readymade...like a cdrom-audio cable or similar06:36
lekerneljtag boards floating freely are 1000 times more messy than this small connector height problem06:36
rohwpwrak: use sticky tape or gaffa on the pcb backside06:36
wpwraklekernel: maybe 10x :)06:37
lekernelis it so hard to find a shorter header? or to cut an existing one?06:37
wpwrakroh: (ready-made) good idea06:37
lekernelI don't understand you...06:37
adamw_lekernel, we haven't decide to cut (this is unusual/bad idea) or to order the one we found in Molex.06:38
adamw_wolfspraul, how far it is from your home to roh or lekernel?06:39
wolfspraulway too far06:40
rohwolfspraul: how long are you in .de?06:40
wolfspraulprobably also next week06:40
wolfspraulI will try to call that reichl guy, get an appointment.06:40
wolfspraulin that case I may even come back to Berlin.06:41
rohi see06:41
wolfspraulcalling right now, nice huy06:42
wolfspraulhe said 'how about monday?'06:42
lekernelbut doesn't answer email :p06:42
lekernelworks for me06:43
rohuh.. dunno if i am ready to do much outside stuff on monday.. want to get proper un-sick first06:43
Action: roh got 'one' of the hacker-pests... seemingly the easy form.06:44
wolfspraullekernel: I'm pretty sure I don't have to explain personal 'pecularities' here, do I?06:44
wolfspraulso I called him, all super easy. he said come Monday morning...06:44
rohkristianpaul: what does it show? GNSS?06:45
wolfspraulI asked what 'morning' means and he said 7-9 AM :-)06:45
rohyikes. on monday? thats crazy06:45
wolfspraulwell, maybe you guys also just call him? I mean he's in the same city...06:45
wolfspraulif we cannot make progress on this in Berlin, I will go to a lab I know in Nanjing06:46
lekernelmonday 7-9 works for me06:46
wolfspraulhe told me Monday is best for him, totally free right now06:46
wolfspraulTuesday = eye doctor06:46
wolfspraulWednesday/Thursday - very busy06:46
wolfspraulFriday could be OK as well06:46
wpwrakkristianpaul: is this good or bad ? :)06:46
kristianpaulroh: some statics so far, wait a bit more for the PRN matching part06:46
wolfspraullekernel: if you have time on Monday, maybe I come to Berlin?06:47
wolfspraulor you can handle it yourself?06:47
rohkristianpaul: i meant.. whats GNSS?06:47
wolfspraulI'm interesting in this guy and what more he can do for us in the future...06:47
kristianpaulwpwrak: i think is not bad, but i'm missing some data concepts for I/Q sampling theory so the data can be analized again06:48
wolfspraulso maybe I should just come there.06:48
lekernelok, do that then06:48
lekernelwhat personal peculiarities are you talking about?06:48
wpwrakwolfspraul: what can he do ? (in broad strokes)06:48
wolfspraulpeople who have this or that preference06:48
wolfspraulyou write an email - no answer06:48
wolfspraulbut what does it mean?06:48
wolfspraulto me: nothing06:48
wolfspraulI don't know. I must have worked with the strangest people on the planet, I feel.06:49
wolfspraulif someone wants to communicate with flying pigeons - fine.06:49
kristianpauli also it was using made matlab.. dammit,  i can find serios examples/aplications for octave/scilab?..06:49
wolfspraulI will dutifully send my messages back and forth that way...06:49
wolfspraulsome get up at 2 PM06:49
wolfspraulsome start at 6 AM06:49
wolfspraulwhatever, you name it...06:49
wolfspraulwpwrak: I was referred to him to help us with ce/fcc certification.06:50
wpwrakah, good !06:50
wolfspraulI was told he's a cool old guy06:50
wolfspraul1-person business06:50
wolfspraula bit chaotic (floor all filled with stuff)06:51
wolfspraulcharges by the hour, very easily approachable and unbureaucratic06:51
wolfspraultakes tons of time to explain things06:51
wolfspraulsounds like our guy, no? :-)06:51
lekernelgood. I also need to borrow a 10+GHz frequency counter for non-commercial nerd research. maybe he could also help ;)06:51
wolfspraulabsolutely. let's ask! :-)06:52
wpwrakhis equipment list looks rather nice :)06:52
wolfspraulI was referred to him, I have reason to believe this is a solid contact.06:52
lekernelanyway let's go there monday morning06:52
