#qi-hardware IRC log for Saturday, 2010-11-13

wolfspraulxiangfu: hey there02:35
wolfspraulI'm just testing the new image a little. I noticed that when I launch StarDict for the first time after flashing, it hangs for a long time, maybe > 60 seconds.02:35
wolfspraulthen it finally comes up02:35
wolfspraulafter this first time, even after reboots, startup time is much better02:35
xiangfuwolfspraul: yes. it's try to index the dictionary.02:35
wolfspraulso is it saving something into the rootfs?02:35
wolfspraulcan we include a prebuilt index in the image?02:36
xiangfuwolfspraul: yes. one .idx file02:36
wolfspraulwe should find a way to prebuild this index and include it in the image (package)02:36
xiangfuwolfspraul: yes. seems we just need copy the rootfs index to the package file.02:37
xiangfuwolfspraul: I will working on that.02:38
wolfspraulah OK, great02:40
qi-bot[commit] kyak: ben-cyrillic: add the required LC_ALL env variable http://qi-hw.com/p/openwrt-packages/5de66bd02:46
wolfspraulxiangfu: when I launch NanoMap, I cannot return to gmenu2x (it will hang after exiting NanoMap)02:48
wolfsprauldo you have this too?02:48
xiangfusame here.02:49
kyakKDLC @C:C )02:49
kyaksame here :)02:49
kyakin fact, this will happen to every Qt app02:50
kyak(tryed with a Qt "hello world")02:50
xiangfukyak: it's uClibc 9.0.30 problem. make QT not exit correct.02:51
kyakxiangfu: i'd say, it's the problem of Qt.. because it appeared after som qt update02:51
kyakthough you are right, it's working fine with newer uClibc02:52
kyakso maybe it's both uClibc and Qt problem :)02:52
wolfspraulwhat are the default uclibc and qt versions in the backfire release? is our image using those?02:55
kyakuclibc is coming from base system, while qt is from feeds02:55
kyakthat's why qt got updated while uclibc basically remained the same02:55
kyaknow we have shiny new qt from latest feeds and uclibc from backfire02:56
wolfspraulwell stability trumps (broken) features02:57
wolfspraulI was wondering how the versions of the OpenWrt releases and the feeds are kept in sync anyway02:59
wolfspraulhow does this work?02:59
kyakfor example, at the time of backfire release, qt was 4.6.2 (http://downloads.openwrt.org/backfire/10.03/xburst/packages/)03:00
kyakright now it's 4.7.0 in feeds03:00
kyakwolfspraul: i'm not sure.. perhaps it's just the latest version of feeds at the time of release, and then feeds continue going forward03:01
wolfspraulso no sync at all?03:01
wolfspraulhow can we 'pin' Qt to 4.6.2 (assuming that will fix the NanoMap bug)03:02
kyaki've been always seening 'feeds' and 'base system' as pretty different parts.. but i could be wrong03:02
kyakone should just be able to get the base system without feeds.. and feeds mechanism would just provide additional packages, if the user wishes so03:03
xiangfuwolfspraul: it's hard.03:04
kyakso it seems that this strategy may lead to discrepancy between "old" base system and "new" feeds03:04
xiangfuif we back QT to 4.6.2. then we back all the package in svn://svn.openwrt.org/openwrt/packages to old version.03:05
xiangfuwolfspraul: there is one method. add QT 4.6.2 to openwrt-package.git. disable the upstream QT. enable openwrt-xburst.git one.03:05
kyakotherwiser, we can try and fix uClibc03:06
wolfspraulxiangfu: I don't understand what you mean with "all pacakges in ... to old version"03:06
wolfspraulfirst I try to understand how the feeds and openwrt releases stay in sync, if at all03:07
wolfspraulI think I remember seeing svn version numbers together with the feed urls somewhere03:07
wolfspraulwe haven't been able to make an image where gtk/sdl and qt all work since June :-)03:08
wolfspraulwith 'all work' I only mean that a simple app can be started and exit without a hang03:09
wolfspraulcan't go much lower than that :-)03:09
wolfspraulso if older versions help, let's do that first to have a stable image, then we can try with newer versions again, but at least we have a stable base03:09
xiangfuhere is the feeds configure file:03:10
