#qi-hardware IRC log for Wednesday, 2010-11-17

qi-bot[commit] Xiangfu Liu: [config.full_system] add dosfslabel, resize2fs, sysfsutils, tune2fs http://qi-hw.com/p/openwrt-xburst/5b3819d00:57
kyakmirko marked ncurses as broken01:08
kyakwithout actually interesting WHY it got in openwrt-packages01:09
xiangfukyak: I am reflash the 2010-11-15 build to nanonote now.01:10
kyakxiangfu: no sense without ncursesw.01:11
xiangfuthen I will add more apps entry in gmenu2x. so if you think which application should add in gmen2x. just let me know.01:11
kyakvim, abook, nightsky :)01:11
kyakalso i was thinking about gmenu2x01:12
kyakand i realized that it makes me sick01:12
wolfspraulkyak: the problem came from mirko's testing branch in openwrt-xburst01:12
wolfspraulbut there was no testing branch in openwrt-packages, and openwrt has no 'pinning' feature to pin feeds at a certain revision...01:12
wolfspraulso we'll sort this out...01:12
wolfspraulkyak: I thought he marked it as broken only in the new openwrt-packages testing branch? or also in master?01:13
kyakwolfspraul: do you know what was the problem?01:13
wolfspraulhe said something in alsa01:13
wolfspraulwhich I think is also because he is on the testing branch :-)01:13
kyakwolfspraul: no, it's marked as broken in master01:13
kyaknot possible01:14
kyakthere is the same ncurses version on testing branch01:14
kyakin fact, i just took it from testing01:14
wolfspraulok let's see01:14
wolfspraulone by one01:14
kyaknto changing anything01:14
kyaklet me explain01:14
qi-bot[commit] Xiangfu Liu: [nethack] [new package] single player dungeon exploration game http://qi-hw.com/p/openwrt-packages/182a82a01:15
kyak'ncurses' is a part of base openwrt system01:15
kyakit is located in packages/ncurses01:16
kyakit is not a part of openwrt feeds01:16
kyakwe used to have ncursesw (ncurses wide version, which does support UTF-8) when we were tracking openwrt trunk01:16
kyakbut there was no ncursesw in backfire.01:17
kyakit was just not backported there01:17
kyakbut since we went back to tracking backfire, we lost ncursesw01:17
wolfspraulok, following01:17
kyaktherefore i added it to openwrt-packages.. thus effectively overwriteing ncurses from packages/ncurses of backfire base system01:17
kyakin fact, ncurses Makefile and pacthes in openwrt-packages is just a copy of ncurses from latest openwrt01:18
kyaknow, we have support for UTF-8 in ncurses01:18
kyakwe had, before mirko marked it as broken01:19
kyakand UTF-8 works i nfact.. tested with a bunch of apps like lynx, abook..01:19
wolfspraulyes sure01:19
wolfspraulbut we will just enable it again :-)01:19
wolfspraulmirko just wants his peace on the testing branch, and as long as openwrt-packages had no matching testing branch, this created a conflict for him, for whatever reason01:20
wolfspraulbut now there is a testing branch in openwrt-packages as well01:20
kyaki'll just ask mirko to read my explanation above.. maybe he will even backport ncursesw to backfire01:20
kyakthis would be the best solution01:20
wolfspraulI don't understand why he marked ncurses as broken on the opwenwrt-packages master branch though01:20
wolfspraulso maybe I'm missing something01:20
wolfspraulkyak: see this mail http://lists.en.qi-hardware.com/pipermail/discussion/2010-November/006141.html01:21
wolfspraulI think we can re-enable ncurses in the openwrt-packages master branch01:22
kyakhm wait a second01:22
kyaki thought openwrt-packages IS a testing repo01:23
wolfsprauleh, that's a separate discussion/definition01:23
kyakfrom what i understood now, we need to merge commits from master to testing01:23
wolfspraulno 'need', if you think it's critical for the next upcoming release, then yes01:23
kyaki think all commit should go there, why not?01:23
wolfspraulcommits go to master first01:24
kyakwhat is broken, is marked as broken01:24
kyakwhat is working, is working01:24
wolfspraulbut I think ncurses was working before, just depends for whom :-)01:24
wolfspraulbecause the openwrt-xburst and openwrt-packages branches are not in sync01:24
wolfspraulI have a related question: is it a good idea in general to have 2 packages with the same name in 2 different locations?01:25
kyaknot the best idea. but it works01:25
wolfspraulwhat if the ncurses in openwrt-packages would be ncurses-utf8 or so, just to give it a distinctive name01:25
wolfspraulwhat would be the consequence? dependency problems?01:25
kyakyes, a lot of packages depend on libncurses01:26
kyakit's not possible just to use another name01:26
kyakas i said, the best would be to backport it to backfire01:26
kyakin fact, it wil not cause any problems because it is tested..01:26
kyakmaybe you rememeber, i asked larsc to backport some commits to backfire to make ncursesw build01:27
wolfspraulyes, remember01:28
wolfspraulthey don't like the idea of backporting ncursesw?01:29
wolfspraulI think the solution with ncurses in openwrt-packages is just fine01:29
wolfspraulmirko's problem comes from him being on the testing branch and running into a conflict01:29
wolfspraulmy understanding is that we can re-enable it on the master branch anyway, but I will confirm with him01:29
kyaki didn't ask to backport ncursesw, i thought we could do  "locally" (i.e. in openwrt-packages) without disturbing anyone :)01:30
kyakin fact, i remember that we had local ncurses once01:31
kyakand then it went upstream01:31
kyakand we removed it from openwrt-packages01:31
kyakit was some work between larsc and... bartbes?01:31
kyaki would like to know about mirko's problem01:31
kyakand conflicts he had on testing branch01:32
wolfspraulhe said "it broke also so I marked it as broken"01:34
wolfspraulnot also01:34
wolfspraul"it broke alsa so I marked it as broken"01:34
kyakdon't follow the connection..01:34
wolfspraulmirko said "ncurses in openwrt-packages broke alsa so I marked it as broken"01:35
kyakwhat he said in git log:01:35
kyak[ncurses] mark ncurses as broken, since it interferes with openwrt/package/ncurses01:35
wolfspraulok let's see what he says later01:36
wolfspraulmy understanding is about the branches, and since there is a testing branch in openwrt-packages now, ncurses can be enabled again in openwrt-packages/master, as long as it works for master01:37
wolfspraulin general overriding the same package name is probably not very clean, but if it works I guess it's OK as a workaround01:37
xiangfukyak:  have you test the if "libncursesw" compile fine after mirko mark the "libncurses" as BROKEN ?01:46
xiangfukyak: I guest it's will still compile fine (ont tested)01:47
kyakxiangfu: uh, no... wait a second.01:47
kyakactually you may be right01:47
wolfspraulyou guys lost me now?01:49
xiangfuMirko mark the another "libncurses" as broken, not "libncursesw"01:49
wolfspraulthe package name is 'ncurses'01:49
kyaktheoretically, it would take libncurses from openwrt-xburst and libncursesw from openwrt-pacakges01:50
wolfspraulah no, the folder is 'ncurses' but the package is 'libncurses' and 'libncursesw'01:50
kyakmirko indeed marked 'libncurses' VARIANT is broken01:50
xiangfupackage name is 'ncurses'  in oepnwrt-package include two package: libncursesw and libncurses01:50
wolfspraulif the 'libncurses' package in openwrt-packages comes out exactly as the upstream one, then why don't we just have the libncursesw package in openwrt-packages?01:57
wolfsprauland if that gets backported to backfire one day, we remove it. or if we update to the next release, we remove it as well.01:58
kyakif it works now as xiangfu suggested, this might be exactly what we have now. just libncursesw package in openwrt-packages02:06
wolfspraulbut then we still need to decide whether we want to remove the BROKEN marked parts (libncurses package), or leave it just as is (maybe will confuse someone later)02:07
kyakactually you are right, we could remove the libncurses completely from Makefile.. marking at as BROKEN could be just easir :)02:09
kyaki wonder why there's still two libncurses in menuconfig..02:13
kyaki also see that utilities depending on ncurses get deselected02:16
kyakit looks like it got even worse now02:18
kyakmarking libncurses as broken in openwrt-packages somehow affects libncurses from openwrt-xburst02:18
kyakand all other paps depending on it02:18
xiangfukyak: yes. those two libncurses all marked BROKEN in menuconfig.02:20
kyaki think im going to fix it as wolfspraul suggested.. leave only libncursesw in openwrt-packages02:21
xiangfukyak: sound good.02:21
kyakxiangfu: btw, yesterday i was able to type russian in Qt app!02:22
wolfsprauland I think by committing to 'master', in both openwrt-xburst or openwrt-packages, we will not intefere with mirko's testing branches and activities02:22
kyakall thanks to kmap2qmap, which convert linux console keymaps to Qt keymaps02:22
kyakit also means that all special keys of Ben are working in Qt02:23
xiangfukyak: good job. :)02:23
kyakalso mirko fixed that annoying bug when Qt app couldn't exit correctly02:24
wolfspraulkyak: well, not in NanoMap yet, it seems02:25
kyakwolfspraul: in NanoMap, too!02:25
wolfspraulyou tried it?02:25
kyakmaybe he strted NanoMap from gmenu2x02:25
wolfspraulI thought mirko said earlier in this chat that nanomap still doesnt' exist02:25
kyakthen it won't exit corerctly because export QWS_DISPLAY=linuxfb is not set02:25
wolfspraulyes, probably from gmenu2x02:26
kyakthis is another problem02:26
wolfspraulso where to we have to export the variable to gmenu2x has it too?02:26
kyakgmenu2x won't read /etc/profile on startup, and doesn't know about those env02:26
wolfspraulwhere do we have to...02:26
kyakwe need to export those before starting NanoMap02:26
kyaksomewhere in gmenu2x menu item02:27
kyakmaybe a wrapper02:27
kyakxiangfu: btw, maybe it's better to start gmenu2x in another way through inittab02:28
kyakcalling it via ash --login02:29
kyakso it would know about env vars02:29
xiangfukyak: for now it's "tty1::respawn:/usr/bin/gmenu2x"02:30
kyakmaybe "tty1::respawn:/bin/ash --login -c /usr/bin/gmenu2x"02:30
xiangfukyak: since the "usr/bin/gmen2x" is already a wrapper. how about add one line read /etc/profile02:30
xiangfukyak: ok. let me try.02:31
kyakxiangfu: good idea, too02:31
Action: xiangfu testing tty1::respawn:/bin/ash --login -c /usr/bin/gmenu2x02:32
xiangfukyak: everytime I exit the 'NanoMap', I must input "ctrl +c " for total exit.02:46
kyakwhat do you mean?02:47
kyakit has has characters left in console?02:47
xiangfu1.  run the "NanoMap -qws" in terminal.  2. exit NanoMap by "ESC".02:47
xiangfuthen the command line not return until I send "Ctrl + C"02:48
kyakyep. same here02:49
kyakmoreover, if yuo press some buttons in NanoMap, you then wil see those characters on concolse after you exit02:49
kyakxiangfu: is that really important? you are returning back to gmenu2x after you exit NanoMap.. if you run it in console, Ctrl+C is the most frequent key combination anyway :)02:51
kyakbtw, running build now.. had to rename ncursesdirectory to ncursesw and cleaned up things related to ncurses from Makefile02:52
