#qi-hardware IRC log for Saturday, 2010-10-30

rafaI know how to write argentina in ch :D01:14
rafa?9รท401:14
logictheoI found Issue 73 "software controlled brightness, but  it's currently not implemented" . Before I went to sleep I wanted read an ebook but the brightness was too bright. Maybe I should change all the text to brown so that it doesn't hurt my eyes(as a temporary 'workaround' solution)?05:19
logictheo(by the way I use 'lynx oz-of-ozma.txt' just as an example of how I currently read ebooks)05:19
kyaksince you are reading in web browser, you indeed can change the text color to grey or brown05:21
bartbesdoes anybody know what this means?05:38
bartbesMakefile:63: *** Download/default is missing the VERSION field..  Stop.05:38
bartbesthat went well...05:49
logictheoI underestimated the ease of changing colors in lynx06:20
bartbesgo bot:06:36
bartbesoh06:36
bartbesoops06:37
qi-bot[commit] bartbes: Updated to new version of nlove (and now uses hg repo) http://qi-hw.com/p/openwrt-packages/c6552a106:38
bartbes(had to rebase, forgot to pull first)06:38
kyakchanging color in lynx as easy as adding relevent html tags06:42
wpwrakhmm, this looks nice: http://github.com/blog/699-making-github-more-open-git-backed-wikis08:14
bartbeswpwrak: it sure does08:30
wpwrakhere's another one, haskell instead of ruby this time: http://github.com/jgm/gitit08:33
wpwrakuses pandoc for conversion. alas, pandoc can't use MediaWiki as input. http://johnmacfarlane.net/pandoc/08:33
wpwrakgithub uses some multilingual markup system, interfacing to markdown, textile, rdoc, org, creole, rst, asciidoc, pod, man (groff), all but the last two completely unknown to me.08:39
bartbesmarkdown is epic08:40
bartbesand textile is what they were using before08:40
wpwrakis markdown good ? at least the basic version looks fairly limited. e.g., no tables.08:41
wpwrakwhich of course spawned extensions, like in pandoc. history repeating itself over and over again and again ;-)08:41
bartbesMarkdown's syntax is intended for one purpose: to be used as a08:43
bartbesformat for *writing* for the web.08:43
bartbesMarkdown is not a replacement for HTML, or even close to it.08:43
bartbes(from a syntax manual thingy)08:43
wpwrakyeah, i saw this08:44
wpwrakmy question is - what would be painless enough to actually make a transition possible ?08:44
wpwraktransition of the qi-hw wiki. i hate web-only interfaces :)08:45
wolfspraulI am pretty sure MediaWiki will continue to be the #1 free wiki software, nothing comes even close to it.08:47
wolfspraulI won't load more tools on my admin shoulders, MediaWiki + Indefero and the assorted tools around them is enough08:47
wolfspraulthere is this MediaWiki filesystem project, but unfortunately it's abandoned08:47
wolfspraulmaybe it will be reactivated one day, who knows08:48
wolfspraulthe moment werner has his filesystem view he won't care about the rest anymore :-)08:48
wpwrakwolfspraul: yeah, the filesystem view is what i'm after ;-)08:48
wpwrakthat is, read and write08:49
wolfspraulsure08:49
wpwrakalas, this means that you really ought to have a version control system underneath08:49
wolfspraulalso my personal opinion, while this gitwiki at github is really cool, and it's no surprise that the GIThub audience likes it, I am always a bit careful when a good tool is over-extended into areas it was not originally designed for08:50
wolfspraullearnt my lesson08:50
wolfspraulI remember how I was totally hooked on extending IRC to multi-party video chat once...08:50
wolfspraulargh08:50
wpwrak;-))))08:51
wolfspraulgit was designed to deal with source codes, and the very delicate dependencies among the many different files08:51
wolfspraulwe all know it08:51
wolfspraulconflict resolutions, cherry-picking, what not08:51
wolfspraulif a wiki is about _documents_, then that's just different08:51
wolfsprauldocuments have embedded graphs, pictures, even videos08:51
wolfsprauldocument != source code, in the end08:52
wolfspraulso let's see08:52
kristianpaulsphinx.pocoo.org08:52
wolfspraulit definitely seems like an improvement from the earlier github wiki08:52
wpwrak"put myself on fire, jumped out of an airplane, into a swamp, in the middle of a war zone" may be an instructive experience, but then i doesn't necessarily mean that you should shun transportation in general ;-)08:52
wolfspraulso no wonder they are forcefully upgrading everyone to the new git-based one...08:52
wolfspraullike I said, when it comes to wikis, I feel very good with MediaWiki08:53
wolfspraulright now08:53
wolfspraulit's a really really great piece of software08:53
wpwraki think the web-based interface can just stay simple. it wouldn't lose any functionality. you can reserve the fancy bits for the command line.08:53