xiangfuhere is the release package svn versions :http://downloads.qi-hardware.com/software/images/NanoNote/Ben/testing/2010-11-07/VERSIONS03:10
kyakwolfspraul: btw, sdl works just fine03:12
xiangfuQT is included by "packages" , now the svn version is 23909.03:12
xiangfufor example we back the "pakcage" to 21808. then all the packages in "package" will all goto old version.03:12
wolfspraulkyak: yes, sdl works now :-) that's good03:12
wolfspraulxiangfu: OK, what do you propose for the NanoMap bug?03:13
xiangfuwolfspraul: I would like use setup one old version in openwrt-package.git.03:13
kyakare there any problems with gtk and uClibc
kyaki remember stardict won't start when uClibc is
xiangfukyak: stardict works fine03:18
kyakmeans that only Qt has problems, right?03:19
xiangfukyak: but when you startup "stardict" in terminal. it's will not exit. it's a directfb bug.03:21
xiangfuor a stardict bug.03:21
kyakxiangfu:  i see03:22
xiangfuwhen stardict exit. there is no stardict thread anymore. but the screen just don't go back to terminal03:22
wolfspraulfirst goal: an image where all apps that have an icon in gmenu2x after booting can at least be started and exited :-)03:26
wolfspraulhow about that?03:26
wolfspraulI haven't seen that in many months...03:26
wolfspraulI wouldn't know how to lower this more, unless we simply remove those apps.03:26
xiangfuall apps? we need work on gmenu2x for make the gmenu2x can start a terminal apps.03:32
xiangfufor now the gmenu2x can not start a terminal apps. like "joe" "mutt"03:33
xiangfumost of package in openwrt is terminal apps,03:36
kyakto start a terminal app, gmenu2x should exit, start this app, and then respawn again03:40
kyaki think it's the same way it works now for dgclock03:41
mthxiangfu: why can't it start them?03:41
mthI have "read" as a menu item (idea from zear) to show the console and to allow the screen to go into power saving03:42
xiangfumth: I don't know the detail . I have add the icon of "joe" "lynx" "alsamixer" ... to gmenu2x. none of them started correct03:42
mthis this with the latest version of gmenu2x? (from projects.qi-hardware.com git)03:42
xiangfumth: yes.03:43
kyakmth: you should share your gmenu icon03:43
kyakthe menu item03:44
mth/boot/local/gmenu2x/sections/applications # cat read03:44
mththe menu entry is only those two lines03:45
xiangfukyak: mth http://downloads.qi-hardware.com/people/xiangfu/gmenu2x.icon.tar03:45
kyakwhat does "wrapper" mean?03:45
kyaklike this03:46
kyakxiangfu: maybe you need to remove that "wrapper" thing from terminal apps03:47
kyaklike mth suggests03:47
xiangfulet me test03:47
mthI don't know what the wrapper option does, but I don't have it here and here it works03:47
kyakmth: do you know how to make "folders" in gmenu? to organize apps03:48
mththis is how gmenu2x handles "wrapper" internally:03:49
mthif (wrapper) command += "; sync & cd "+cmdclean(gmenu2x->getExePath())+"; exec ./gmenu2x";03:49
mthI have no idea why those extra steps would ever be necessary03:50
kyakmaybe to save memory by not running gmenu2x while the app is running03:51
bartbesgmenu2x is closed03:51
bartbeswith the wrapper it reopens03:51
bartbesthat's all03:51
kyakthe more i use gmenu2x (though i don't use it much), the more i want X and a decent window manager03:52
bartbessince gtk and qt work, can't you like.. create a gtk/qt wm for the time being?03:54
bartbes(you being general, not saying you have to do it)03:54
mthkyak: gmenu2x is not running anyway when it starts something03:59
mthat least on Dingux, it relies on init to respawn the menu when the started app exits04:00
mthafaik the same approach is used on the NanoNote04:00
xiangfumth: can you test in your nanonote. http://pastebin.ca/1989851. it's not working here.04:00
mthxiangfu: I only have a Dingoo though, not a NanoNote04:01
kyakmth: yeah, i got it04:01
mthI do have alsamixer though, so I can test it04:01
mthxiangfu: it works fine here04:03
mthI see the ncurses interface, can change volumes and leave with the button that's mapped to Esc04:04
xiangfumth: ok. it's not woking in Nanonote.04:04