qi-bot[commit] Xiangfu Liu: add more apps entry in gmenu2x, abook, bc, nightsky, sc, vim http://qi-hw.com/p/openwrt-xburst/08534bd02:53
wolfspraulwell it should exit without ctrl-c, I think02:54
kyakxiangfu: you can test Qt with other apps from qt4-demos02:54
kyakwolfspraul: it exits, but some crap is left in terminal02:55
kyakso you have to ctrl+c to clean up your command line02:55
wolfspraulmaybe not highest priority, but in general I'd say cleanliness is important :-)02:56
wolfspraulthere's a lot of strange left-over texts and blinking in general when flipping around apps on the NanoNote02:56
wolfspraulbut it's not top priority right now I'd say, compared to many lower hanging fruits we can go after02:57
kyakxiangfu: try /usr/share/Qt4/demos/textedit/textedit -qws to see how incredibly slow it is :)02:59
kyakin fact, X11+gtk1 is faster than Qt03:00
xiangfukyak: add "/bin/ash --login -c " to inittab or add "source /etc/profile" to  gmenu2x both make NanoMap works. which one do you think is better :)03:07
kyaksource /etc/profile i think is better.. to avoid creating of parent process of gmenu2x (overhead)03:08
qi-bot[commit] Xiangfu Liu: read the /etc/profile before run gmenu2x http://qi-hw.com/p/gmenu2x/6e5fa8c03:11
qi-bot[commit] Xiangfu Liu: [gmenu2x] update, read /etc/profile before run gmenu2x http://qi-hw.com/p/openwrt-packages/dca493903:12
xiangfukyak: we are not select qt4-demos by default. I need compile them for test.03:26
kyakxiangfu: i have it compiled, can put to somewhere on fidelio03:27
kyakoh no i haven't :)03:28
kyakcleaned it03:28
wolfsprauldoes it make sense to include them? to quickly make people see what is possible and what is not possible?03:29
wolfspraulwe continue to have gtkperf in the image, and I think it's nice for that purpose03:29
kyaki agree03:29
xiangfuwolfspraul: gtkperf, we can start it but we can not exit normal, must start another shell "kill" it. that why gtkperf not in gmenu2x03:30
xiangfukyak: what is the different about qt4-demo, and qt4-example ?03:30
kyakxiangfu: in my understanding, "demos" are applications, which are ready to use.. "examples" are just samples of different UI elements and Qt technoquies03:32
wolfspraulxiangfu: no need to add it as an icon, that's a bit too much03:32
wolfspraulI think it's nice for someone who thinks about developing to quickly get an idea of the potential of the platform, or where it stands03:32
wolfspraulbut that potential developer may very well go to the command line first and start it from there03:33
kyakxiangfu: btw, in my somewhat minimal build gtkperf can be started fine, but stardict fails to start.. i'm still trying to figure out what does stardict lack for, since it works in full_system03:35
xiangfukyak: should we change the "ben-cyrillic" to "y" now?03:36
kyakxiangfu: i'm not sure. It makes changes in /etc/profile adding LC_ALL=ru_RU.UTF-8 env, and it also enables keymap switching via "Qi" key.. i don't think everyone needs it.. Somebody can even get confused - like "oh, and what's..03:41
kyak..that nice key with some Chinese symbol? Oh, no, i'm typing some weird language!".. Let users enable it via opkg install ben-cyrillic03:41
xiangfukyak: ok.03:41
xiangfukyak: from the source code. the after display the "loading... ", the stardict try to access network03:46
kyakis connected to my laptop at this moment, the network is up and running... from strace, one of the thrteads is segfaulted03:47
kyaki think i need to build gdb, because strace is not showing any more info03:48
xiangfukyak: can you send me the ".config" you using . I will try to test here.03:49
xiangfukyak: also I found the "Terminal" section not show up in gmenu2x. :(03:50
kyakxiangfu: yeah, i know...03:50
kyakxiangfu: http://downloads.qi-hardware.com/people/kyak/tmp/.config_stage1503:51
kyakyou need to run menuconfig and enable stardict03:51
wolfspraulkyak: it would be nice to have cyrillic working by default, same as others like chinese, japanese, korean, hebrew, arab, etc. _BUT_ you are right, we cannot just set LC_ALL to ru_RU03:51
wolfspraulso there would need to be some way to switch/cycle through03:52
wolfspraul(btw, I also think we should have an easy way to cycle through console fonts, with some hotkey)03:52
kyakwolfspraul: i was thinking about some first-time configuration wizard.. let people choose time zone, language etc03:52
wolfspraulsure, and it could be re-run from an icon in gmenu2x03:53
wolfspraulbut I do think more languages should be preinstalled in the image first, we do have space for some fonts and keymappings and stuff03:54
wolfspraulcyrillic definitely03:54
kyakmaybe it could be done in Qt, since we have it running :)03:54
xiangfuanother thing . we should add "poweroff" in gmenu2x , right? :)03:54
wolfspraulthen we also have lots of customers in Japan03:54
wolfspraulso better out-of-the-box Japanese would be cool03:54
wolfspraulwell let's start with cyrillic, since we have you :-)03:55
kyakrun fbterm, get Japanese display support :)03:55
wolfspraulget that working very well...03:55
wolfspraulyes sure03:55
wolfspraulbut we also need to have the fonts installed, needs to be thought through a little03:55
wolfspraulwe get there03:55
kyakxiangfu: we have a real button for power off :)03:57
kyakmaybe for "reboot"03:57
xiangfukyak: maybe. reboot, suspend, poweroff, .. :)04:05
kyakwhy not? i wonder if there is a way to ask user to confirm his action in gmenu2x04:07
kyakor maybe just to have a simple Qt wrapper for all those reboot, suspend, poweroff04:07
wolfspraulreboot is just off & on, no?04:08
wolfspraulsuspend - ideally the nano could do that himself on inactivity, until then or to expedite it there may be a key...04:09
xiangfuwolfspraul: while the NanoNote reboot. we can press "U", it's a easy way to goto usbboot mode.04:09
wolfspraulyes :-)04:10
wolfspraulI was wondering whether we need 2 different actions - off & reboot. Or just 'off'.04:11
wolfspraulreboot = off + on, no need for a special action, imo04:11
kyakwolfspraul: i use "reboot" pretty oftern.. it's faster than turning off, the holding the button to turn it on...04:11
wolfspraulok then :-)04:11
kyakxiangfu: regarding the suspend, i noticed that the screen won't go blank when gmenu2x is started04:12
xiangfukyak: when we press POWER buttion. there is a hardware delay. that's why it's slow. :(04:12
kyakxiangfu: btw, is it necessary to delay those 3 seconds before start? is it done to prevent turning on by accident?04:13
kyakxiangfu: oh, another idea - have the brightness control in gmenu2x :) since you implemented it.04:14
kyakmaybe two buttons - increase 5 % and decrease 5%04:14
xiangfukyak: in gmenu2x there is one setting option. "LcdBacklight" 0 ~ 100.04:17
xiangfukyak: we just need change the gmen2x code to implement it04:17
kyakdoes it work?04:17
kyakso i thought :)04:17
xiangfukyak: seems we just need change one line :fopen("/sys/class/backlight/pwm-backlight/brightness", "w+");  gmenu2x.cpp:151:04:19
kyakxiangfu: is there some hotkey to change the brightnes? or how it is done?04:20
kyakor it only changes when you start some application?04:20
kyakgmenu2x source - would be great if it realyl was that easy :)04:21
xiangfukyak: if we have "cmdpad" or "triggerhappy hotkeydaemon" then it's easy create hotkey.04:22
wolfspraulxiangfu: what happened to triggerhappy upstream?04:26
wolfspraulif they don't add it, let's just add it in openwrt-packages then it doesn't matter how slow upstream is :-)04:26
xiangfuwolfspraul: Florian have reply the email says, it's ok. but nobody have commit it to upstream.04:27
wolfspraulkyak: about the turn on/off speed, I doubt it's that much of a hardware issue, although individual units may differ in the speed. but I can tell you for sure that the proprietary software we are replacing is much snappier and faster in both turn on and turn off.04:27
wolfspraulfor the proprietary software, you see a reaction (boot screen) after < 1 second for sure when turning on.04:27
wolfspraulbasically you press the button down, and before you lift your finger back up you see a reaction04:28
wolfspraulit boots to a fully working system in ca. 4 seconds04:28
wolfspraulturn off < 1 second for sure as well, maybe even < .5 seconds04:28
wolfspraulso that's the gold standard :-)04:29
wolfspraulxiangfu: I saw that mail, well if nobody applies it then I suggest we just move forward in openwrt-packages04:29
xiangfuwolfspraul: ok.04:29
wolfspraulit doesn't cost us much time to add it there (10 minutes or so), and if upstream adds it fine, it takes another 10 min to remove.04:29
qi-bot[commit] Xiangfu Liu: update the brightness file path to NanoNote http://qi-hw.com/p/gmenu2x/be9c6b904:30
wolfsprauland it becomes more accessible to people, even others can add the openwrt-packages feed if they like04:30
qi-bot[commit] Stefan Tomanek: add triggerhappy hotkey daemon http://qi-hw.com/p/openwrt-packages/4dcfd3804:32
xiangfukyak: the brightness setting in gmen2x works pretty good04:46
qi-bot[commit] Xiangfu Liu: set default backlight to 25% http://qi-hw.com/p/gmenu2x/a4b8aa404:50
kyakwolfspraul: understood about turn on/off04:57
kyakxiangfu: good! will check it..04:58
qi-bot[commit] kyak: removed ncurses completely; have ncursesw build http://qi-hw.com/p/openwrt-packages/057dd6a05:00
qi-bot[commit] kyak: Merge branch 'master' of projects.qi-hardware.com:openwrt-packages http://qi-hw.com/p/openwrt-packages/dc9a53405:00
kyakxiangfu: wolfspraul ncursesw should build fine now, no conflicts with package/ncurses05:01
wolfspraullet's see whether mirko is still happy as well05:01
kyakof course05:01
qi-bot[commit] Niels: add layer that displays  points of interest from a .osm file http://qi-hw.com/p/nanomap/2d6e62a05:35
qi-bot[commit] Niels: add monav plugins from http://code.google.com/p/monav/ and use them to calculate routes http://qi-hw.com/p/nanomap/e18428c05:35
qi-bot[commit] Niels: draw start and end markers only when they are set http://qi-hw.com/p/nanomap/fd81fcf05:35
qi-bot[commit] Niels: emit a signal if the gps device got a fix http://qi-hw.com/p/nanomap/0dd2f3305:35
qi-bot[commit] Niels: draw different symbol if gps device got no fix http://qi-hw.com/p/nanomap/889c3fb05:35
qi-bot[commit] Niels: use monav plugin to do address look ups http://qi-hw.com/p/nanomap/a8c636005:35
qi-bot[commit] Niels: move route marker only if routing information is available for that point http://qi-hw.com/p/nanomap/6c6fabb05:35
qi-bot[commit] Niels: do not crash if plugins could not be loaded http://qi-hw.com/p/nanomap/8a3346405:35
qi-bot[commit] Niels: do not crash if no city is selected http://qi-hw.com/p/nanomap/223efe705:35
qi-bot[commit] Niels: a little start at getting routing directions http://qi-hw.com/p/nanomap/5e06ce005:35
qi-bot[commit] Niels: draw icons of the points of interest loaded from a .osm file http://qi-hw.com/p/nanomap/bdecafa05:35
wolfspraulwow that's so cool05:37
wolfspraulI want to make a list of cities where our customers live, and include maps of those cities :-)05:38