wpwraklast time i checked, the conflict resolution model was along the lines of "someone else changed what you were editing - please reload and do it again". has this changed ?08:55
wpwrak(speaking of mediawiki)08:55
wolfspraulhe, don't know08:55
wolfspraulthis happens so rarely... it's not sources :-)08:56
wpwrakit must happen a lot in the edit wars on wikipedia :)08:56
qi-bot[commit] Xiangfu Liu: Revert "Replace OpenWrt banner with a custom NanoNote banner." http://qi-hw.com/p/openwrt-xburst/7000e6e10:09
qi-bot[commit] Xiangfu Liu: move the NanoNote banner to data/ folder http://qi-hw.com/p/openwrt-xburst/70a813710:09
qi-bot[commit] Xiangfu Liu: move NanoNote special files to data/ folder http://qi-hw.com/p/openwrt-xburst/d5e9dcd10:17
viricHello10:17
viricI'm trying to write an udev rule for when the 'usb0' network device appears (nanonote)10:17
viricHow to match it? SUBSYSTEM=="net", NAME=="usb0" does not work for me10:18
viricah, found in the openmoko web, I think10:18
qi-bot[commit] Xiangfu Liu: move NanoNote special files(build, reflash_ben.sh) to data/ folder http://qi-hw.com/p/openwrt-xburst/cae161510:26
kristianpaulhttp://ninapaley.com/mimiandeunice/wp-content/uploads/2010/10/ME_218_ColdVirus.png :)10:32
kristianpaulwolfspraul: http://ninapaley.com/mimiandeunice/wp-content/uploads/2010/10/ME_219_MoneyOnTheInternet1.png10:34
kristianpaulFIY http://ninapaley.com/mimiandeunice/category/ip/10:34
kristianpaulbbl10:34
wpwrakalso interesting - a distributed bug tracker: http://bugseverywhere.org/be/show/HomePage11:34
fossroxheya DNS777 :D12:02
DNS777hola :)12:02
viricI want to put images accurately on screen...12:48
virichow much do I have to worry about the special LCD in the nanonote?12:48
viricand what is the way to worry about it?12:49
kyakviric: why do you have to worry?13:10
kyakuse fbi or imgv for viewing images13:11
viricI'm writing my own viewer13:12
viricI want to display 15000x15000 pixels jpegs13:12
viricAnd both fbi and imgv load all the uncompressed image to memory.13:12
viricso, no way.13:12
viricDo they have code for the special screen of the nanonote?13:13
kyakyou check?13:15
viricI did not check13:15
kyakyou should at least understand how they work before trying to do something of your own13:15
viricI know how they work13:16
viricI have zoom working and I'm doing the scroll.13:16
kyakbut you didn't check, how do you know?13:16
viricAh, sure I checked imgv and fbi for the image loading part.13:16
kyakzoom is working and scrolling, too in fbi13:16
viricIt is not that I *wanted* to write my own :)13:16
viricI know that they work. But they cannot load a 15000x15000 jpeg in the nanonote :)13:16
kyaki'm not sure my laptop can do that ;)13:17
kyakwhat are you trying to do anyway?13:17
virickyak: I mean that with libjpeg, it's not needed to load all the uncompressed pixels into memory.13:17
virickyak: but that is what both imgv and fbi do.13:17
kyakcan you write a patch for them not to do that?13:18
viricI don't mind having repeated decoding.13:18
viricFor fbi, impossible. It's all designed for having the image whole on memory13:18
viricFor imgv... it would be about making the code much bigger.13:18
kyaki found fbi pretty fast.. again, what is your target? why you attempt to view this big images in Ben?13:19
viricI have maps.13:19
kyakit's like 45 screens to scroll to view the whole image (without resize)13:19
viricI want to use the nanonote as map viewer.13:19
viricSure, I'll have zoom.13:19
kyakhave you tryed NanoMap?13:19
viricMy maps are cartographic13:20
viricnot PGS13:20
viricnot road-related13:20
kyakthat'a 19th century, my friend13:20
kyak:)13:20
viricI don't want to use it as map for a city13:20
viricI do mountain climbing :)13:20
viricI also use compasses :)13:21
kyakok13:21
kyakhow about that - you can split your big image into pieces13:21
kyakand write a script for viewing them13:21
kyakusing imgv or fbi13:21
viricIt's far easier to write a SDL program, for me :)13:21
viricand I enjoy the process, too. So, all fu.13:21
viricfun13:22
kyakwell as you wish.. if you ever succeed, please let me know :)13:22
viricok, don't worry13:22
viricbut I imagine I'll do something simple that will "work for me" :)13:22
kyakactually, NanoMap is working this way13:22
kyakin has tiles and downloads/displays them13:23
viricAnd how much will take a 15000x15000 map ? :13:23
viric:)13:23
kyakyou can calculate13:23
viricIn fact 5000x5000 would be enough13:23
virica map I have here of that size takes 10MB.13:23
kyaki used NanoMap with ~1Gb data, which i pre-downloaded on my laptop13:24
kyakit ran smooth13:24
kyaktiles might be the way to go13:24
kyakinstead of messing with memory/not memory :)13:24
viricno, I don't like tiles :)13:24
viricnanomap is for another job, I think.13:25
viricI only want to be able to display big pictures.13:25