wolfspraulbut let's see, we have to try...06:52
kristianpaulI like that website design :-)06:54
adamw_nice! the equipment lists!06:55
kristianpaulalso expertise06:55
rohlekernel: im asking around a bit about frequency counters... laforge doesnt have one that fast.. but i got a tip06:57
kristianpaulAny one around have the printed copy and CD for the "A Software-Defined GPS and Galileo Receiver" book. (maybe)07:05
kristianpaulOr know somebody which have it?07:05
lekernelwho needs that? http://lea.hamradio.si/~s53mv/navsats/theory.html :)07:06
wpwraklekernel: very old-school graphics :)07:09
kristianpaulvery old heavy stuff too :-)07:10
wpwrakkristianpaul: that's the problem with those ex-commies - they're really really good at math :)07:24
wpwrakHURD loop ;-)07:29
lekernelyeah, where all the beauty of the GNU shines :)07:30
wpwraklekernel: you say it's all bull ? :)07:31
lekernelnot all, but mostly - that's why it's only a gnu07:31
kristianpaulalso gnu is brand ie gnuplot is not gpl or gnu project it self07:33
kristianpaulwpwrak: remenber our talk abot I/Q and 2 bits, (sign & mag), i ask you now, how the USRP handle/format sign and mag?08:19
kristianpaulso if the USRP data ouput is 16 bit I/Q it means there are 16 bits I and Q interleaved? right08:20
kristianpaulsoemthing like: IQIQIQIQIQIQIQIQIQIQIQIQIQIQIQIQ08:21
wpwrakkristianpaul: err no, it's not bit-serial like that. Q and I have different channels in hardware.08:25
wpwrakkristianpaul: the FPGA then puts data into consecutive words08:25
kristianpaulthose words are signed?08:26
wpwrakkristianpaul: i think so, yes08:26
wpwrakhmm, something is coming ... just don't know whether it's three bens or a letter informing me that i have to appear at customs to personally retrieve them. let's hope for the former.09:43
larscor it's a surprise gift09:51
wpwraksurprises coming from customs are rarely good ones ...09:52
kyakwpwrak: thanks for the debwrt link, i'll have a look10:28
wpwrakkyak: if they're as extreme as openwrt, their work may not be directly useful. but they'll have to solve a number of problems that, according to zrafa, still prevent debian from being a good choice for such small devices. so in the worst case, if there are any results at all, they'll make it easier to do something interesting for us10:31
larscafaik it's a debian with openwrt kernel patches10:37
wpwrakoh, nothing more ? i was hoping they'd tackle the big task of splitting packages into smaller chunks as well10:39
wpwrakwell, maybe they're not aware of that problem yet. that would be an interesting surprise. heh heh :)10:40
kyakstill trying to persuade this guy...10:42
kyakwell, at least the image can be built now when libiconv, gettext and glib2 are overriden10:43
wpwrakkyak: i think the two of you are talking about different things10:43
kyakwhy is that?10:44
wpwrakkyak: maybe try to see if you can grab him in IRC. could be easier to converge.10:44
kyakyes, might be a good idea10:45
wpwrakjow says there's no difference. you imply there is and then you explain why the change makes things hard.10:45
wpwrakso it seems that the first question should be why jow thinks there is no difference in function10:45
kyaki told him about the "no difference". then i again explained why it is needed10:46
wpwraki.e., maybe there is indeed something wrong in libiconv that made it functionless ? or maybe he doesn't understand your use case ? or maybe he does understand but rejects it ? many possible interpretations, and progress will be difficult if you can't find out which one it the right one10:46
kyaknow, he said "it is the same, but 10x less size"10:47
kyakthen i shown him that it's not the same10:47
kyakbeacuse i.e. fbterm can't be linked10:48
kyakand because he himself has to unstub it all the time10:48
kyakand i also asked if it was the same, why *-full versions are there?10:48
wpwrakyes, but the action to take depends on why he thinks it's the same while it apparently isn't10:48
kyakthat might be in his next answer10:49
wpwrake.g., if it indeed be a fully equivalent replacement, then the fix would be to enhance the stub accordingly10:49