kyakwolfspraul: maps are in png, and they occupy rather much space :) for example, it took around 1 Gb to download Frankfurt and suburbs to the maximum zoom level05:50
wolfspraulkyak: well let's see. we start, learn, improve, etc.05:51
wolfspraulI'm a huge believe in OpenStreetMap05:51
kyaknice commit to NanoMap above, need to update it in pacakges,too05:51
kyakOpenStreetMap is reat05:51
wolfsprauljust the beginning05:51
wolfspraulI think it will be bigger and more meaningful than Wikipedia, say in 10 years05:51
wolfspraulwe live in our environment after all :-)05:51
lekernel_kristianpaul: (video on milkymist, and others) http://milkymist.uservoice.com (or better, send a patch :p)05:52
kyaki wonder how and where Niels is using NanoMap05:55
kyakhe added a lot of GPS support05:55
kyaklooks like NanoMap is tending to a desktop application :)05:56
kyaklatest git version won't compile -\05:57
kyakerror: QObject: No such file or directory05:57
wolfspraulkyak: hmm :-)06:01
wolfspraulit's still called NanoMap, so let's hope for the best...06:01
kyaki think Niels was not testing with Ben -\06:12
kyakit's broken at least from Oct-3006:12
wolfspraulok let's hope the memory requirements don't go through the roof06:15
wolfsprauland general speed etc.06:15
wolfspraulwe see06:15
qi-bot[commit] kyak: added qt4-demos and qt4-examples to full_system http://qi-hw.com/p/openwrt-xburst/dc0a85006:27
qi-bot[commit] kyak: updated gmenu2x to latest git. http://qi-hw.com/p/openwrt-packages/15da1f006:41
qi-bot[commit] Xiangfu Liu: [gmenu2x] update http://qi-hw.com/p/openwrt-packages/c68f12b06:41
qi-bot[commit] Xiangfu Liu: [gmenu2x] update to 20101115 http://qi-hw.com/p/openwrt-packages/812dd9006:41
qi-bot[commit] Xiangfu Liu: [gmenu2x] update, read /etc/profile before run gmenu2x http://qi-hw.com/p/openwrt-packages/1ba8cb606:41
kyakwolfspraul: do you know why http://downloads.qi-hw.com/ is not working? :)06:43
wolfspraulkyak: not setup06:43
wolfspraulqi-hw.com was only meant to work for selected URLs, as a shortener06:44
kyaksame as fidelio.qi-hw.com06:44
kyakah ok06:44
wolfspraulI mean if you like short urls, can can add those06:44
wolfspraulbut then we can even do qi-hw.com/f for fidelio :-)06:44
wolfspraulbut it adds maintenance overhead, so I do the short URLs one by one06:44
wolfspraulalso I don't want multiple urls pointing to the same place, it always causes headaches later06:45
kyakthis makes sense06:45
wolfspraulI setup qi-hw.com only for the commitlog06:45
wolfspraulso the lines coming out here in IRC are short, make sense, but are valid/active urls people can click on06:46
kyakyeah.. i remember that06:46
kyakdon't bother, i was just wondering if this is ok06:47
kyakB_Lizzard: hi! how's your OE-based distro going?07:16
B_LizzardFine, fine07:16
B_LizzardMade some optimizations here and there07:16
B_LizzardI'll need to start working on some applications07:17
B_LizzardPIM and the like07:17
kyakoh, great.. is PIM an input method daemon?07:17
kyakB_Lizzard: i remember you had a nice shutdown dialog. Is it Qt? Did you release the source code? i think it would be nice having this in openwrt...07:21
B_Lizzardkyak, Personal Information Manager07:21
B_LizzardContacts, Calendar etc07:21
B_Lizzardkyak that's just a shell script07:22
kyakah, yes, it's defintiely a worthy application.. could be tricky to fit it to 320x240 though07:22
kyakB_Lizzard: a shell script? what does it use as a graphics?07:22
qi-bot[commit] Mirko Vogt: remove <triggerhappy> since it went upstream http://qi-hw.com/p/openwrt-packages/18ed9b507:22
kyaki see.. some X11 thingy, obviously07:23
B_Lizzarduh huh07:24
kyakB_Lizzard: let me know when you have something new and interesting, i'm willing to test. Always many new ideas from jlime guys :)07:30
B_Lizzardthanks, I sure will07:31
wpwrakkyak: btw, when installing packages, does openwrt use opkg, like jlime does ? i.e., do you also have the "opkg needs swap to even run" problem ?07:58
kyakopenwrt uses opkg07:58
kyakopenwrt runs fine without swap07:59
wpwrakhmm. maybe that's just because there are fewer packages :)07:59
wpwrakbut you're using the same. that's already good. thanks.08:01
qi-bot[commit] Niels: first try to fix building with OpenWRT toolchain that does not include QtCore and QtGui in the include path http://qi-hw.com/p/nanomap/da88ec208:14
qi-bot[commit] Niels: second try http://qi-hw.com/p/nanomap/88239a208:22
kyakwpwrak: what are you using yourself with your Ben?08:24
kyakwpwrak: i've noticed your increased interest to jlime toolchain recently :L)08:24
wpwrakkyak: for now, my bens still have openwrt in NAND. one of them now has a uSD with jlime. (the other is glued to a board, waiting for the cnc mill to become available and the pcb scan to proceed)08:26
kyakwhat's cnc mill?08:27
wpwrakmy plan is to migrate to jlime, mainly because of the much larger choice of packages08:27
kyakhow long does it take for one scan?08:27
kyakwhat are the packages that you need and they are not in openwrt?08:27
kyakmany questions :)08:27
wpwrakkyak: this kind of device: http://www.rolanddga.com/products/scanners/mdx15/08:28
kyakyou don't have this at home, do you? :)08:29
wpwrakkyak: (packages) nothing specific, because i haven;t done much with the ben yet. it's just very likely that i'll eventually run into things that jlime/OE has that openwrt doesn't.08:29
wpwrakkyak: sure i do :)08:29
wpwrakit's not as monstrous as you may think. a bit larger than a printer.08:30
kyakwpwrak: my opinion: openwrt has around ten times less pacakges than OE, but it doesn't matter since openwrt is used by people and their needs are similar. howvere, with OE you have high changes to run into application that..08:30
kyak..might not work at all because it's untested08:30
wpwrakkyak: it lives is a box with thick sound-proof walls. so i can even do some light work a night, without getting death threats from the neighbours :)08:31
kyakwpwrak: show us a pic sometime :)08:31
wpwrakwolfgang took one when he was visiting earlier this year: http://en.qi-hardware.com/wiki/File:Roland_MDX-15_scanning_nanonote.jpg08:33
kyakmy point is - out of 2678 packages of openwrt, you should find all what you need08:33
kyakif you don't then you'll port  :)08:33
wpwrakthe picture shows the device scanning a ben. here's the result of some of the milling it did (mold for counterweights): http://downloads.qi-hardware.com/people/werner/cw/cw-mold.jpg08:34
kyakwpwrak: indeed not a big machine at all!08:34
kyakwpwrak: oh! it "prints" in wood?08:34
kyaki though it was plastic08:34
kyakoh, i got it wrong... it's a scanner, not a printer08:35
wpwrakit mills :-) it's a subtractive process, no additive. kristianpaul's machine "prints". mine has a "drill" that removes material08:35
wpwrakit's scanner and mill. it has two different heads. when i install the scan head, it's a scanner. when i remove the scan head and install the milling head, it's a mill.08:36
kyakdoes it create 3D models while it mills? or it uses 3D models to mill?08:36
wpwrakit moves the head along a so-called toolpath. the result depends on what that toolpath looks like.08:36
kyakcool.. what was your aim when you bought it?08:36
wpwrake.g., the mold was made by removing wood in a sequence of layers. i also cut pcbs with it, etc.08:37
wpwrak(pcb) e.g., these pcbs were cut with the mill: http://downloads.qi-hardware.com/people/werner/f32x/c2ben-run1.jpg08:38
wpwrakor this one http://downloads.qi-hardware.com/people/werner/wpan/atusd/tethered.jpg08:38
wpwrak(aim) mainly to learn about cnc an to be able to make things like cases08:39
kyakwhat about the stripes on PCBs?08:39
kyakhow the "got inside"?08:40
wpwrakyou mean the traces ? (the copper, covered with tin)08:40
kyakyeah, those08:40
wpwrakthey're not inside. they're on top of the board.08:40
kyakyou just glued it?08:40
wpwrakthe boards come covered with copper. then i "print" the layout on the copper and etch away the copper i don't needs.08:41
kyakvery intestering.. can you that with knife? :)08:41
wpwrak"print" = i print the layout on a special paper with a laser printer, then transfer it with heat+pressure (e.g., with a clothes iron) to the PCB08:42
wpwraketching is a chemical process :) i use a mixture of peroxide and muriatic acid08:42
kyakso you actually burn those traces out08:43
wpwrakit's not as scary as it may sound :)08:43
kyaknot cutting them08:43
wpwraknot burning. etching.08:43
kyakok, etching :)08:43
wpwrakthe transfer of the layout is just hot enough to transfer the toner, not to affect board or copper08:43
wpwrakit's the acid that does the work :)08:43
kyakso you're actually to produce your home-made PCBs...08:44
kyak*able to produce08:44
kyakthis is awesome08:44
wpwrakit's a nice capability to have :)08:44
kyakhow much does it cost?08:44
kyakone PCB08:44
wpwrakduh. hard to calculate.08:45
wpwrakdo you include machine use ? tool use ? (tool = the "drill") work ?08:45
kyakbut much less than if you have ordered in on factory?08:45
wpwrakno idea ;-) probably. and it's fast. i can go from layout on the pc to having the board ready for soldering within about an hour.08:46
rafabuilt emacs... what should be do?.. I just learnt C-x C-c :)08:46
kyakinclude the PCB cost and chemical costs (so only expendables)08:46
kyakor consumables08:46
kyakhowever it's called :)08:46
kyakrafa: take care not to get RSI :)08:47
kyakuse vim for your own good08:48
wpwrakthe chemicals are very cheap. the boards ... about USD 6 for a 10x15 cm board. how many circuits i can make from that one depends on the size of the circuit.08:48
wpwraksomething between 2 and 12 :)08:48
kyakwpwrak: do i understand correct that you don't need the millling machine to do etching?08:49
wpwrakrafa: i was about to give you a nice example, but then i noticed that i don't have an emacs on any of my machines ;-)08:49
wpwrakrafa: i think Esc X doctor  should do something nice08:49
wpwrakkyak: the mill is just for cutting the board. i could also do this with a dremel if i wanted to. of course, with the mill it's much more precise.08:50
wpwrakkyak: and boards with a compex shape are no fun with a dremel08:50
kyakdo you use some kind of press to put that laser printing on PCB?08:51
wolfspraulkyak: there is no way Werner can beat costs, probably not even on 1 PCB. but the 2 main advantages are gaining knowledge, and turnaround speed08:51
wpwrakkyak: e.g., in this case: http://www.almesberger.net/misc/ben/wpan/board-100813.jpg08:51
rafakyak: ;-))08:51
kyakwolfspraul: are you saying that one factory made PCB is cheaper than 6 USD? (not mass production)?08:52
wolfspraulkeep in mind what Werner can do there :-) 2-layer etc08:52
rafawpwrak: I ran.. it is saying me : "I am the psychotherapist. Please, describe your problems. Each time you are finished talking, type RET twice."08:52
wolfspraulcheck out batchpcb.com, for example08:52
wpwrakwolfspraul: actually, having small quantities of boards made isn't that cheap. after all, they need to recover their setup cost08:53
rafawpwrak: woww! :D08:53
rafawpwrak: and it has colors :)08:53
kyakwpwrak: very nice..08:53
