#qi-hardware IRC log for Saturday, 2012-04-14

qi-botThe build was successful: http://fidelio.qi-hardware.com/~xiangfu/build-nanonote/openwrt-xburst.full_system-20120412-1915 00:01
yossdoes anybody know how  to calculate PER out of SNR values?08:58
DocScrutinizeranybody heard sth of mickeyl? it's over 2 months now that he last showed up09:25
DocScrutinizer~seen mickeyl09:28
infobotmickeyl <~mickey@openmoko/coreteam/mickey> was last seen on IRC in channel #openmoko-cdevel, 72d 11h 23m 56s ago, saying: 'gnight'.09:28
wpwrakhave you tried sending him am mail ?10:35
DocScrutinizerwpwrak: sent him a SMS11:13
DocScrutinizernot sure about the right number though11:14
DocScrutinizerwell, the number once must've been the right one, and the SMS got delivered11:15
wolfspraulI think he wrote it on his blog, no?12:27
wolfspraulhe is busy with his ios development firm and taking care of family12:27
wolfspraulyou should try to contact his company about wanting to develop a new ios app, or applying for a dev job (he hired his 3rd developer recently, according to blog)12:28
wolfspraulI don't know more than the blog, but the blog is pretty clear to me...12:28
DocScrutinizerwolfspraul: <mrmoku> 05-04-2012 12:26:12 < mickey_office!~mickey@e182221015.adsl.alicedsl.de: we're finishing a bunch of apps this month, so i hope things will slow down soon12:39
DocScrutinizerso everything fine"12:39
DocScrutinizerjust not very nice to scare people like me with not showing up on IRC at all for >2 months12:40
wolfspraulsomehow it seems I understood the situation quite well :-)12:41
wolfspraul1. family is priority #112:41
wolfspraul2. making money, in his case with ios apps, is priority #212:41
DocScrutinizerno priority #1 could stop me from cranking up IRC at least once or twice a week, if only to idle12:41
wolfspraulif open harwdare fails, I try my luck with a dev job at mickey's :-)12:42
wolfspraulok but you are special12:42
DocScrutinizerno, mickey is special - he's #openmoko GC12:42
wolfspraulexactly :-)12:43
DocScrutinizerwell, if he doesn't care at all anymore, then he should nominate a co-GC12:44
wolfspraulmaybe he can join qi in a few years12:44
wolfspraulwhen the kids are in school12:44
DocScrutinizeror pass the job to somebody else completely12:44
wolfspraulany takers? maybe he would be happy if someone takes over...12:44
wolfspraulI don't know12:45
DocScrutinizerI don't mind12:45
wolfspraulhow many active gta01 and gta02 users are still around?12:45
DocScrutinizererr, #openmoko != gta0*12:45
DocScrutinizeror - to put it simple and generic - openmoko.org >> openmoko.com12:47
DocScrutinizerand for IRC all this doesn't matter12:49
DocScrutinizeruser counts is all that matters12:49
DocScrutinizerfor all the #openmoko* channels12:50
DocScrutinizerthe GC owns all of them12:50
DocScrutinizeranyway, scratch that all - all caused by mickey's annoying habit to change nickname12:59
DocScrutinizerand then not even registering to use his common cloak13:04
wpwrakiOS firm ? so DocScrutinizer's fears that something horrible and tragic may have happened to him were justified13:15
viricwow, linux.com. congratulations wpwrak !13:55
wpwrakthanks ! :) nice article indeed13:55
wpwraksome of these days, i should finish my work on the atusb driver :)13:56
wpwraki mean it works. just not very well.13:57
viricwpwrak: can't you get kernel programmers to do this?14:06
wpwraksure, someone else could do that too. it's both ends that need to change, though - atusb firmware and kernel14:08
viricthe fw is the 8051 code?14:09
wpwraki have some almost-finished code. just need to find the time for a context switch.14:09
wpwrakit's AVR14:09
wpwraknot that you'd notice much of a difference. since it's written in C anyway14:10
DocScrutinizerreal men code in FORTRAN. If it can't get done in FORTRAN, do it in assembler. If it can't get done in assembler, it's not worth being done at all14:12
DocScrutinizerotoh the real programmer can write FORTRAN code in any programming language, even C ;)14:14
DocScrutinizeron a more serious comment: have you checked how good assembler from C is for AVR?14:15
DocScrutinizers/assembler/machine code/14:15
DocScrutinizerwpwrak: ^^14:18
lindi-DocScrutinizer: you are asking how good avr-gcc is?14:20
DocScrutinizerkinda, yes14:20
wpwrakDocScrutinizer: i optimized C to assembler on some occasions on AVR. i could usually get efficient code. but it needs a bit of effort. picking the right register sizes, ordering code the right way, etc.14:21