wpwraker  .. if it is indeed intended to be ...10:49
kyaka lot of things can be chosen in openwrt by configuration10:51
kyakeven enabling wide character support in underlying libs10:51
kyakbut using full versions of libiconv and libintl (and accordingly in all other dependent apps) is not blocked10:51
kyak*is now blocked10:51
kyaka user will have to modify Makefiles for that10:52
wpwrak(many options) seems that there's no good mechanism for having variants for the same functionality, though (or otherwise, you would have used that one)10:52
wpwrak(variants) e.g., package X needs a libc, but doesn't care which one. or an editor. or a shell. whatever.10:52
kyaksomething very bad has happened10:55
kyakfrom my opinion, since those stub libs appeared, openwrt is seriously broken10:56
lekernelwell, it's using GNU software, so it's obviously broken :)10:56
lekernelbut as long as it somehow works, why care...10:56
Action: wpwrak is looking forward to savour the sheer elegance and speed of sebcc10:57
lekerneljust use llvm...10:57
Action: kyak movie time...10:57
kristianpaulI wonder if linux can be compiled with llvm11:01
lekernelno, it's full of GNU extensions and therefore equally broken11:02
lekernelthe freebsd kernel can11:03
kristianpaul(freebsd) oh, so the finally drop gcc?11:03
lekerneli've heard of some effort to mimic the gnu extensions in llvm, but I don't know what the state is11:03
lekernelthey're trying to... I don't know when it'll be the main compiler11:04
kristianpaulSure all we know when, that will be on the news everywhere ;-)11:04
lekernelwhat I love is when the GNUtards blame Microsoft for trying to discreetly put loads of proprietary extensions on every computer11:05
lekernelcf. autoconf, gcc, ...11:05
lekernelautoconf is horrible when used on a gnu system and cannot be named when used on a non-gnu system11:06
lekernelit's full of obscure checks to make sure you're using the GNU stuff everywhere11:06
kristianpaulGNU is the whole thing  indeed :-)11:07
lekerneland it infects the source of many software packages like nasty windows malware11:07
wpwrakwheee ! my UPS dodged a brown-out ! :)11:10
wpwraklekernel: just don't use it ? i never saw a need for autoconf and friends11:12
lekernelI'm not using it, but many software packages do, including libraries and tools that I need and for which there is no equivalent without autocrap11:13
lekernelif GNU/Autocrap was merely self-contained, I wouldn't make such a fuss. but the problem is that shit is everywhere11:14
wpwrakhmm, libs and stuff usually don't leak their autojunk much. at least not in my experience. it just becomes part of the build instructions.11:16
wpwrakthe only real annoyance are those who assume that you know the whole automake, autoconf, configure canon by heart and don't even provide a script to run the whole mess11:17
lekernelyeah, and good luck fixing those when they break (and they DO break, all the time, on non-gnu systems, sometimes clearly on purpose)11:18
lekerneljust how many different and patched autocrap versions you can count in the openbsd ports tree11:19
wpwrakah, interesting. well, happily i'm on linux all the time :)11:19
wpwrakbut it is kinda ironic if autocrap forces you to a specific kind of platform11:21
lekerneland they call that tools to make software portable. this is not true. they're GNU propaganda tools, which attempt to conceal the fact that they make porting software hard on non-GNU platforms by making you believe software is hard to port in general11:21
wpwrakand soon we'll find out the truth behind JFK, 9/11, and most of all, Elvis :)11:22
lekernel"let's check that the cos function, mandated by many standards, exists in libm - and do that in a way that stresses either the shell, compiler, linker or both, so it can break on a non-GNU system"11:23
lekerneletc. etc.11:23
lekernelautocrap is all about those (and puking out shell scripts that are more unreadable than the blobs the FSF makes so much fuss about)11:24