wpwrakwolfspraul: and the ultra-cheap places are a fedex away ...08:53
wolfspraulyes true, fedex = 30+ USD08:53
wolfspraulor it's terribly slow again08:53
kyakis there a way to do 2-layer PCBs at home? :)08:53
wolfspraulyour process has some unbeatable advantages, like I said the 2 that stand out for me are knowledge and turnaround speed08:54
kyakglue together two 1-layers PCBs :)08:54
wolfspraulfor example for developing a PCB antenna :-)08:54
wpwrakwolfspraul: yup. they're the key advantages, particularly turnaround.08:54
wolfspraulfor something like batchpcb.com you will need x weeks, who knows how many08:54
kyakit's super rapid indeed08:55
wolfspraulthey bundle boards together08:55
wpwrakkyak: 2 layers is not a problem. you can buy boards that have copper on both sides. then you apply the same process on both sides. either at the same time or sequentially.08:55
kyakwpwrak: what about 4?08:55
wpwrakkyak: both at the same time is faster, sequentially is more precise. so you can choose :)08:55
wpwrakkyak: > 2 is a very very different story ....08:55
wolfspraulwpwrak: what I meant is at 4 layer, or more, HDI and what not, it's obvious that home-DIY is not an option. not today, not tomorrow. who knows when, if ever...08:55
wolfspraulbut that's not the point, the point is to focus on where it excels, and it does for many very specific problems08:56
wpwrakwolfspraul: yup. 4 layers, small vias, etc., all that's not an option, unfortunately08:56
wpwrakwolfspraul: there's some gruesome process for doing via "properly" but it's with quite a lot of chemistry. metal-acid solutions and such.08:57
kyakwpwrak: do you call the connection between PCBs a "via"?08:57
wolfspraulbatchpcb.com has a 10 USD setup fee, 2.50 USD / square inch for 2 layer, plus shipping08:58
kyaki always called it a "hole"08:58
kyakor whatever :)08:58
kyakbetween l[3~;ayers08:58
wolfsprauland you can even do 4-layer at batchpcb at 8 USD / square inch08:58
kyakterminal glitch08:58
wolfspraulin the US it's probably a really cheap option, since they can ship with US Postal mail08:59
wpwrakand multilayer needs more chemistry, high pressure, and so on. you can probably do it at home if you're really determined. powerful presses aren't so expensive. but it turns what would be a relatively straightforward and easily automated process into a lot more manual work. and one problem of the manual bits is their error rate.08:59
wpwrakkyak: via = hole with copper that connects both sides08:59
kyakwpwrak: thanks ;) i ahve to translate sometimes things to English and i called that "interconnection hole"09:00
kyakbut i was understood :)09:00
wpwrakkyak: yeah, it's just understandable :)09:00
wpwrakrafa: and, did it help ? ;-))09:01
Guest90570can anyone assist me with an asus motherboard purchase?09:01
kyakwpwrak: i assume you want rafa to run into one of your tricks, don't you? :)09:02
wpwrakkyak: naw, i just showed him something where emacs is better than vi :)09:02
kyakthat's a hotkey in emacs to clean swipe your hard disk and call a doctor automatically :)09:03
rafawpwrak: I need to re read the whole chat with the doctor... the nn screen is small.. but naving emacs text is not hard it seems. He is mainly saying me that I am the problem!!.. Is there some way to change the psychotherapist there? :)09:03
wpwrakrafa: we should invite silvina to write an update for doctor.el ;-)09:05
wpwrakrafa: oh, i heard a story that there's some guy who did the following experiment: he got two psychiatrists and told each the other was delusional, thinking he was a psychiatrist. then he set them loose on each other ;-) alas, i don't know how this turned out09:07
kyakin turned out in some butt kicked09:07
kyakand double bill :)09:08
wpwrakkyak: i'm not sure this is an approved treatment method. they do have electroshocks, though ...09:08
rafawpwrak: silvina would do better questions for sure :)09:08
kyakhmm i just noticed there are three simultaneous builds on build host :)09:10
kyaki stopped mine for good09:10
wolfspraulkyak: why? too slow?09:11
kyakpretty slow yes09:11
kyakload average: 2.47, 2.85, 2.9909:12
wolfspraulgood, at least we are not wasting money09:13
kyaknot a bit of it! :)09:13
wolfspraulwell let's see09:14
wolfspraulthe buildhost will probably get more work over time09:14
wolfspraulsay for example if we do the planned layout history brdhist09:14
wolfspraulor other server-side scripted stuff09:14
wolfspraulso maybe if we are all still happily building there, it's time for a more powerful machine :-)09:15
wolfspraulkyak: but right now I'd say if you just wait a day or so, it should slow down. mirko is trying to get his release out, and xiangfu is constantly building something because he plays with the config files so much.09:15
wpwrakwe'll eventually manage to bog it down ;-)09:16
wolfspraulwhen I play/test the config files, I also like to just let the buildhost crunch on them... slow or not... so I can focus on more low-latency stuff on my notebook09:16
kyakwolfspraul: not a problem.. i'm building on my host, and buildhost is interesting to me because it's 64bit. But i don't really have anything to test right now09:16
wolfspraulsure sure, but I want you to be happy about the buildhost, not turn away in disgust :-)09:16
wolfspraulso please keep the feedback coming09:16
wolfspraulI was a bit stingy with the 29 EUR buildhost, because I first wanted to see how much use we can squeeze out of the machine09:16
wolfspraulbut now we already use it for 2 really nice use cases: schhist, and people building openwrt images09:17
wpwrakwolfspraul: btw, you mentioned that you're planning to do more things with boom sometime soon. about when, and what ? i'd like to make sure to set aside a bit of time for that.09:17
wolfspraulmy todo juggling09:17
wolfspraulyes I'd love to start09:17
wolfspraulyour Xue TODO list is awesome!09:18
wpwrakthanks :) i notice that it silenced them ...09:18
wolfspraulI would most likely start there.09:18
wolfspraulAndres already gave me commit permission, he he09:18
wolfspraulbig mistake! :-)09:18
wolfspraulI will apply my super-superficial jtag-serial skills on that much bigger project now :-)09:19
wpwrakgood luck with the beads/inductors/filters ! ;-)09:19
wpwraki notice that they're a recurrent problem :)09:19
wolfspraulmany things to learn09:19
wolfspraulbut it's OK09:19
wolfspraulthose are valuable missing pieces in our copyleft hardware story09:19
wolfspraulso I'm up for it09:19
wolfspraulabout timing, don't know09:20
wolfspraulno need to reserve time for my stupid questions09:20
wolfspraulI come from the manufacturing side09:20
wolfspraulfirst priority is still milkymist one rc209:20
wolfspraulmany things moving there, like roh's cases, jtag-serial daughterboards, power adapters (local regulations), etc. etc.09:20
lekernelwolfspraul: what xue todo list?09:20
wolfspraulafter that is Xue, but I will probably start to get my hands dirty before Adam jumps into Xue09:21
wpwrak(timing) it's about having time to enhance boom where it's needed and such. e.g., by adding groups of components and such. i basically work these things on an on demand basis.09:21
wpwrakthat is, if i don't need anything new, very little happens.09:21
wpwrak(power adapter) oh dear. you let yourself get caught in this quagmire ...09:22
wolfspraullekernel: http://lists.en.qi-hardware.com/pipermail/discussion/2010-November/006097.html09:22
wolfspraulwpwrak: don't worry I understand that09:22
wolfspraulwe cannot all point fingers at Werner09:22
wolfspraulI think I get the architecture of boom, it has a strong and extendable design09:22
lekernelah, yes09:23
wpwrakwolfspraul: (boom) i also want to rewrite the whole boom core in C. it's already getting quite slow and we're not nearly at a comprehensive set of components.09:23
wolfspraultrue, it's slow, but then again, we are still in the < 30 seconds range09:23
wpwrakwolfspraul: (boom) and then, add some goodies like a part database browser. right now, it's too hard to find things there.09:24
DocScrutinizerI got a whole 'book' full of beads/filters, if you have any use for that :-P09:24
wolfspraulDocScrutinizer: hey Joerg!09:25
wolfspraulwe need you! :-)09:25
wolfspraulare you going to 27c3 btw?09:26
DocScrutinizernope, no money09:26
wolfspraulis Dieter going?09:26
DocScrutinizerdon't know09:26
wpwrakthen, the workflow also has some issues. e.g., right now, i let it proceed past errors, so that one can ignore problems at one stage the work on bringing the things that almost work to completion. this is nice to far. but it has the disadvantage that it's easy to overlook problems. example: if a component can't be matched, it will not be fed to the inventory lookup. so you get a BOM simply without that component, with only a warning in09:26
wpwrak an early stage but no further complaint.09:26
wpwrakso lots of little issues that need addressing at some point in time. just let me know when you'll embark on that journey and i can do a bit of trailblazing. easier for you if things get cleaned up before you hit the problems than encountering them and being uncertain whether they're part of the design or just bug :)09:28
wolfspraulDocScrutinizer: ok let me email a friend of mine in Berlin see whether I can find a cheap place to crash09:28
wolfspraulI start to like the underground life :-)09:28
wolfspraulI would love to hookup and get an update on things, maybe we find a way...09:29
DocScrutinizerwolfspraul: don't bother about a place to stay. I probably could find such place. But tickets were sold 11.11. and travel, diner etc aren't for free09:30
wolfspraulI'm not planning to stand in line for those tickets anyway :-)09:31
kyakwhat's 27c3?09:31
wolfspraullet me see what I can do, also find out whether Dieter is going09:31
wpwrakwolfspraul: how do you plan to sneak in ?09:31
wolfspraulI haven't started any serious 27c3 planning yet09:31
wolfspraulno sneak in, I just stay out09:32
wolfspraulI don't like how they are managing access anyway. I like public space.09:32
wolfspraulshould be held in a public space I think.09:32
DocScrutinizerkyak: world's largest hacker meeting09:32
kristianpaulwolfspraul:  nano could do that himself on inactivity <- power button press 1s suspend, more  poweroff09:33
kyakDocScrutinizer: it's soo cool google doesn't know it ;)09:33
kristianpaulwolfspraul: propietary sofware for ben have suspend support?09:33
DocScrutinizeror maybe at least the most honored one09:33
wolfspraulgood question don't know09:33
wolfspraulprobably not, it's so fast on power on/off09:33
kristianpaulkyak: nanomap  <- hope it sill fast and seems to be right now09:33
kristianpaulwolfspraul: yeah current OE?Jlime image wich have suspend support thanks to larsc dies after some hours09:34
kristianpaulmay be power arranage was not tought for suspends stabillity..09:34
kristianpaulwprak jlime, ohh i tought you had bad time in openmoko mostly because bitbake bloatness ?...09:35
kristianpaulwpwrak: but indeed lots of packages are always good :)09:35