lindi-at least in the past it did not have separate address space support yet, I think this was changed recently14:21
lindi-other issue was that it created quite a lot of extra overhead for interrupt handlers easily14:22
viricthere is that c compiler that can do both 8051 and avr, no?14:23
viricsccs or something like this...14:23
viricah, no more AVR, they say.14:23
IerthHi, sorry for my bad english. I'm a quite new user of nanonote and I can't reflash it because I can't install xburst-tools : a dependance don't want to be compiled. Can you help me please ?14:23
viricpost the error14:24
Ierthgtype-desc.c:9164:31: erreur: subscripted value is neither array nor pointer nor vector14:25
viricwhere did you get xburst-tools from?14:25
viric(to check that exact file)14:25
DocScrutinizerI frequently get baffled when running Lauterbach against ARM code generated from C with RCT (RVT?) compiler. 14:25
IerthSorry, the dependance is cross-mipsel-linux-gnu-gcc. I'm on Archlinux, sow i get it and xburst-tools from AUR.14:26
DocScrutinizersometimes it's next to impossible to understand which line of c code the current instruction belongs to14:26
lindi-gcc, gdb and systemtap at least understand line numbers14:27
DocScrutinizersure, but what to do when your line 800 gets interleaved partially between lines 830, 850, and 910?14:28
lindi-DocScrutinizer: you make smaller functions :)14:28
DocScrutinizerso the compiler can inline them later on?14:29
DocScrutinizerwhich it does a lot14:29
viricIerth: ah, that file is in the mipsel gcc, or in xburst-tools?14:30
viricIerth: maybe you should contact the maintainer of that AUR thing. All I can do is to help you install the nix package manager, and have xburst-tools from nix.14:32
viricbecause I use every oportunity to make people use nix14:32
IerthHum... I think it's just a file which need xburst-tools. Yaourt say me (in french) : ==> Dépendances de xburst-tools : - libusb (déjà installé) - confuse (paquet trouvé) - cross-mipsel-linux-gnu-gcc (compilation depuis AUR).14:33
IerthBut I don't know nix, what is it ?14:34
virica package manager14:34
virichm don't you understand what fails?14:35
viricI mean... I thought Arch users were aware of compilation troubles, building packages, ... :)14:35
IerthHmm, sometime i understand, but not now. u_u14:36
viricdo you know C?14:37
IerthI'll ask on archlinux's channel. I know C, but not enought to understand the error line I pasted you.14:38
DocScrutinizerlindi-: in fact the compiler does the equivalent to code, of transformation (a+b)^2  -> a^2 + 2ab + b^214:38
viricthe error line is not enough14:38
viricwe'd have to see the contents of that file, and maybe of other files too.14:38
DocScrutinizerthen on the 2nd form it does weird optimizations and spreads it all across the code14:38
DocScrutinizeramazing optimizer14:39
IerthUrf. I'm not used to go so far when some installation falls. I generally search the package somewhere else or search a solution on Internet.14:41
virichm ok14:43
wpwrakDocScrutinizer: that's exactly what compilers should do :) gcc with SSA should also do things that are pretty non-obvious. SSA = http://en.wikipedia.org/wiki/Static_single_assignment_form14:43
viricIerth: well, it's hard to help without detailed information about your system and what is it doing at installation time (versions involved, etc.)14:44
lindi-llvm also does quite many interesting optimizations14:44
viricIerth: maybe you should find some good willing person that uses Arch linux, reproduces your problem and fixes it.14:44
DocScrutinizerwpwrak: :-D14:47
IerthHmm, maybe. Do you want a pastebin about the installation (maybe you will see what file crash the installation) ?14:48
DocScrutinizerI probably should get used to using -O0 or whatever when debugging code at work14:49
viricIerth: but I'd need the exact version too14:49
viricIerth: if that's included, fine14:49
viricbut the fix will involve changing the AUR recipes, sure14:49
viricAnd I've no idea about how to do that14:49
IerthOk, the compilation is very big, I'll post just the 5-6 lines before the error. Programm with version : cross-mipsel-linux-gnu-gcc 4.6.0-214:55
viricthe compilation of what?14:58
IerthI'll try to contact the person who made this package to tell him the error.14:58
viricno, make the full paste if possible14:58
viricah, that's better14:58
virichave you heard about electricity counters 'Pachelbel'?15:00