wpwrakwell, such breakage may not necessarily be deliberate. i've quite often run into quirky environmental issues i hadn't anticipated. and propaganda was the last thing on my mind ...11:25
lekernelwell maybe11:25
lekernelnever attribute to malice that which can be adequately explained by stupidity ;)11:25
wpwrakyeah :)11:25
wpwraka particularly nasty source of breakage is l10n. even seemingly harmless things like  for n in [a-z]; ...  become unreliable11:27
mthlekernel: there is a modified gcc that emit llvm bytecode, using that it should be possible to compile sources that use GNU extensions and still use llvm for the final code generation11:31
lekernelthanks. but I use clang and no GNU extensions - except those that are implemented on other major systems; sometimes they manage to do things that don't suck technically, like asprintf()11:34
mthyes, that's better for new code, but it can be useful to have an alternative for existing code11:35
kristianpaulHow could be easier add 6 bits to a 2bit data in order to be easilly processed?..11:59
kristianpaulalso that bloat a the whole thing..12:00
wpwrakhow about worrying about optimization after the algorithm works ? :)12:05
kristianpaulyes, it just was a suguesstion from the guy who is helping me to analize the data12:09
wpwrakah, i see12:12
kristianpaulhe said google python unpack for more details...12:13
bartbesdoesn't it make more sense to stuff 4 'data units' in a byte instead?12:14
kristianpaulthats how it is right know12:15
kristianpaulbut seems i hard to process data wich is orinally 2 bits wide in that way..12:16
kristianpaulI dont know i need read more about it12:16
bartbesbitshift to the proper position12:16
bartbesthen %412:16
bartbes4? no12:17
bartbesor wait..12:17
bartbesyes, 4, I was right after all :P12:17
wpwrak& 3 :)12:17
bartbesjust bitwise12:18
wpwrakuint8_t in, out[4]; for (i = 0; i != 4; i++) out[i] = (in >> (2*i)) & 3;12:18
wpwrakno need to figure out fancy packing functions :)12:19
qi-bot[commit] Werner Almesberger: atusb/fw/include/at86rf230.h: started updates for AT86RF231 http://qi-hw.com/p/ben-wpan/337e5d213:02
qi-bot[commit] Werner Almesberger: atusb/fw/include/at86rf230.h: make one section per register, not one per field http://qi-hw.com/p/ben-wpan/bd5b00813:02
qi-bot[commit] Werner Almesberger: atusb/fw/include/at86rf230.h: added remaining AT86RF231 values http://qi-hw.com/p/ben-wpan/4387d8413:02
qi-bot[commit] Werner Almesberger: atusd/cam/Makefile: updated for cameo-based process (forgot to commit this) http://qi-hw.com/p/ben-wpan/4ef7a8213:02
qi-bot[commit] Werner Almesberger: libatrf: new function atrf_identify to identify the chip http://qi-hw.com/p/ben-wpan/a9321ce13:02
qi-bot[commit] Werner Almesberger: atrf-txrx: set CRC and transmit power depending on chip (AT86RF230/231) http://qi-hw.com/p/ben-wpan/02e394a13:02
qi-bot[commit] Werner Almesberger: atusb/fw/include/at86rf230.h: many corrections, some additions http://qi-hw.com/p/ben-wpan/f67a79b13:02
qi-bot[commit] Werner Almesberger: atusb/fw/include/at86rf230.h: make 231-centric http://qi-hw.com/p/ben-wpan/309a5ed13:02
qi-bot[commit] Werner Almesberger: atrf-txrx: added support for AT86RF231 continuous transmission mode http://qi-hw.com/p/ben-wpan/b2d90cd13:02
wpwrakhmm ... no notification for the commits i just pushed ?13:05
mthmaybe the flood protection cut them off?13:05
wolfspraulwpwrak: what notification?13:08
wolfspraulthere were 9 commits that just came through here... (6 minutes ago)13:09
wpwrakwolfspraul: these are the ones. only took unusually long.13:24
wpwrakmeanwhile ... wpwrak->bens += 3; /* fedex just brought them :-) */13:24
wolfspraulbut the commits showed up 2 minutes before you posted that they are missing?13:48
wolfspraulmaybe something in the irc servers?13:48
wpwrakoh, that's odd then. probably irc indeed.13:49
wpwraki waited 1 minute after the push before asking about them13:50
wolfspraulyes, 3 minutes on irclogs...13:50
wpwrakheh. the laws of relativity also apply in irc, it seems :)13:50
wolfspraulroh: for the m1 case, how can I tell which side of the top and bottom plates are meant to be inside, and which one is outside?13:53