qi-bot[commit] Xiangfu Liu: add wallpapers from community contributions http://qi-hw.com/p/gmenu2x/7d1a69909:35
kristianpaullekernel: feature request page for milkymist looks empty, are u blogging or telling the world about it? :-)09:36
wpwrakkristianpaul: i had a horrible time at OM with OE until i stopped trying to use it as a build environment :)09:37
kristianpaulkyak: 27c3 nice event i discovert it last year, seems germany is a good place for geeks or related people :)09:37
wpwrakkristianpaul: once others provided the packages and i could just ask them for anything i needed and tell them where i kept my code and they would package it for me, things went very smoothly09:37
wpwrakkristianpaul: so the key to OE success is outsourcing ;-)09:38
kristianpaulwpwrak: sure relay in others is good :)09:38
kristianpaulthanks rafa B_Lizzard !! :)09:38
kristianpaulwpwrak: ah you made deals with rafa just because both live near ;-)09:39
kyakkristianpaul: yeah, looks interesting! though geeks are usually antisocial and only gather together to drink beer ;)09:39
kristianpaulwolfspraul: indeed why bother when device is really fast powering up09:39
DocScrutinizerkyak: you'd be surprised09:39
kristianpaulwolfspraul: about this propietary sofware, it is a OS or a standalone app? i'm just curios what they find better to run on the ben09:40
wolfspraulnot sure, probably some microkernel plus apps09:40
wolfspraulI don't know much about it, never cared09:40
kristianpaulmicrokernel :D09:40
wolfspraulI could ask and all, but why...09:40
kristianpaulyeah time wasted09:41
wolfspraulit works well though, and still beats us on several key metrics09:41
wolfspraullike screen stability (there is still more flickering with Linux than with the proprietary sw), boot time, shutdown time09:41
wpwrakkristianpaul: naw, it's coincidence that rafa is doing just what i like :)09:41
kyakwhen you are talking about "proprietary software" for Ben, you mean it's original electronic disctionary instance?09:41
wolfspraulbut we got A LOT closer over the course of last year, and I am optimistic about the future09:42
kristianpaulwolfspraul: at least have a better idea of its featires in order to know in wich it still beating09:42
wolfspraulyes, it's a whole suite of software09:42
wolfspraulincluding makeup tips for girls and what not09:42
wolfspraulit's nice, definitely a nice package09:42
wolfspraulall sorts of useful little things09:42
wolfspraulkristianpaul: those 3 stand out, after that we should be OK: screen stability (flicker), boot time, shutdown time09:43
DocScrutinizerhave fun with screen flicker!09:44
DocScrutinizera bitch09:44
wolfspraulDocScrutinizer: it got much better over time. What we have today on OpenWrt, and probably also Jlime/Debian is definitely usable09:44
DocScrutinizer1st you need to make sure if it's LCD or backlight09:44
kristianpaulDebian? :O09:45
kristianpaulah *probably*09:45
DocScrutinizerseen BL flicker on several devices09:45
wolfsprauljust when comparing say the latest openWrt image and proprietary software side-by-side, it's still evident that ours has more flickering09:45
wolfspraulthis does differ slightly from one device to another, but even if you look at 10 of them, we need to admit that for some reason their software produces a more stable image09:46
wolfspraulbut again, it got A LOT better over the last months09:46
wolfspraulit wouldn't even be high on my priority list anymore with where we have it now09:46
kyakhas it been improved from the software side?09:46
wolfspraulI just answered where the proprietary software still beats us, and this is one of them.09:46
DocScrutinizerso it's LCD driver and settings?09:46
wolfsprauloh sure, a lot09:46
wolfsprauldrive IC settings, timings, clocks, etc.09:46
wolfspraullarsc knows the whole story09:47
DocScrutinizerand scheduler, IRQ handlers... whatnot09:47
wolfspraulit's good now, usable!09:47
kyakcan't we just copy those? ask the original manufacturer about those settings09:47
wolfspraulthe proprietary sw is still a tad better, that's all09:47
wolfspraulit's not so easy, we spent a lot of time on it already09:47
wolfspraulscreen flicker is good now09:47
DocScrutinizeryeah, a buzz-class problem09:48
wolfspraulwe have lower hanging fruits to go after09:48
wpwrakkyak: not a bad idea. provided contamination issues aren't a problem09:48
DocScrutinizerthough this is fixable by sw it seems09:48
DocScrutinizersettings as in register values aren't (c) anyway09:49
DocScrutinizerread out and copy09:49
kristianpaulkyak: printer <- afaik i cant scan ,but i dont really care for that09:49
DocScrutinizerfor timing issues caused by IRQ response latency that's a completely different game09:50
DocScrutinizeryou might have to do awful things like sync scheduler timedisk to VSYNC09:51
DocScrutinizer...completely rework IRQ handler structure, process table handling, etc09:52
DocScrutinizerplus investigation/debugging is a nightmare09:52
wpwrakDocScrutinizer: why would IRQ matter ? i think it's flicker of static content, not necessarily flicker caused by updates unsynchronized with vertical blanking09:53
DocScrutinizerNo idea about how you manage your framebuffer09:54
wpwrakDocScrutinizer: (reg values) the issue with looking at the sources is that they won't just contain the register settings. so someone could claim that you've copied some of the other stuff09:54
DocScrutinizerwhat is locking what in hw, etc09:54
qi-bot[commit] Xiangfu Liu: [config.full_system] add nanoterm http://qi-hw.com/p/openwrt-xburst/5a89e8109:55
DocScrutinizerwpwrak: (flicker) even such nice effects like VDD noise caused by different CPU loads might cause timing jitter on an oscillator...09:56
wpwrakDocScrutinizer: okay, that's evil ;-) but if the original sw doesn't have a problem there, chances are we shouldn't either09:57
DocScrutinizerthere's zillions of possible contamination pathes for things like LCD jitter09:57
wpwrakDocScrutinizer: besides, there's an *army* of caps in the LCM ;-)09:57
DocScrutinizernr1: stop *all* processes, see if flicker changes09:58
DocScrutinizerto rule out it's an actual sw load and operation induced issue09:58
wpwraki wonder if it's possible to properly quantify the damn thing. that way, one could compare different register settings and such.09:59
DocScrutinizerthen go insert NOP into driver loops etc09:59
DocScrutinizer(quantify) that'S another problem that quite resembles buzz issue10:00
DocScrutinizerIOW: sounds like fun10:01
DocScrutinizerto quantify you first need to *understand* the nature of flicker. Do proper video recording - then analyze frame by frame, and exactly spot the differences that cause the flicker10:03
DocScrutinizerwill teach you a lot about possible root causes10:04
wpwrakvideo recording may be tricky. any changes with a photo transistor or such ?10:04
DocScrutinizernope, not as long as you don't even know if it's a weird form of interlace flicker aka hsync freq interference, or a moving black/bright bar, or whatever10:05
DocScrutinizersporadic noise causing a dither of pixels that are off the correct value10:06
wpwrakin either case, you should be able to detect a variation of brightness within a sufficiently small area. granted, with just one "pixel", you won't be able to tell how this relates to adjacent areas10:07
DocScrutinizeror maybe just a VSYNC jitter that makes the img jump up and down10:07
wpwrakokay, perhapt the first step would be to find patterns where it does/doesn't happen10:07
wpwrakthat ought to tell something about the nature of the flicker10:08
DocScrutinizerand proper video recording and A/B or differential analysis10:08
DocScrutinizerstart with a plain white screen and see if it still is flickering10:09
DocScrutinizerdo same with a gray screen10:09
DocScrutinizerthen go to different checkboard patterns10:09
wpwrakthen hline/vline pattern, etc.10:09
wpwrakor maybe primary colors first10:10
DocScrutinizerstill a crappy video that can be viewed frame by frame is a powerful tool10:10
DocScrutinizershould be done with same freq as VSYNC, or a multiple10:11
DocScrutinizersome few cameras can do slomo10:12
wpwrakmay be difficult to find such a camera. i mean without spending money :)10:12
DocScrutinizereven if you can't get slomo/sync, it's still worth a try10:12
DocScrutinizeryou may 'test' camaras in shop10:13
DocScrutinizeror visit you nearby peoples' media center10:14
DocScrutinizeryou even might strob the BL :-P10:14
kristianpaulHey i have my Canon camera running chdk i cat take shots up to 1/100k aperture speed !10:14
kristianpaulif that may help10:14
DocScrutinizersounds good10:15
DocScrutinizernow if you could sync it to VSYNC...10:15
kristianpaulDocScrutinizer: what aperture speed should i try?10:15
DocScrutinizeror flash an LED during VSYNC strobe, and just take enough pictures to get a few that were taken same relative point in time in relation to VSYNC10:16
DocScrutinizerI'd suggest both a aperture time of = one frame of LCD aka VSYNC period, and ultrashort, as short as possible, and hope you catch some unambiguous takes where you can understand what's going on10:18
DocScrutinizerdoing this with the patterns mentioned above may be a tedious way to get same results like with a proper slomo cam10:19
DocScrutinizeralso scoping LCD bus VSYNC, HSYNC, pixel clock, and data should tell you something10:20
Action: xiangfu will remove the fn-handle project in projects.qi-hardware.com since we have triggerhappy or cmdpad in openwrt which is much better.10:20
DocScrutinizeractually should tell a lot10:20
DocScrutinizerdo this for both sw stacks, the flickering and the non-flickering one10:21
DocScrutinizerpst the scope screenshots or data files10:22
DocScrutinizerso nerds like me can drool over them10:22
DocScrutinizerthat's probably a job for wpwrak10:23
DocScrutinizeras he excels on such tasks10:23
qi-bot[commit] Xiangfu Liu: remove fn-handle and cmdpad, add triggerhappy http://qi-hw.com/p/openwrt-xburst/63e598b10:24
wpwrakhe's currently busy in a completely different war ...10:24
DocScrutinizerwpwrak: btw how's RF noise in your new flat?10:24
wpwrakhmm, about 2 dB ambient, i think. most of the noise seems to be homemade10:25
DocScrutinizermuch better for scoping, than when living in a giant microwave oven :-P10:25
wpwraki now filter my antenna signals with: 1) the USRP's tuner, 2) decimation, 3) limit the frequencies in the FFT, 4) throw away too small amplitudes, 5) reject total power that's more than 2 dB from the average, 6) do all this over 100 runs.10:26
DocScrutinizerhah, still that antenna design?10:27
wpwrak(that's for measuring the antenna's frequency-dependent sensitivity)10:27
wpwrakyeah. messy business without the right tools ...10:27
DocScrutinizertoldya, didn't I10:27
qi-bot[commit] Xiangfu Liu: remove fn-handle http://qi-hw.com/p/openwrt-packages/1cf24a110:28
wpwrakwith the right equipment, it would be connect, press a button, and there it is10:28
wpwrak(microwave oven) my scope is quite happy with the new surroundings :)10:28
DocScrutinizerthen send some meaningless diagrams to customer, which are puzzled by one diagram not matching the other :-P10:29