IerthVery big compilation, my terminal can't contain all. So, wait a little please, I will do again it.15:01
Ierth*"I will do it again" I guess.15:03
Ierthviric: tada http://pastebin.archlinux.fr/pastebin.php?dl=43887415:20
viricIerth: there is no error there15:21
viricIerth: I think it cut the file.15:22
Ierthviric: Heck ! You're right, it's cut, the file is too big. So : http://www.toofiles.com/fr/oip/documents/log/cross-mipsel-linux-gnu-gcc.html15:25
viricah, it's an error in the recipe for gcc...15:27
viricThat's really a job for the Arch linux people.15:28
viricunrelated to xburst-tools15:28
IerthUnrelated to cross-mipsel-linux-gnu-gcc too ?15:31
viricno. it's an error in that pkgbuild15:32
viricno = yes, it is related15:32
IerthThanks very much. And I saw a broken link in the wiki. Here : http://en.qi-hardware.com/wiki/Xburst-tools. The link must be http://aur.archlinux.org/packages/xburst-tools/PKGBUILD instead of http://aur.archlinux.org/packages/xburst-tools/xburst-tools/PKGBUILD.15:38
viricIerth: you can update it15:42
viricOr anyone with login could. :)15:42
IerthI'll create an account so. :)15:43
viricthank you!15:43
IerthIt's nothing.15:44
viricwpwrak: you have the judgment of Videla...15:46
viricwpwrak: here the stupid right wing government is menacing Argentina due to the Repsol YPF nationalisation15:47
wpwrakviric: of course, what they're doing with YPF isn't particularly bright either ...16:09
Aylahi guys16:20
Aylais it possible to flash u-boot from the Nanonote itself?16:20
Aylaprovided I have access to the MTD device on the kernel16:20
larscsince the first few blocks need a special layout it doesn't work i think16:25
viricwpwrak: who are they'16:28
viricwpwrak: Repsol is the big shame of the country. But our fascist government things it's of great value.16:28
Aylalarsc: booboo made that: http://code.google.com/p/dingoo-linux/source/browse/trunk/nandfix/nandfix.c16:30
Aylathat was supposed to create a working image to be flashed with usbboot, I think16:30
wpwrakviric: "they" = the argentine national government 16:33
viricwpwrak: I don't know the details.16:33
viricwpwrak: but that made the Spanish gov say "Any attack against Repsol is an attack against the Kingdom of Spain!"16:34
viricShameful, because most company owners aren't event from Spain16:34
viricAnd of course has nothing to do with the citizens of Spain.16:34
viricBut here we have very easy paths between the companies linked to Energy, and the Government16:35
viricWhen someone in the government finishes his charge, goes working in an Energy company as 'adviser'.16:36
wpwrakyeah, that's pretty common everywhere :)16:36
viricno matter the ministry. Education, Health, ... :)16:37
wpwrakbut what else but make threats can they do if the argentine government is basically trying to confiscate YPF ?16:37
wpwrakthey justify it with YPF not investing enough. of course, with government-controlled prices, government-imposed restrictions on currency export, and so on, it's not surprise that investment would be slow16:38
wpwrakalso, none of this bothered them during the last 9 years. why suddenly now ?16:39
viricI don't doubt there have been very doubtful movements in the acquisition of YPF by Repsol16:39
viricHere this happens regularly... when the right wing comes to power, they start up selling public owned assets.16:40
viricWhich "of course" then they are unrecoverable (unless they bankrupt and have a big debt hole the goverment has to pay)16:40
wpwrakwell, YPF and repsol went through several iteration. and the "sell all assets" policy ended well over a decade ago.16:41
viricwell, I'm deciding more on prejudice of other movements I've read from Repsol, than on the concrete study of this case.16:41
wpwrakviric: nothing in argentina's current economical policy is deserving of your sympathy. after the elections, they went from dubious and awkward to stark raving mad16:54
wpwrakviric: e.g., one recent stunt was to institute a de facto ban on the import of books and other printed material17:02
wpwrakviric: of course, they didn't cal it a ban. instead, they came up with the idea that the lead content of the ink would have to be checked before books, journals, etc., can enter the country17:03
mthAyla: afaik the issue is that the Dingoo has 4K flash pages while the JZ boot code can handle 512-byte or 2K pages17:08