mthah, I don't have the very latest version of gmenu2x installed04:05
mthI'll retest with the latest04:05
xiangfumth: ok. there I can check the Dingoo rootfs ?04:06
kyaki think Ben his it's own version of gmenu2x anyway04:06
mthxiangfu: the one I'm using is slightly modified, but it's based on the rootfs built by booboo using buildroot04:11
mthkyak: yes, but we're using the qi-hardware branch/fork as well04:11
mthxiangfu: I had some problems with the "input.conf" in git not matching the Dingoo's keys, but now the latest version of gmenu2x is running04:28
mthand it starts your alsamixer config without problems04:29
xiangfumth: thanks.04:29
xiangfufor now I have no idea why it's not working in nanonote. :(04:29
mthwhat happens on the nanonote then?04:30
xiangfumth: the system goto terminal. and there is 'alsamixer' thread running,  but the terminal display nothing.04:33
kyakxiangfu: what if you started gmenu2x directly from terminal? would it display alsamixer?04:34
xiangfumth: the gmenu2x is autostart in nanonote.04:34
kyakcurrently gmenu2x is not attached to any terminal, in runs in backgroudn via start-stop-daemon04:34
xiangfukyak: yes. when direct start from terminal . all works fine.04:34
kyaki see..04:35
xiangfukyak: do you know how to make it attach to one terminal??04:35
mthxiangfu: if you go to the /proc subdir corresponding to the pid of alsamixer, is there anything useful in the 'fd' directory there?04:35
qi-bot[commit] Ayla: Added the input.conf file for the Dingoo A320/A330. http://qi-hw.com/p/gmenu2x/e3dffeb04:35
qi-bot[commit] Ayla: Merge branch 'master' of projects.qi-hardware.com:gmenu2x http://qi-hw.com/p/gmenu2x/aebd03104:35
xiangfumth: wait I need restart my NanoNOte.04:35
mth0, 1 and 2 are /dev/tty1 here04:36
mthstart-stop-daemon maybe redirects output to /dev/null or to log file04:36
kyakxiangfu: either start it without start-stop-daemon (thus creating additional unnecessary sh process - remember we discussed?) or rework the way gmenu2x executes apps...04:37
mthI'm starting gmenu2x from inittab04:37
xiangfumth: http://pastebin.ca/1989890. here is the 'fd'04:37
mthxiangfu: stdout (1) and stderr (2) redirected to /dev/null, that would explain why you don't see anything04:38
xiangfumth: yes. it's /dev/null04:38
xiangfumth:  very thanks.04:45
mthyou're welcome04:45
xiangfukyak: yes. now I am also try to start gmenu2x in inittab.04:45
xiangfumth: kyak very thanks. now the terminal apps works fine in nanonote.04:50
xiangfuI will commit the change. :)04:50
xiangfumth: in Nanonote I must  export SDL_NOMOUSE=1 for make sure the gmenu2x correct start.04:52
xiangfumth: is that ok I add this line to gmenu2x.sh04:53
xiangfuin gmenu2x.git04:53
mththat's needed for Dingoo as well, I wonder why it's not in the script already04:54
mthplease add it04:54
qi-bot[commit] Xiangfu Liu: add export SDL_NOMOUSE=1 to script file http://qi-hw.com/p/gmenu2x/aac04da04:58
kyakxiangfu: you add it to inittab. I personally don't use it much and will replace with sh anyway05:19
kyak*you can add it05:19
qi-bot[commit] Xiangfu Liu: add ks7010 firmware http://qi-hw.com/p/openwrt-xburst/2e9ef2606:49
qi-bot[commit] Xiangfu Liu: add gmenu2x to inittab, remove init.d/start http://qi-hw.com/p/openwrt-xburst/fce185006:49
qi-bot[commit] Xiangfu Liu: add a lot of apps entry in gmenu2x http://qi-hw.com/p/openwrt-xburst/ad70ede06:49
qi-bot[commit] kyak: lynx: enable SSL and UTF-8 support http://qi-hw.com/p/openwrt-packages/47d100906:54
kyakit seems to me that keyboard layout switching can be done in DirectFB level with dfbinput. Alternatively, qt-embedded has kmap2qmap utility and QWSKEYBOARD env07:52
kyakand it seems that gtk doesn't have anything for that, so perhaps dfbinput is the right thing to try..07:53
kyakso few information about that.. people are mainly trying to fight with i18n input/output in console or X1107:58
kyaknobody cares about qt or gtk running in fbdev :)07:58
lekernel_kyak: +108:17
lekernel_everyone seems to prefer the X11 bloatware...08:17