wpwraki'm working on the meaningless diagrams ...10:29
wpwrak... just did a first full run. now the 2nd is in progress.10:29
wpwrakthen i'll see how much the same configuration differs from itself ;-)10:29
DocScrutinizerand add some nonsensical tech speack, and make customer move speaker and eventually abort project10:29
DocScrutinizerhell, if time had bothered to send a "commercial" antenna from arbitrary garmin et al devices to that antenna design house, back when10:31
wpwraki think i'll just add filters until i get some result. worst case, i reverse the amplitude filter, reject the signal and just average over the noise. that should be nice and even ;-)10:31
DocScrutinizerthe GPS ant for gta0310:31
wpwrakyou're saying they did their measurements wrong ?10:32
DocScrutinizerwe spotted that, and clearly told, but nobody cared10:32
wpwraki remember there were some oddities10:32
wpwraklike the same point in one plane looking quite different in the other plane10:32
DocScrutinizerthe value for 0° was differing in X and Y plane10:33
DocScrutinizeryep, exactly10:33
wpwrakbut then, i don't know to what extent such things can be expected to happen. e.g., if there's an implied change of polarization10:33
DocScrutinizerwell, anyway a "this diagram doesn't 'look' nice enough" isn't a valid rating for a GPS antenna10:34
DocScrutinizeresp if there's quite a lot of other oddities in same test series, that aren't explained at all10:35
DocScrutinizerthey shipped some crap as they knew we never can check if it's proper or not10:35
wpwrakyou mean OM already had earned a reputation ? ;-)10:35
DocScrutinizerthat antenna house for sure knew what we possibly can do and what we definitely can't10:36
wpwrakthen they knew a lot :)10:37
DocScrutinizerthat's not difficult to know10:37
wpwraki mean the sheer quantity of data. particularly in the latter category :)10:38
DocScrutinizera) OM was rather open. b) rather noobish in some respects c) we asked for lowest price and probably hadn't any quality requirements10:38
DocScrutinizeryeah, that's true as well :-P10:38
qi-bot[commit] Xiangfu Liu: add poweroff.conf for triggerhappy http://qi-hw.com/p/openwrt-xburst/5d74e2310:38
qi-bot[commit] Xiangfu Liu: add Berlin as the NanoMap example maps http://qi-hw.com/p/openwrt-xburst/d5c1bde11:10
wpwraknice. reproducible results :) http://downloads.qi-hardware.com/people/werner/tmp/refant.png11:13
qi-bot[commit] Xiangfu Liu: add nightsky icon png file http://qi-hw.com/p/openwrt-xburst/7b42d0811:24
xiangfuwpwrak: there are two different lines :)  don't understand the hardware :)11:35
wpwrakxiangfu: it's a measurement of the strength of a signal sent from one device to another, at various frequencies. the objective is to found out if there are any frequencies that don't work well. e.g., 2410 doesn't look too good in this example.11:37
wpwrakxiangfu: the two graphs differ only a little, so i think (hope :) that's good enough11:38
kristianpaul-30db looks for a few meters...11:40
xiangfuwpwrak: thanks .11:40
kristianpaulhow is that in mili Wats ?11:40
wpwrakkristianpaul: decibel of whatever ;-)11:41
wpwrakkristianpaul: the distance is about 3.5 m11:41
wpwraki don't have any calibrated equipment, so i can't tell what the energy really is. nor can i compensate for frequency-dependent sensitivity of the receiver or the "fixed" antenna11:43
wpwrakkristianpaul: to make it more interesting, the sender is "imperfect", namely an atusb board. so absolute values mean nothing. the only thing that matters are differences at the same frequency.11:48
qi-bot[commit] Werner Almesberger: usrp/: new tools to run, evaluate, and plot a full frequency scan http://qi-hw.com/p/ben-wpan/3684c0711:52
qi-bot[commit] Werner Almesberger: usrp/README: clarified distance between antennas and fixed typos in examples http://qi-hw.com/p/ben-wpan/b7ad7f211:52
qi-bot[commit] Werner Almesberger: usrp/fscan: fail if the directory name begins with a dash http://qi-hw.com/p/ben-wpan/e31144f11:52
qi-bot[commit] Werner Almesberger: usrp/plscan: instead of `echo ... sed ...`, just use ${n%%.*} http://qi-hw.com/p/ben-wpan/45d594611:52
qi-bot[commit] Werner Almesberger: usrp/README: described how to perform and evaluate a measurement run http://qi-hw.com/p/ben-wpan/66f6ad611:52
wpwrakgood. wolfgang's scripts resist meta-characters :)11:52
kristianpaulwpwrak: you're afraid to accidentally attack qi servers ? ;-)11:53
wpwrak"afraid" may be the wrong word. more like "curious" :)11:54
wpwraknow the same experiment, at 2.1 dB less ...11:55
kristianpaulwpwrak: yes sorry, you dont read like a afraid person :) NO way11:55
wpwrakye who enters RF, abandon all fear :)11:57
kristianpaulfreedom not fear :-)12:08
kyakhm, triggerhappy won't build12:13
kyakevtable_KEY.h:244: error: 'KEY_RFKILL' undeclared here (not in a function)12:13
qi-bot[commit] Werner Almesberger: usrp/plscan: truncate title at the last dot, not the first http://qi-hw.com/p/ben-wpan/59beaba12:14
wpwrakkyak: i thought you guys were a bit trigger-happy with kicking out all the old stuff so quickly :)12:15
kyaktriggerhappy makes me sad :)12:16
wpwrakkyak: because it doesn't build ? or because of the work you put in the other stuff ?12:17
wpwrakhmm, i guess i should have squared the values after all. ah well, it's just a scale variation ...12:51
wpwrak(refant is +2.6 dBm, ref-0.5dBm is +0.5 dBm, ref-2.2dBm is -2.2 dBm.) http://downloads.qi-hardware.com/people/werner/tmp/refant-txpwr.png12:52
wpwraknow, let's try a PCB antenna :)12:53
kyakwpwrak: i didn't put anything into fn-handle, it was xiangfu13:08
wpwrakkyak: so he undid his own work. better for you :)13:23
kyakheh yeah13:24
kristianpaulsige used to have just real output..14:05
kristianpaulso the examples i got for se4110 are slight outdate for se4162 until i get it real data again, or decide to work with complex14:06
kristianpaulah is tricky14:07
kristianpaulthe decrease some clocks rate to the half14:07
kristianpaulhmm good point :)14:08
kristianpaulhmm is a code reused from USRP..14:10
Action: kristianpaul got the firmware for the gn3s GPS SiGE dongle14:12
qi-bot[commit] kyak: build abook against ncursesw http://qi-hw.com/p/openwrt-packages/ed2d09c14:13
kristianpaulGPIF... ??14:16
qi-bot[commit] kyak: use more sane 'linux' TERM http://qi-hw.com/p/openwrt-xburst/70c8ce314:18
qi-bot[commit] kyak: updated Qt keymap http://qi-hw.com/p/openwrt-xburst/581829714:18
qi-bot[commit] kyak: call setfont2 before starting abook http://qi-hw.com/p/openwrt-xburst/21f6b0614:18
kristianpaulkyak: had you tried nanomap recently?14:18
kyaki did14:19
kristianpaulkyak: btw i have new clean system. let me know when the repo is ready to test14:19
kristianpaulkyak: how fasts feel?14:19
kyakNanoMap is working very good14:19
kristianpaulis it last commit or same as some moths ago/14:19
kristianpaulgreat !14:19
kristianpauli want try its gpsd support14:19
kyakthe NanoMap if from September git i think14:20
kristianpauland las time i tried it was outdate14:20
kyakso it is14:20
kristianpauland i tried copmile upstream version and boom14:20
kristianpaullots of erros and not clear support from Niels14:20
kristianpauli think all is lists..14:20
kyaki think he did something today14:20
kyakto make it build in openwrt14:20
kristianpaulgood !14:20
kristianpaulfinally :)14:21
kristianpaulkyak: are you using xianfu scripts to compile openwrt?14:21
kyakso you can try.. just update the git commit in NanoMap/Makefile14:21
kyakno, actually i'm not..14:21
kristianpauli was thinking get it all over again14:21
kyakkristianpaul: regarding the repo, it's always ready to test :)14:21
kristianpauli was wating qt issue get solve14:22
kyakthe image right now looks very good14:23
kyaksome of the most annoying bugs are fixed14:23
kyakthough i wonder who made that wall paper14:24
kyakwith Socrate14:24
kyakthe icons are unreadable!14:25
kyakjust awfull wallpaper14:25
kristianpaulwell it looks good in wikicommons14:27
kristianpaulbut indeed i like the one before it14:27
kyakit looks good when you see it on PC screen14:27
kyakimaging small icons all around it :)14:28
kyaki'll try to make a screenshot ]14:28
kristianpaulhehe yeah is fun14:28
kyakkristianpaul: judge by yourself, if "alsamixer" text is readable14:39
kyakbelieve me, it looks much worse on the screen14:39
rafa:) ...he.. there is an icon for bc ? ;-)))14:41
kyakrafa: of course :)14:41
rafanow.. there is a "ls" missing icon yet14:41
kristianpaulthe head of the guy si a icon !!14:41
rafa[patch] new icon launcher : pwd14:41
kyakrafa: you're misjudging.. bc is an interactive application14:42
kyakwho cares if it's console or not14:42
kristianpaulso not put wallpaper with heads on it14:42
rafakyak: I use bc every day.. but it is just a command.. still if it has a interactive mode.. :)14:42
rafaI just would add a shell icon.. and user know what to do14:43
kyakrafa: http://projects.qi-hardware.com/index.php/p/openwrt-xburst/source/tree/master/data/qi_lb60/files/usr/share/gmenu2x/sections14:43
kyakwhat other icons are there :)14:43
rafakyak: if user does not know what bc exists to be ran then he will not know what to do when he runs it14:43
kristianpaulnow it looks like dingux screen but more blacky ;)14:43
kristianpaultry avoid so much console base apps icons14:44
kyakrafa: so you are saying that the user can't type 1+2 in the window that will open?14:44
kristianpaulat least you made a nice guy like the sound recorder rafa did14:44
kristianpaulwich btw is already ported?14:44
kyakkristianpaul: i'm not inot the gui thing anyway.. also not interested in drawing icons :)14:45
kyakkristianpaul: what's ported you say?14:46
rafakyak: no, the user will not write "1+2" if he runs calc and he does not know what is calc14:46
kristianpaulrafa: do you have upstreamed code for the soundrecorder you made in last jlime beta?14:46
rafakyak: sorry.. no calc.. bc.. which is worst14:47
rafathe name does not say anything14:47
kyakso what? he will discover14:47
rafakristianpaul: it is already there on front your eyes :D.. sound recorder is just a simple script of 20 or 30 lines ;)14:47
rafakristianpaul: let me find the file14:48
kyak"GMU" doesn't say anything as well14:48
rafakyak: yes.. sure.. bc is too intuitive :D14:48
kyakor lynx14:48
kyakwhat the hell is lynk?14:48
kristianpaultext based browwer14:48
kyakkristianpaul: that was irony :)14:49
rafakyak: the user sees the icon.. it says "bc".. cool.. maybe it is a game.. he runs it.. the main bc screen says :14:49
rafabc 1.06.9414:49
rafaCopyright 1991-1994, 1997, 1998, 2000, 2004, 2006 Free Software Foundation, Inc.14:49
rafaThis is free software with ABSOLUTELY NO WARRANTY.14:49
rafakyak: intuitive14:49
rafaFor details type `warranty'.14:49
rafakyak: he will start to write " 2+5"  3*4"?  :)14:49
rafakyak: and now the fun part.. the user needs to know how to exit ...14:50
kristianpaulkyak: it is14:50
rafabc: ctrl+C nothing.. "exit+ENTER" nothing14:51
rafahelp+ENTER nothing14:51
kyakrafa: ok, i have very small idea about user-friendliness..14:51