wolfspraulit looks like one side of the edge is a bit rounder, maybe that's the one that should point outwards...13:53
wolfspraulhey I just realized - the power connector says "DC 5V IN", maybe I would add "TIP +" if there is enough space.14:01
wpwrakis the ubiquitous "disc in arc" symbol encumbered ?14:03
wpwrakzrafa: hmm, why do i need to  apt-get install libconfuse0  in order to install jlime in NAND ?14:06
wpwrakor is this a (missing) dependency of xburst-tools ?14:06
wpwrakzrafa: (hwiki/JlimeDownloadsAndInstallation )ow about putting the s/NAND_FORCEERASE... in the section about installing xburst-tools ?14:09
wpwrakzrafa: also, why the 201002 version ? that one has at least two major known bugs (one parser bug, and the u-boot-killing nand erase wrap made famous by debian)14:11
wpwrakzrafa: at least in 201012, NAND_FORCEERASE is already set. one item less to worry about;-)14:17
zrafawpwrak: it was a requirement for xburst-tools, maybe you do not need if you have another version14:20
wpwrakChecking 454656 bytes... Comparing 454656 bytes - FAIL at off 2048, wrote 0x0, read 0xff  :-(14:22
wpwrakso far, so bad ... let's try another one ...14:23
wpwrakgrmbl.the same with #214:26
wpwraklet's try xburst-tools 201007 ...14:28
wpwrakthe same. but the device was already booted. rebooting ...14:29
zrafawpwrak: NAND_FORCEERASE: yeah, that should be on xburst-tools section.14:30
zrafainstalattion section14:30
wpwraksame error. very peculiar14:31
lekernelwolfspraul: you'll love this: http://www.renesasrulz.com/community/rx-contest14:35
wpwrakthey must be SO desperate ;-)14:37
wolfspraulhmm, I don't fully get it. but ok.14:39
wolfspraulthe limited experience I had with Renesas was positive.14:39
wolfspraulbut I think the Japanese in general have their whole own take on the proprietary/free/secrecy thing, on their cultural (Japanese) background.14:39
wolfsprauland I have no clue about that, so it's kinda mysterious :-)14:40
wpwrakinteresting. didn't usbboot -c ... but instead brought up usbboot interactively and pasted the commands individually. this time it worked.14:40
wpwrakzrafa: "The system will not boot if there were problems, and it will be harder to get usbboot again." euphemism of the year 2010 ? ;-)14:41
lekernel"Trust us, YOU WANT THIS KIT. IT'S COOL!"14:42
wpwrakthe nice thing about this flash failure is that it leaves just enough of the system intact that reset+U still works14:42
lekernel(from http://www.renesasrulz.com/community/rx-contest/instructions)14:43
wpwraklekernel: guy at a bar, to a girl "please please sleep with me ! i'm the coolest dude !" starts crying "PLEASE ! PLEASE !"14:44
lekernel"Demonstration Kit featuring the RX62N series 32-bit MCU, which has on-chip flash memory and enhanced communication functions, including an Ethernet controller, USB 2.0, and CAN. " seems rather boring to me. except that I could use the CAN to talk to that stupid black box they put in my car and is causing me trouble atm14:44
wolfspraul"login or register to RenesasRulz"14:44
wolfspraulI think that's really cute actually. They try to be hip...14:44
wolfspraul"But wait, there's MORE..."14:46
wpwrakxiangfu is on ia32 or x86-64 ?14:46
kristianpaulwpwrak: you meant owrt toolchain?14:47
lekernelspeaking about CAN and ODB2, is there any open source solution to fix car calculator problems?14:47
lekerneldidn't find any :(14:47
wpwrakkristianpaul: naw, xburst-tools. seems that i found another flashing bug. (see above)14:48
lekernelgoogling these keywords yields craploads of forum posts without much brilliance and dubious products14:49
wpwrakzrafa: i think jlime still needs the  touch /etc/shadow14:52
wpwraki wonder how hard it would be to have the "nice" console font right after booting15:00
wpwrakwell, after reset, during boot15:01
wpwrakzrafa: grmbl. how do you exit the audio player ? :)15:07
kristianpaulwpwrak: esc, alt + f4?15:12
kristianpaulah gmu15:12
kristianpaulf1 then alt + enter15:13
wejpyou don't need to press f1 first15:13
wejpit is always alt + enter, unless the keymap has been changed of course15:13