lekernel_and usermode applications that talk directly to I/O devices08:18
lekernel_try google android maybe...08:25
lekernel_btw, why didn't google use X11? :D08:25
rafalekernel_: well, google android does not use any application for linux, no just x11. .. it just uses linux kernel if I read correctly.. no shells, no initscrips, no linux fhs08:29
rafalekernel_: it is like android is not a linux distro, just uses the kernel08:29
rafamaybe them hate GNU and dont want to be called gnu/linux :)08:30
qi-bot[commit] Xiangfu Liu: [uboot-xburst] remove debug code http://qi-hw.com/p/openwrt-xburst/ce4de7508:55
kristianpaulDictionary+Wikpedia+Music http://en.qi-hardware.com/wiki/File:NN_stickers_drafts.png :D09:37
kristianpauland the first one09:38
wpwraklekernel_: you may enjoy reading this one: http://lwn.net/SubscriberLink/413335/2c202976e5743e89/09:40
wpwrakkristianpaul: yet another X without X, eh ? ;-)09:42
wpwrakkristianpaul: in fact, all these things remind me of the heydays of ATM. back then, they all recognized TCP as the "killer application of IPv4". so they tried to come up with alternatives. TCP without IP, TCP without TCP, and so on. it all led to nothing.09:44
kristianpaulexcuses just excuses09:45
wpwrakkristianpaul: the arguments were of course complexity (both intrinsic and due to unnecessary features), efficiency, and so on. much like what X11 is getting accused of.09:45
lekernel_yeah, nothing except google android, iOS and Mac OS X, all incenditally featuring a better and faster user experience than bloatware based on x-windows09:45
lekernel_and more technically I think no one would argue that the fact that X talks directly to I/O devices instead of the kernel is a bad thing09:47
wpwraklekernel_: but they all live in parallel universes. you may as well list the text console :) i think can all agree that the text console works very well in its specific domain.09:47
lekernel_mac os x, a parallel universe?09:47
wpwraklekernel_: if you and bleeding edge performance, you often need to cut out the man in the middle :)09:47
lekernel_X is crap, period!09:48
wpwraklet's talk about this again in ten years ;-)09:49
wpwrakin that thread, a few good points are being made about legacy features that indeed don't make much sense anymore. of course, their main impact would be code size and porting effort, not so much real-life performance.09:51
lekernel_isn't X all about those features?09:52
lekernel_on top of crappy implementations...09:52
lekernel_anectodally, which just crashed again on my machine this morning...09:53
rafalekernel_: is not mac os x propietary?09:53
lekernel_sure it is, but at least it doesn't suck technically09:53
lekernel_seems it takes a big software company to make bold technical decisions like Unix without X09:54
lekernel_or, more exactly, successfully implement them :)09:55
kristianpaulhow you could know what the "successfull" really is if this is not open?09:57
kristianpaullekernel_: are you aware of a libre sofware or language for generating timing diagrams?10:15
kristianpaulmaybe latex?10:15
kristianpaulwell i could do the module in verilog and the view it on gtkwave but the idea is not simulte every time i need a timing graph10:15
wpwrakgnuplot ?10:16
kristianpauldont know, but make sense10:17
kristianpauli'm not fan of gnuplot10:17
kristianpauli'll read a bit more about it10:18
wpwrakdata files for gnuplot are very very simple.  <x-coord> <y-coord>\n<x-coord> <y-coord>\n ...10:18
wpwrakwant a new line ? ... <x> <y>\n\n<x> <y>\n...10:18
wpwrakdon't have x coordinates but a constant "clock" ? just omit the x coordinate10:19
kristianpaulwpwrak: ah gnuplot is what you use for the usrp signaling visualization isnt?10:20
wpwrakgnuplot is what i use for probably 99% of all graphing tasks :)10:21
kristianpaulwell i'll follow you then10:21
kristianpaulfoudn it http://www.gnuplotting.org/defining-piecewise-functions/11:00
wpwrakkristianpaul: it's probably way too complex to express your graph algorithmically. better to make a file with the coordinates and plot the file11:04