wpwrakviric: even better, if these things were sent by courier, there would be only one customs checkpoint where they all are held - the international airport of buenos aires. so if you, say, live in mendoza, you'd have to travel some 1000 km to buenos aires to do the customs procedure to get your books17:08
mthto work around that, the first half of the 4K page contains 2K of data and the second half contain the OOB area17:08
mth(so most of the second half is unused)17:08
wpwrakviric: following the angry reaction of the literate class, the government eventually decided to drop that thing. although they didn't exclude bringing it back later17:10
Aylamth: nandwrite can write a "raw" image without adding OOB data, so maybe we could prepare an image with a custom OOB layout17:11
mthit's no problem if the OOB data for the 4K page is written as usual17:11
mthit's just that you have to package a 2K page + its OOB and put that inside the data area of a 4K page17:12
mthI don't know what OOB format the boot code expects, but booboo must have figured that out17:12
Aylait's on the page I linked earlier17:17
viricwpwrak: I have one friend that bought by postal mail a Geiger counter... and imagine, it even has a 'calibration body' :)17:19
viric(2nd hand)17:19
viricso you have a thing boxed in lead, that can be opened to calibrate the counter17:19
viricwpwrak: well, democracy is about some actions and reactions, and you can't expect it to be most efficent making everyone happy17:20
viricas we use to say here: "we all know that dictatorship is more efficient"17:21
viricI'm amazed that your goverment gave a peny on the reaction of 'the literate class'17:21
wpwrakviric: (boxed in lead) ah, thought it would be a bag of Pu-238 powder ;-)17:22
wpwrak(literate class) yeah, it's not usually their style to back down from mistakes17:22
virichere they would laugh at any reaction :)17:32
virichaving won the elections, here means "the democracy gives us full power for four years, not asking any opinion until the next elections"17:33
wpwrakoh sure. here too. they don't like public protests they didn't organize themselves, though.17:35
whitequarkbut what exactly would "kill" the motherboard here?17:47
whitequarkspurious currents? electrolysis byproducts? and what'll be if I would just wash them off?17:47
rohhey there18:30
roh .oO(wasnt argentinia bankrupt? and spain... well they are soon too.)18:31
rohin bankrupcy things do not need to make sense neccessarily.. check out berlin. capital of germany and still bankrupt for over a decade18:32
viricwpwrak: here the police is getting stronger weapons against "attempts to destroy the established public order". This week they changed the laws so "pacific resistance to police orders" is considered "attack to an authority".19:00
viricwpwrak: and "organizing massive public meetings for possibly violent acts" changed into "belonging to a criminal organization".19:01
viricThey say they would make "vandalic acts" to be treated as "terrorism".19:02
viricAnd I think Spain has such shameful laws, like "torture" being allowed if it is "against terrorism".19:02
viric(vandalict act = burying a dust bin on the street)19:02
viricwhitequark: maybe it has to do with the materials of some capacitors/resistors, or having some electrolytic having inverse polarity, ...19:05
viricbah, it's my blind guess. who knows. :)19:06
viricOf course, if any electronic gets wet, the first thing to do is to remove any power supply, be it cable or battery19:06
viricwhitequark: E>BO MB> =5 >4=0 :0?;O19:07
DocScrutinizerwhitequark: electrolysis creating ions which are radicals (H+, OH-  iirc), both quite aggressive21:21
DocScrutinizerif you wash dirt (oxidized metal) off before anything got rotten fubar, then dry it, usually all shall still work21:22
DocScrutinizerI do that frequently21:23
DocScrutinizerthe "trick" is to remove any power before permanent severe damage happens, then thoroughly wash off all dirt with water and other solvent liquids, then the most "tricky" part: dry it all thoroughly. washing with 99% alcohol as last rinse helps a lot on that, as 99% alc is rather hygroscopic21:25
DocScrutinizerof course you may encounter damaged electronic parts due to changed "circuit" caused by "shorts" from water21:26
viric96% alcohol is not good enough?21:27
DocScrutinizerthen no washing will help :-D21:27
DocScrutinizerhmm, dunno21:27
DocScrutinizerthe higher percentage the better21:27