wejpin the new version i have changed it to Alt+Q, because so many found the alt+enter so confusing15:14
wpwrakwejp: ah, alt-enter works. thanks !15:15
wejpoh and you can always change it to something else, if you don't like it. just edit the nanonote.keymap file15:15
wpwrakopkg install ldd15:30
kristianpaulwpwrak: they guy is helping me (Fabrizio) is using a python code to analize the data, i pushed him to realize the code publically so 2 bits support  can be added15:35
kristianpaullet's see how it goes15:35
kristianpaulNow time to increase a bit the fpga buffer size :D15:36
wpwrakgaah. those things really suck without counterweight ! un-be-lief-able that anyone would even think of selling such crap :)15:42
wpwrakthere, fixed :)15:55
kristianpaulzrafa: Hey how are you?16:18
kristianpaulI was trying the toolchin guide for jlime but i end with this http://paste.pocoo.org/show/316805/16:19
kristianpaulI must crosscompile due it seems something is ran gcc on the SIE is just reset the ethernet over usb link :/16:19
kristianpauloh solved16:26
kristianpaulsome symbolic links.16:26
kristianpaulneat, crosscompile works very well16:36
kyakis pretty awesome17:34
kyakwhat did you use for visualization?17:35
wpwrak(the wlan is pretty far away, that's why it's a little weak)17:35
kristianpaulwpwrak: neat !17:35
wpwrakatrf-rssi uses sdl-gfx17:35
kristianpaulthe visualiation is kind retro, (wich make it more awesome i think)17:35
kristianpaulI see you're runing Jlime ;-)17:36
kyakwpwrak: what's the current status of wpan? is it working and you can send/receive data?17:36
kristianpaulwpwrak: can you zoom in on screen a bit more?17:36
wpwrak(jlime) of course. i lack the youthful energy to mess with openwrt :)17:37
wpwrakkyak:i'm currently tweaking the RF behaviour. send/receive worked already a very long time ago. but there's no stack yet - just a utility to send/receive "bare" frames17:38
wpwrakkristianpaul: i'll take a picture just of atrf-rssi ... actually, i can do that on the PC. easier :)17:39
kyakwpwrak: i really hope your work will get implemented in some future versions of NanoNote (and soon). I lack wireless capabilities so much :)17:40
kristianpaulwpwrak: nice work on gui. it even have shadow17:42
wpwrakkyak: the idea is to make a few experimental boards soon. their rf probably won't be optimal, but they should be good enough for playing.17:42
kristianpauloh i remenber know you draw the numbers your self17:42
wolfspraulthat is such a cool video!17:42
wolfspraulI assume it's not fake, right? :-)17:43
wpwrakkyak: oh, and i'm forced to switch to a different chip. i didn't like this idea at first, but the new chip (which is very similar and nearly pin-compatible) has a number of nice features. including a high-speed modes where it gets up to 2 Mbps. real throughput with this may be in the order of 0.5 Mbps.17:43
wpwrakwolfspraul: it's all real ;-)17:44
wolfspraulso that measures live traffic as you see it on the antenna/the rf chip?17:45
kyakwpwrak: this is all very cool, the only issue i think can be that people don't have wpan-enabled access points to connect Ben to, am i right?17:45
wolfspraulwhat is it useful for?17:45
wolfspraulkyak: first, they can have a second NanoNote :-)17:45
wolfspraulor, we are nice, we are making both 8:10 card and usb versions17:45
wolfspraulso you can have one in your (Linux powered) notebook17:46
kyakoh yes, and then one NanoNote can be connected via USB and "share" Internet with another :)17:46
wolfspraulor router, for that matter17:46
wolfspraulcorrect :-)17:46
wolfspraulbut there's a usb variant17:46
kyaki see, it sounds so cool17:46
kyakis there any schedule, even approxiamte? :)17:47
wpwrakwolfspraul: this just shows the signal strength. it doesn't actually demodulate or decocde things.17:48
wpwrakwolfspraul: the purpose is a) just to for me to verify things, abn b) to have an idea where there is a quiet place in the local spectrum17:49
wpwrakkyak: the USB board will be similar to this one: http://downloads.qi-hardware.com/people/werner/tmp/nemesis.jpg17:50
wpwrakkyak: minus the cable, and ~2 mm longer17:50