kristianpaulscript > file > plot11:07
viricwejp, wpwrak: ah, this talks about batteries, MCUs, regulators... http://www.silabs.com/Marcom%20Documents/Resources/LowPowerSolutions.pdf13:52
mthkristianpaul: I use matplotlib for Python to create graphs: http://matplotlib.sourceforge.net/13:59
viricwpwrak: Do you think I could program this with free tools? http://www.silabs.com/products/mcu/lowpower/Pages/C8051F92x-93x.aspx14:09
wpwrakviric: to a very large extent, yes. sdcc doesn't seem to support the C8051F9xx yet, but it supports 8051 in general many other C8051Fxxx MCUs.15:04
wpwrak(maybe 9xx support is already there, too. didn't search very hard)15:05
wpwrakC2 looks the same, so you can probably use my stuff from f32xbase15:07
wpwraknote that there's no USB in these chips.15:07
qwebirc21411Hello, has anyone where gotten the Ben NanoNote working with the micro-sdio wifi card (SDW-823)?15:15
viricwpwrak: what about uploading the compiled code? easy?15:32
viricwpwrak: a 3v3 serial port would do?15:33
viricI'll check later the documentation15:33
viricI only wondered, if they document how to program it, or instead they only say "click this on Our Program"15:33
wpwrakviric: i'd use a ben for the upload :)15:33
wpwrakwith one of these cables: http://downloads.qi-hardware.com/people/werner/f32x/c2ben-run1.jpg15:34
wpwrakthe programming interface isn't completely documented. but you can find the information ... :)15:35
viricI can find it? :)15:39
wpwrakthis is a good hint for the missing bits: http://coding.derkeiler.com/Archive/General/comp.arch.embedded/2004-06/1134.html15:39
viricI better start taking notes15:39
viricI thought I could even use their temperature sensor!15:39
wpwrakor just use f32xbase/f32x - it supports some of the functionality of C2, namely flash read/write and boundary scan :)15:40
wpwrakdidn't bother with breakpoints and the like, though15:40
Action: FrankBlues waves15:55
qi-bot[commit] kyak: updated gmenu2x to latest git. http://qi-hw.com/p/openwrt-packages/afd682216:37
qi-bot[commit] Werner Almesberger: The eda-tools welcome the arrival of fpd2pdf http://qi-hw.com/p/eda-tools/4ae9a0916:52
qi-bot[commit] Werner Almesberger: scripts/fpd2pdf (usage): don't call the input files .pdf - they're .fpd ! http://qi-hw.com/p/ben-wpan/bb5486416:52
qi-bot[commit] Werner Almesberger: scripts/fpd2pdf: moved over to eda-tools http://qi-hw.com/p/ben-wpan/6f8665e16:52
kristianpaulwolfspraul: Just telling you the debuging process is done now http://en.qi-hardware.com/wiki/GPS_Free_Stack#Debugging. and now i'm working on the implementation of gps data using SIE board, why not Ben?20:14
kristianpaulwell seems its hardware is not capable of getting data as faster as i spected (8mhz) and so far just 2mhz acording some tests made by wpwrak20:15
kristianpaulanyway i'll dedicate later some time in the university lab  in order to do my benchmark, so SIE will be really usefull after all :)20:16
kristianpauli'm not sayinng i'm kicking out ben from my way, but for now SIE will help to develop then later optimize and fit gps stuff on the the nanonote20:23
wolfspraulkristianpaul: sure, wpwrak already mentioned it to me20:49
wolfspraulof course, this is the right approach20:49
wolfspraulfirst we need to get it to work and learn, then we see what is the cheapest tech we can find to implement it20:50
wolfspraulnice pictures!20:51
rafawolfspraul: hi.. I have uploaded :22:22
rafa- a free of patents repository built from OE22:23
rafa- 4 rootfs images : 2 bootstrap (one for NAND and another for SD) and 2 full jlime beta4 (one for nand and another for SD).22:24
rafawolfspraul: all this stuff is free of problematic packages. The repo is sane, so any package there is okey to install. Also, I removed from the full jlime beta4 image the packages with patented technologis. I replaced that stuff with a theora player.22:25
rafawolfspraul: maybe I should upload some screenshot and write a wiki page about?22:25
--- Sun Nov 14 201000:00

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