viricI had a spray at home that was very good at removing oxid from boards...21:27
viricI had totally ruined boards that were left new21:27
DocScrutinizerafter all you want to evaporate the liquid, and when there's 4% water in it that's worse than 1%21:28
viricwell, usually I dry some days.21:28
DocScrutinizerweeks rather than days21:28
DocScrutinizerwater can be very persistent when in a sub-millimeter gap under a chip21:29
viricthere are always chances :)21:29
DocScrutinizerthat's why alc21:29
DocScrutinizerpresurized air also nice21:30
DocScrutinizersome 5 to 10 bar21:30
DocScrutinizershould be waterfree though ;-D21:30
DocScrutinizerpressurized air can hold amazing amounts of water21:31
virichow alcohol helps removing water in small gaps?21:31
viricI'm quite bad at chemistry21:31
viric(yes, I remember quite well the water in pressurized air from a compressor)21:31
DocScrutinizerit mixes extremely well with water, dilluting and replacing it21:32
viricah ok21:32
virichm fire alcohol is 98% iirc... but has that ink.21:32
DocScrutinizerand alcohol is less conductive than water21:32
DocScrutinizerso water, air, alc, air, heat&time. Yielded good results21:34
DocScrutinizerall of that plenty of it21:34
DocScrutinizerexcept heat which shouldn't exceed 80°C21:34
viricha, I used alcohol to cool down things ;)21:36
viric80°C? hm so little?21:36
viricalcohol in boiling water would fire? grr21:36
viricDocScrutinizer: ethanol has an ignition temperature of 425°C21:42
viric(for that 96% thing I have)21:42
viric(I took that temperature from a search result of a search engine :)21:57
DocScrutinizerdoesn't matter, your hardware won't cope with high temperatures for long times, some might even fail in minutes at >100°22:19
Aylais it possible to change the console font at runtime?22:29
whitequarkDocScrutinizer: yeah, I thought so22:56
whitequarkthanks for confirmation22:56
DocScrutinizerwhitequark: YW22:58
DocScrutinizerwhitequark: btw if voltage is under electrolytic limit of water, probably way less will happen. Alas I can't recall if electrolytic threshold of water was 1.2V or 1.8 or even >222:59
DocScrutinizeryou can design circuits that will work perfectly for months under water, without any isolation or sealing case23:00
whitequarkoh, now that's something entierly new for me23:00
whitequarknone of my chemistry classes mentioned that23:00
DocScrutinizerno sarcasm please23:00
mthfont change at runtime is possible: I saw that happen during the boot sequence on earlier SUSE releases23:01
DocScrutinizerI just thought it's worth mentioning it as usually people tend to forget23:01
mthalthough that was with an actual text mode, not fb console23:01
whitequarkDocScrutinizer: no sarcasm23:01
whitequarkI really didn't know about that23:02
whitequarkand none of them really did mention it23:02
whitequarkand so I was extremely surprised, because, when I still was at university, not only they were kind of important for my specialization, but I also enjoyed them and still remember every interesting fact23:03
whitequarksame for high school23:03
whitequarkI shoulda ask my friends. maybe it's just my memory23:04
DocScrutinizer>>In turn, the minimum electrolytic potential for electrolysis of water or 1.23 V. Thus, any current (I) at applied voltage (V) greater than 1.23 V is an overvoltage and results in waste heat23:07
whitequarkah, THAT. just did not recognize the concept when it was named in English.23:08
whitequarkof course I did seen that.23:08
whitequarkeven was on my exam23:09
DocScrutinizerat less than 1.23V deionized water doesn't conduct at all23:09
whitequarkit's just called "RedOx potential" here, not "electrolytic limit"23:10
DocScrutinizerand thus will not cause any harm on electronic circuits that have <1.23V at/between any two points23:10
DocScrutinizerwell, no clue23:10
whitequarkanyway, I did not seen that applied to electronic circuits23:10
DocScrutinizerprobably redox potential is the absolutely correct term23:11
whitequarkI'll take that into account. iirc STM32LV can work at 0.8V...23:11
whitequarkthis also means that LVDS links aren't affected by water. interesting.23:12
DocScrutinizerI've seen circuits built exactly with this purpose and then shown off at conventions23:12
DocScrutinizersome field busses might use the principle23:12
whitequarklvds links I've seen were strictly indoor, probably due to high losses in wires23:13
--- Sun Apr 15 201200:00

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