kristianpaulwpwrak: in the video the only Ben with wpan was that?, i mean you just we're measuring your place?17:50
wolfspraulsure for finding quiet spectrum it's nice17:51
wpwrakkyak: also minus the connector at the bottom. that one's for the initial flash load (there's a microcontroller on the board)17:51
wpwrakkristianpaul: yes, and then i let a laptop (about two rooms away) hammer its channel for a few seconds17:52
kristianpaulwpwrak: you should attach a gps (wich is easy btw) go give a walk around neigboorhood and map spectrum as well17:52
kristianpaulI mean at least first floor measurements ;-)17:52
wpwrakkristianpaul: hah, i can just do a wifi scan. then i know pretty well what's crawling around here. i don't think there are too many people who are emitting anything different from WLAN at 2.4-2.5 GHz.17:54
wpwrakwell, maybe microwave ovens. mine has enough leakage to knock out the wlan (already operating at its limit)17:54
kristianpaulok ;-)17:54
wpwrakkyak: estimate ... hmm, around 2-month-ish, i'd say. i need to test the new chip. then a few RF improvements in the USB board. next, see how the one for the ben behaves (haven't done any "good" measurements of that one yet)17:56
wpwrakkyak: then i can send some samples to wolfgang. we also need to figure out how to do initial flashing and the testing.17:57
wpwrakadd one "the"17:57
wolfsprauland I hope we find time for some boom and other process innovation17:58
wolfspraul(on my end too, I have a nice set of todos I want to get done before/with ben-wpan)17:58
wpwrakwhat would they be ?17:59
wolfspraulin boom: mouser18:00
wolfsprauluse the new cmdline options to auto-generate some files18:00
wolfspraul(kicad based)18:00
wpwrak(mouser) very good18:00
wolfsprauldocument the process well, and ideally little by little automate it more, with Makefiles etc.18:00
wpwrak(new files) have you seen that i already do that ? (in ben-wpan/bom/)18:01
wolfspraulso the path from projects server to actual run becomes better documented and better automated18:01
wolfspraulall not without reason, i.e. we want to actually produce the boards asap, but we also want to invest a little for the future18:01
wolfspraulthe usual mix that we do already for 1.5 years :-)18:01
wpwrakyeah :)18:01
wpwrakregarding the process, i think the way boom currently interfaces with the world around it is a little too hardcore for most people18:02
wolfspraul(I'm just explaining this to kyak so he has an idea for the schedule, I think werner and I are already mostly on the same page)18:02
wpwrakit also lacks transparency. e.g., it's difficult to track why boom did this or that. so there's still quite a bit of work waiting.18:03
wolfspraulsure. if you can cut out a little improvement and get it done, that's great.18:03
wolfspraulI have my share to do, need to dig deeper...18:04
wpwrakso i wouldn't "sell" boom in its present state as a solution. more as a tool you can use to provide a solution, but ideally without making people spend time to figure it all out themselves (unless they like the challenge, of course ;-)18:04
wolfspraulthe direction is good18:04
wolfspraulwe just need to keep improving little by little18:04
wolfspraulif we improve nothing, it's bad. if we try too much at once, we will fail as well.18:04
kyakwpwrak: it's very promising and nice! and USB board seems to be very small18:05
wpwrak(little improvement) my current plan is to just patch it up if necessary, so that it doesn't fall apart in mid-flight. the big cleanup will need a re-implementation. everything in C, so that the data structures are cleaner, with interactive browsing of things.18:06
wpwrakin particular, i want to bring it closer to kicad. make it part already of the design process, not something that runs like an afterthought18:06
wpwrake.g., when i add a resistor, i'd want to select the symbol in kicad and then let kicad send the data it has on that symbol to the bom processor (this interface could and should be generic and relatively simple)18:07
wpwrakin the bom processor, one would then refine the query until something suitable is found in the catalog. then that data could be sent back to kicad.18:08