kyakctrl+c will exit btw14:51
kyak(interrupt) Exiting bc.14:52
kristianpaultake off the baterry always works ;-)14:52
rafakyak: CTRL+C14:52
rafayou sure?14:52
rafakristianpaul: battery : yeah ;))14:52
rafakyak: so it is intuitive if ctrl+c works and he never realized that bc is a calculator?14:53
kyakrafa: you are right that "bc" might mean nothing to a regular user. i'm not sure though that there is 1 % of users among 1000 buyers that don't know what bc is14:53
kyakand there is certainly noone who doesn't aware about ctrl+c14:53
kristianpauljust let the icons with colors kyak14:53
kyakkristianpaul: i don't argue, just do those icons :)14:54
Action: kristianpaul well i have commit permisions too just wait i got this source code again 14:54
kristianpauli meant remove the console based apps14:54
rafakristianpaul: can you check /usr/bin/recorder.sh ?14:55
rafakristianpaul: that file is a simple script which does the recorder application :)14:55
kyakkristianpaul: you might want to discuss it with someone else.. i wasn't the one who added them, and i'm good with them14:55
kyakin fact, i don't have gmenu2x started by default, so...14:56
kristianpaulrafa: damn no i let my ben at home :(14:56
kristianpaulrafa: ok14:56
kristianpaulkyak: i will14:56
Action: kristianpaul writing mail14:56
rafakyak: I have seen users asking how to install applications.. so I am sure that a big % of nn users do not know bc14:56
rafakristianpaul: I can upload the file.. that is okey.. it is just a text file14:57
kyakinstallign applications is a specific thing to every distro14:57
rafakristianpaul: if you want to check now14:57
kristianpaulrafa: later :)14:57
kyakbc is a common knowledge :)14:57
rafakyak: yes sure :)14:57
kristianpaul"Debian is slow, as far as applications... well, I can run emacs. This                                                        "14:59
kristianpaulis a huge win. But it is damn slow to power on14:59
kyaklong boot time15:00
kristianpaulmm nighsky need icon15:00
kristianpaulalsamixer.. well it need a better gui15:01
kyakgui for alsamixer is too mcuh :)15:01
kristianpaulyou should see jlime :D15:02
kyaki think i didn't?15:02
kyak*you think15:02
rafakristianpaul: qi openwrt can not run X app15:04
kristianpaulkyak: B_Lizzard version in OE15:05
kyaki know15:06
kyakrafa: is jlime also upstream?15:07
rafatelnet, ftp, python, perl, gdb, ed are all interactive.. should have icons=15:09
rafakyak: yes and no.. current beta 4 has a lot of ideas already on upstream. Blizzard is doing that work.. if he thinks that some ideas from betas are okey for upstream he sends patches to kriss and kriss push that on OE.. There are other ideas from B_Lizzard which goes to upstream as well. SO well, that is all stuff from jlime.. so yes.. jlime is on upstream15:11
kyakrafa: do you know only "hot" and "cold" and never heard about "warm"? A calculator, even console-based, should be available in one button click. It is also reasonable to add some other usefull console apps which a frequently..15:12
rafakyak: btw, you said that maybe some applications on jlime repo might no work .. can you tell us which ones? (I am talking about repi on qi servers built from OE)15:12
kyak..used. This should be adequate and nobody is trying to add all interactive console apps in gmenu2x.15:12
kyakrafa: haven't tryed the image from qi repo15:13
kyaklast time, opkg worked strange15:13
kyakalso i couldn't mount my datafs with mtd-utils15:14
kyaki didn't have a closer look15:14
rafakyak: what do all of that have to do with jlime repo ?mount problems and opkg strange)15:14
kyakjlime is mostly interesting with it's applications.. many ideas like nupdf, fbida, whatelse came from jlime15:15
rafakyak: please.. add just a Packages file text on your qi openwrt system with an index of 15000+ packages and your opkg on your qi openwrt will not work. Also.. mount problems to mount datafs is because the kernels in jlime and openwrt has different nand partitions15:16
kyakrafa: i installed mtd-utils with opkg from OE. it doesn't work15:16
wpwrakrafa: (bc) if you have only 2000 packages, you can't be picky ;-)15:16
kyakthis answers your questions "what apps might not work" >)15:16
rafakyak: it is not a problem of mtd-utils.. the problem to mount is because kernels have hardcoded the nand partitions15:16
kyakare you sure?15:16
rafakyak: yes15:16
kyaki'm not :)15:17
kyakit's just ubiattach15:17
kyakwhy would it care?15:17
rafakyak: you can not ubiattach if kernel thinks that partitions are different15:17
wpwrakrafa: correction. you very much hope you can't ... ;-)15:17
kyakrafa: i'm not sure i understand you15:18
rafakyak: and that does not answers "what apps mights not work".. because you are wanting to mount a data partitions which kernel does not know15:18
kyakthis is just your explanation about why this very application doesn't work15:18
rafakyak: mmmh?.. are you a troll?15:19
rafakyak: qi openwrt kernel has a different partition table inside than jlime kernel.. you can not mount your qi partitions on jlime because kernel think different15:19
kyaknot understandable..15:20
wpwrakkyak: how exactly do the mtd-utils fail ? maybe you and rafa are talking about different things15:20
kyaki can use it as i use "mount"15:20
kyakwpwrak: uh.. i don't remember now how exactly it failed15:21
rafaha.. funny15:21
wpwrakkyak: well, which of the mtd utilities did you try to run ?15:21
kyakwpwrak: really, don't bother about it..15:21
wpwraki guess package cross-use isn't much a priority then ...15:22
Action: kyak sleep();15:23
wpwrak(reading about debian on the list) quite scary ...15:27
wpwrak(regarding the wallpaper) the high-contrast background does indeed look awful15:28
kristianpaulwpwrak: saw the mail about NAND speed compared to uSD ?15:32
wpwrakkristianpaul: yeah. dunno. someone should benchmark them.15:34
rafawpwrak: the socrates wallpaper looks a bit suspicious as well.. the image looks like a modified version of a socrates picture.. is that CC?15:34
wpwrakkristianpaul: raw (device) and through file system.15:34
wpwrakrafa: apparently yes. wolfgang checked it.15:34
rafawpwrak: good painter then15:35
wpwrakrafa: wee, not CC. i think it's public domain. because its copyright has already expired.15:35
kristianpaulrafa: is public domain15:36
rafakristianpaul: okey.. in Argentina if you use a public domain stuff for commercial you need to pay somethin to the art national organization.. no idea the exact name in english. in spanish it is "fondo nacional de las arter"15:42
wpwrakwow. they killed public domain ? brilliant concept :)15:44
rafakristianpaul: but I guess that it should be some public domain stuff from argentina15:44
rafakristianpaul: well, it is not clear.. maybe they just want people to pay15:45
wpwrakrafa: like with any other taxes ;-)15:46
rafawpwrak: yeah.. better for them if it is not clear15:53
rafaso they can get money for anything15:54
calamarzwpwrak: re. debian speed, I still have to measure boot time and whatnot, but I don't have the feeling it's slow. maybe because I didn't try openwrt/jlime..16:25
calamarzbut in general quite happy with its responsiveness, specially since I disabled console-setup and stayed with in-kernel fonts16:26
calamarzmatchbox is a bit slow to load, but once there it behaves pretty well (using fbreder on X from time to time)16:27
kristianpaulkyak: http://paste.debian.net/100080/16:41
kristianpaulhope u still awake..16:41
wpwrakcalamarz: the swap requirements are a bit scary. and swap usually means slowness ... but well, as long as it doesn't feel slow, nobody cares how much slower it is than it could theoretically be :)16:52
qi-bot[commit] Andres Calderon: BOOKSHELF has been added http://qi-hw.com/p/xue/91c1a6917:21
wpwrakhmmm. cable assembly with antenna and cable assembly without antenna (just ending in open U.FL connector) = almost the same signal strength. i don't like that.17:38
DocScrutinizerwpwrak: X-D17:42
DocScrutinizerwpwrak: btw you noticed we ( paul fertser, me) "finished" on N900 hostmode?17:45
DocScrutinizerteaching Nokia what's possible with their hw :-P They always claimed it's not feasible17:46
DocScrutinizermentorgfx musb-core a bitch, so everything is a bit clumsy still. But basically it 'just works'17:47
wpwrakDocScrutinizer: oh, sounds cool. why did they claim it wasn't possible ?17:52
DocScrutinizerthey messed up everything you could17:54
wpwrakhmm, now that sounds vaguely familiar ;-)17:56
DocScrutinizermusb-core has a statemachine. this SM changes to hostmode/A_device on a msg from PHY via ULPI. This msg is sent by PHY when it detects ID pin is gnd. Alas our PHY isn't connected to ID pin17:56
DocScrutinizerthey also didn't bother to connect the PHY chip pin ID_IN to a GPIO or sth, they hardwired it to ID state B17:58
wpwrakso you had to override the state machine ?17:58
DocScrutinizerthey missed a capacitor for USB VBUS chargepump in PMU17:58
wpwrakso you added a cap ?17:59
DocScrutinizernope, I found the charger chip has a boost mode :-D17:59
DocScrutinizercharging reversed :-P17:59
wpwrakoh :)18:00
DocScrutinizeryou 'just' need to understand how to play your game with that chip, while fouling the Nokia closed blob BME battery management entity that likes to keep full control over the charger18:02
wpwraknice platform :)18:02
wpwrakand in the next version, if nokia is still around to make one, it will all be different :)18:03
DocScrutinizerit will be friggin meego then18:03
DocScrutinizerthough I heard they keep bme for meego :-P18:04
wpwrak"if everyone hates it, it must be good" :)18:04
DocScrutinizermost secret magic IP about how to charge a LiIon CC/CV X-P18:05
wpwraki'm sure they greatly advanced the state of the art ;-)18:05
DocScrutinizerwhile not getting the fact it's actually bq24150 charger chip that does all that18:05
DocScrutinizerthe mean part about that is they don't even publish API/ABI specs, so NFC how to manage e.g 911 calls mustn't shut down on bat low, or may proceed even on overtemp. Also no idea how to tell it does proper shutdown, overtemp throttling, and whatnot in a normal situation18:09
DocScrutinizersure we can go brute force, and shutdown hard on OT / UV18:09
DocScrutinizerbut we don't even receive a single word about where OT starts, from Nokia18:10
DocScrutinizerif they would tell us, they might get sued when we manage to kill ourselves in a LiIon explosion  X-P18:12
DocScrutinizerlawyers - MEH18:12
DocScrutinizerwell, cya18:12
wpwrakhmm, seems that one of my antennas doesn't work. the other just looks weird. good that i made a pair of them. let's see how this continues ...18:18
wpwrakDocScrutinizer: (lawyers) ah yes, the dark side of liability ... cya18:19
wpwrak(liability) if there was a device that would eliminate all car accidents caused by the driver but would fail in a way that causes an accident at 1% of the current accident rate, that device would not get deployed because the company making the device would be liable for those accidents, which would be particularly disastrous in countries aligned with US laws. better to have 100x the number of fatalities etc. but someone else who has to18:23