qi-bot[commit] kyak: ben-cyrillic: updated some glyphs, added pre-png file http://qi-hw.com/p/openwrt-packages/0be554f15:20
qi-bot[commit] Werner Almesberger: boom/Makefile: new target "db" to build or update the database http://qi-hw.com/p/eda-tools/6ba3a5618:29
qi-bot[commit] Werner Almesberger: manu/Makefile: run a recursive make on all subdirectories http://qi-hw.com/p/eda-tools/a6c65c718:29
wpwrakphew. now this works again, too.18:29
qi-bot[commit] Erwin Lopez: Uploaded the first Beta to test in the development group http://qi-hw.com/p/sie-ceimtun/8f2a9cc19:36
kristianpaulwolfspraul: i have some delays here about the scope meter, so i deciced not forgot it but debug later and trust the data from the SiGE evb, so i hope have some raw data this sunday :)20:56
wolfspraulkristianpaul: he. no worries.21:05
wolfsprauldelay = you are working on the right thing :-)21:05
wolfspraulyou should ask wpwrak about it...21:06
wpwrak;-))21:06
wolfspraulhe's the master of producing great results, but also the master of delays21:06
kristianpauli'll do the CPLD thinf that wpwrak is askign me for long time ago ;)21:06
kristianpauljsut in case21:06
kristianpaulthats was actually i wanted to debug signal first21:06
kristianpaulbut damn we need copyleft scopemeters ;-)21:07
kristianpaulyeah i want delay to make sure all works as is supposed to, not suppose that it will work after powering up21:07
kristianpauls/i/why21:08
wpwrakkristianpaul: what i think would be nice to have is a number of fast but low-resolution ADCs. that way, they're not so expensive. then, hook all this up to an FPGA and connect a few GB of PC memory on the other side.21:21
kristianpauli got one ADC 10Mhz from Ti21:22
kristianpaulbut was before you advised me i need at least 80Mhz..21:22
kristianpaulhmm21:22
kristianpaulgood point21:22
kristianpaulfast no res21:22
kristianpauls/no/low21:23
kristianpaulfew GB..21:23
wpwrak10 MHz is DC ;-)21:23
kristianpaulhe21:23
kristianpauli just have 800Mb ram actually :-/21:23
kristianpaulbut yeah ram is cheap those days..21:23
kristianpaulwpwrak: are you aware of soem of this ADC you recommend from TI?..21:24
wpwrakthe trickiest bit would be visualization21:24
kristianpauls/from TI/in TI stock21:24
kristianpaulindeed21:24
kristianpaultiming..21:24
wpwrakhmm, i found something that looked halfway promising, but i don't remember from which company. 6 bit resolution.21:25
kristianpauloh21:25
wpwraki think you need at least 5 bits for signal analysis. 6 should be quite comfortable. then make 8 or more channels.21:26
kristianpaulwith 3 channels is enought for now21:27
kristianpaulor i'll end doing a logic analizer.. with the SIE21:27
kristianpaulbut i *really* want explore/debug analog signaling before jump digital.21:27
wpwraki'm surprised you're so worried about signal integrity here. had some bad experiences ?21:28
kristianpaulno bad no good. so i want check21:28
kristianpaulno precious experiences ;)21:28
kristianpaulprevious*21:28
wpwrakyou're not telling me this is your first circuit ? :-)21:29
kristianpaulas i said i can ignore this, mae  SPI-to SPI on the CPLD21:29
kristianpaulno no21:29
wpwrak(ignore) yup. maybe there's even some test pattern mode or such ?21:30
kristianpauli can look for a pattern on data out right21:30
wpwrak(ignore) in any case, you can tell by the clock whether your signal is right or wrong. clock is usually the fastest and thus most sensitive.21:31
kristianpaulthe light will come when i got my first PRN code founded ;-)21:31
kristianpaulwpwrak: i was thinking same about clock21:31
kristianpauli have thec counter :)21:31
kristianpauli'll move to digital..21:32
kristianpaul:-)21:32
kristianpaulif something go wrong (readings) i can have the lab next week i think21:32
kristianpauland debug21:35
wpwraksounds like a plan :) clocks are tricky anyway. it's easy to get a crystal wrong, etc.21:36
kristianpaulwpwrak: (first circuit) in terms of radio receiving signal (not counting classical RF modules), yup21:37
kristianpaulhmm21:37
kristianpaulwell yes i know i must do some clock domain sync21:37
kristianpauli was reading about it sebastien push me as soo he noticed i was doing a counter21:38
wpwrak(clock domain sync) you mean in the spi-spi bridge ?21:39
kristianpaulwell not21:40
kristianpaulin the counter part21:40
kristianpauli'm not sure if apply for SPI-to-SPI21:40
kristianpauli was planing a 3 state buffer and no more..21:40
kristianpaulor do some "memory" blocks21:41
wpwrakhmm. i think the tricky area when it comes to clock domains will be the access demultiplexing logic. of course, if you have dual-ported ram blocks, they may take care of some of the headache.21:43
qi-bot[commit] Erwin Lopez: Changed src in Beta with example to all 4 PWM at 100% http://qi-hw.com/p/sie-ceimtun/cc225fc23:28
--- Sun Oct 31 201000:00

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