wpwrakthis would replace the current (manual) eeschema-digikey loop and the eeschema-boom loop when adding hints to steer boom in the right direction. you could of course still do things in this way.18:10
wpwrakthe interface on the boom side could be something similar to the parametric search in the digi-key catalog. once you've selected something, boom could put the parameters in the clipboard, and kicad could retrieve them from there and complement the symbol information18:11
wolfspraulok you have the best overview18:12
wolfspraulI will just add things on the surface, like mouser.18:12
wolfspraulthat's a good contribution18:12
wolfspraulyou need to pull the core forward for the time being18:12
wpwrakmouser will be great. i'm sure there will be surprises :)18:13
wpwrakby the way, another interesting project would be making the pcb equivalent of schhist18:14
wolfspraulfor sure, but I don't think I get to it the next 1-2 months or so18:15
wolfspraultoo many things already18:15
wpwrakhehe :)18:15
wolfspraulI need to spend some time to get the cmdline stuff upstream18:15
wolfsprauldon't want to leave too much one-off garbage behind18:15
wpwrakyup, that's important18:16
wolfspraulbrdhist is definitely on the wish-list, I am hoping you start it one day :-)18:16
wpwrakagreed. that's just maintenance nightmares piling up18:16
wolfspraulif not, I am not against taking up the shovel, it's just not high enough in the priority list yet18:16
wolfspraulwe do need it, for sure18:16
wolfspraulI think a proper boom is even more valuable though.18:17
wpwrakdifficult comparison :) they have different audiences and different complexity18:18
wolfspraulwhat they have in common is only that both are needed for a proper process18:19
wolfspraulfor totally different reasons18:19
wolfspraulschhist is there, and robust, and working. so I consider that one 'done' for now...18:19
wpwrakyeah. schhist is a useful tool already. what i should add some day is a version/mode in which it just compares two revisions locally. like git-diff18:21
wpwrakright now, i sometimes commit stuff, wait until schhist has updated, and then check on the web if i guessed my changes right ;-)18:22
qi-bot[commit] kyak: ash --login as login shell in fbterm (in gmenu2x) http://qi-hw.com/p/openwrt-packages/e7e18ea18:41
qi-bot[commit] kyak: small optimization when starting ben-ash-rus http://qi-hw.com/p/openwrt-packages/a1c1ed318:41
bartbeskyak: I thought gmenu2x was fixed so --login wasn't needed anymore?18:44
kyakit is required when starting fbterm18:45
kyakcause otherwise fbterm would start /bin/sh18:46
kyaktherefore, it would not read profile, and all env and aliases etc are empty18:46
bartbesso it's an fbterm issue18:48
Josh478Anyone around?19:23
wpwrakkristianpaul: for some future studies ;-) the 231 actually does support antenna diversity and external RF front-ends20:26
kristianpaulwpwrak: oh nice to know that ! [external RF front-ends]20:37
wpwrakof course, you would have to design a new board for that ... :)20:38
wpwrakhere's antenna diversity: http://www.atmel.com/dyn/resources/prod_documents/doc8158.pdf20:40
wpwrakalas, the document about an external rf frontend isn't accessible at the moment. not sure why.20:40
kristianpaulThis will be my next goal http://www.zarlink.com/zarlink/hs/82_GP2021.htm20:43
kristianpaulnot the IC but what it does20:43
kristianpaula good deal will be a small factor fpga with low consume or power saving profile20:44
kristianpaul(talking about sige rf zigbee-like)20:44
kristianpaulhmm, there is a way to pass data between different vim processes20:47
wpwrak:w /tmp/xxx20:48
wpwrak:r /tmp/xxx20:48
wpwrak(next goal) yup, correlation ... should be fun :)20:48
kristianpaulwell i cant address more that 4096bits, so i need a read pointer now.. (if i consider increase buffer size)22:56
zrafawpwrak: dont use opkg.. use jlime-pkg23:22
zrafawpwrak: audio player: yeah.. alt+enter is not so intuitive23:22
--- Sat Jan 8 201100:00

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