wpwrak pay.18:23
wpwrak(for those interested in bad antennas) http://downloads.qi-hardware.com/people/werner/tmp/ant120.png18:26
emebwpwrak: the flip side of that is when someone invents a safety device and then mfgs get sued for not deploying it.18:26
emebwhich is particularly egregious when that device is patented & bears royalties18:26
emebsee this: http://www.sawstop.com/18:26
wpwrakemeb: yeah, that variant exists as well18:26
wpwrak(antennas) refant is one of the usual 2.45 GHz WiFi antennas. ant-120B and ant-120B are the atusb/atusd antenna but 20% larger, in slightly different positions. ant-none is the same setup, with the antenna disconnected. ant-120A is another 120% antenna (identical design). still some way to go ...18:28
wpwrakemeb: solution: just outsource sawing to a place where fingers are cheap ;-)18:30
emeband safety laws non-existent...18:30
wpwrakthat usually goes hand in hand :)18:31
emebnot an issue for big business, but kind of a show-stopper for small guys & hobbyists.18:31
emebbut then we're all better off just sitting on the sofa watching TV - that's safer than having a hobby. :)18:31
wpwrakwell, as a hobbyist you can use whatever unsafe saw you choose :)18:31
emebwpwrak: not if the mfg can only sell 'safe' saws at a substantially increased price.18:32
wpwrakebay :)18:32
wpwraksynonyms ! :)18:33
emebI'll just play with computers.18:33
Action: emeb gets back OT18:33
calamarzwow... I got to have netsurf running on nn on framebuffer... awesome indeed X-D19:55
calamarztaking shot as soon as I come to terms with fbgrab :p19:56
xiangfucalamarz: I am try to create the openwrt package of netsurf.19:57
xiangfucalamarz: did you create the openwrt package already?19:57
calamarzxiangfu: no, I compiled it on debian19:57
calamarza hell bunch of libs btw19:58
xiangfucalamarz: ok19:58
xiangfucalamarz: I am working on the openwrt side :)19:58
DocScrutinizerwell, my hobbies I mustn't even tell about19:58
calamarzxiangfu: I had to turn -Werror off on a couple of places btw19:59
xiangfucalamarz: thanks for the info. I got the first error:)  Unable to find library for: BMP20:04
calamarzxiangfu: you'll have to package most of the libs on the top of that page too... can also disable some options on Makefile.config20:07
calamarzman this thing is awesome: html5, svg... and even js on a couple of releases :) how could we miss that for so long?20:09
wolfspraulcalamarz: sorry I stole the news of your file upload before your post :-)20:57
wolfspraulI didn't realize you had _just_ uploaded the file...20:58
calamarzhehe... wolfspraul: I was also thinking about wiki reader21:12
calamarzI only have to find a 12 or 16 Gb microsd... full english wikipedia dump is 6.4Gb, plus +3Gb of xapian indexes... I have to find my way thru the bisonparser for wiki markup...21:14
wolfspraulcalamarz: it's a long story :-)21:16
wolfspraulthings to consider are tables, images, special characters (unicode) including math symbols21:16
wolfspraulWikimedia Foundation provides nice XML dump files, and being able to reuse them is a big win because we are talking about gigabyes and gigabytes of constantly changing data21:17
wolfspraulthen there is zim, which first renders the xml (mediawiki markup) into html21:17
wolfspraulbut last time I checked, wmf support for zim was still mostly lip service21:17
calamarzugh... yep, I was leaving them apart for now.. I was following this simple approach here: http://users.softlab.ece.ntua.gr/~ttsiod/buildWikipediaOffline.html21:17
wolfspraulthat is there are very few zim files around, maybe it got better now21:17
wolfsprauland even then the problem is a nice zim reader that runs snappy on 32mb ram etc.21:18
wolfspraulfinally, some practical problems like how to deal with links, and how to have a really great search engine, full index of all words maybe?21:18
wolfspraulalso, I am thinking more and more that a small subset might make more sense, I'm thinking about 'math'21:18
wolfspraulso one could imagine a script that is given some category names, and then fetches all articles in those categories21:19
calamarzfull index will take too much space, I'm afraid..21:19
calamarzyep, that "sets" is promising... I think you can do it in the export features in mediawiki, right?21:19
wolfspraulthe amount of data would be much less, so making a reader that is good (fast) on all these things like images, index, etc. is easier21:19
wolfspraulah yes, good idea21:19
wolfspraulor you can just crawl through the pages21:20
wolfspraulwell it's all just ideas, lots of work and I hesitate to start a development effort that will collapse under its weight21:20
wolfspraulmust be something super maintainable that reuses as much as possible of other free efforts21:20
calamarzwhat I didn't like about the zim project was the reader of their own21:20
wolfspraulfor embedded images, there is another problem with the licenses21:21
wolfspraulthat's why wmf does not offer images included21:21
wolfspraulthe standardization of licenses in images is too weak, so there are many exceptions for this and that picture or use case21:21
wolfspraulso embedding pictures only works with a lot of manual work (fetching yourself), and at your own risk too because you may accidentally trample over some of the restrictions that are listed in the metadata21:21
wolfspraulso that's about everything I know we need to consider :-)21:22
calamarzI think something more unix-like could be much more powerful... I was thinking about doing some simple experiment with a modified micro_httpd and text only mode...21:22
wolfspraulyes of course, sounds good!21:22
wolfspraulI'm also for simple bottom-up approach21:22
calamarzso images looks hard... but having only text + latex rendering would be great21:23
wolfspraulas for zim, well sure there are things one might like or not like. but if it's an active project and they get things to work, more power to them.21:23
wolfspraulI haven't checked in a while though, like I said last time I felt a little unfortunate that I think the WMF was not really honest with them.21:23
wolfspraulthe reality is that the WMF does not support offline efforts because offline also means they (the wmf) is disconnected from potential donors21:24
wolfsprauland they don't like that :-)21:24
wolfspraulthey have quite a few hungry mouths to feed nowadays :-)21:24
wolfspraulso zim is in this state of great lip-service support, little action (from the wmf)21:25
calamarzyep... this morning I was thinking wikipedia has grown too much to be manageable grassroots-like...21:25
wolfspraulbut then I did see some zim files recently, maybe they have managed to start some server processes to crunch data21:25
wolfspraulthat would be very helpful21:26
wolfspraulat these sizes of data, we need to consider even the data itself as a given, and valuable21:26
wolfspraulbecause once you start dealing with/downloading/formatting millions of articles and dozens of languages, it becomes an effort not very dissimilar from say the Linux kernel or so :-)21:26
wolfspraulso when I see a nice large file like a dump or a zim file, and I know there are people maintaining and updating these files, I think about a way to reuse those files as is, without touching the innards of the files21:27
wolfspraulso yeah, if we can do a unix text-only approach that is still fast, why not21:28
wolfspraulmaybe you take a wmf xml dump file, and you generate some kind of index into it21:28
wolfspraulthen have a console app that allows you to lookup and extract articles from it21:28
calamarzthat was does the guy in the tutorial I pasted21:28
wolfspraulcalamarz: you think that project stands out?21:29
calamarzhe indexes with xapian inside of the bzip files21:29
wolfspraulI know there are some efforts like that, but everytime we looked closer, we found some big showstoppers.21:29
wolfspraullike huge memory use, very slow, etc.21:29
calamarzyeah, I did not try yet on the nn, but the parser in the example is using php... we should find a way of reusing some c/c++ parser to turn the xml chunks into html21:30
calamarzwas thinking about doing some quick and dirty trick like getting the article from the local url, extract xml from the bzip file, process and redirect to the parser output...21:33
wolfspraulwell any insights, experiments, progress you can make along these lines would be very appreciated21:34
calamarzone of the limitations is this indexing takes only titles (that's the +3Gb for the full dump)21:34
wolfspraulI'm just reading the page now, I am sure someone looked into this earlier but then gave up for some reason21:34
wolfspraulbut now that the overall image, system, software etc is getting better and better maybe we are ready to look back at this :-)21:35
calamarzI should get a bigger microsd in a few days... and then will spend a weekend tinkering, will share my outputs for sure :)21:35
wolfspraulspeaking about big microsds, I think nobody tried 32gb cards yet21:37
wolfspraulthey exist now, but quite expensive so last time I decided to not buy one and wait a little instead...21:37
wolfspraulsomething like 80 USD in China, I think21:37
calamarzhttp://svn.wikimedia.org/viewvc/mediawiki/trunk/parsers/flexbisonparse/ <- this is the parser I wanted to look at...21:40
wolfspra1lrafa: the socrate picture is 'free' as in the wikimedia foundation definition22:26
wolfspra1lin fact it's from their servers22:26
wolfspra1las far as I'm concerned, I am following the wmf on those matters22:26
wolfspra1lno -nd, no -nc, no fair use22:26
wolfspra1lbut everything else: yes, is 'free'22:27
wolfspra1lthere are huge amounts of free content, especially old stuff like literature, music (recording needs to be free as well), paintings etc.22:28
wolfspra1lentire photograph libraries being donated to Wikimedia Commons22:28
wolfspra1lI fully intend to use this stuff on copyleft hardware, either with good viewing software, or make it downloadable, or bundle it right away, even to only fill the storage space the device comes with anyway22:29
wolfspra1lnow, if some countries restrict the concept of 'public domain', or any other restriction or taxation of 'free', then still first of all I will follow what the WMF does22:30
wolfspra1lit's a good baseline to follow I think, a baseline upon which to build exceptions22:30
wolfspra1lOpenWrt proper boots into a home screen that greets you with a recipe for an alcoholic drink22:30
wolfspra1lI am sure if you read the law well enough, that you would have problem with this kind of thing in Saudi Arabia, UAE, and who knows where else22:31
wpwraki rather like that bit :)22:31
wolfspra1llet's not even get into the extremely differing definitions of 'pornography'22:31
wolfspra1lthe WMF has had their share of heat already over that...22:32
wolfspra1lso if you are telling me that Argentina has special rules if money is made with 'public domain artworks', well, fine22:32
wolfspra1lI wait until the case becomes real, for now I just follow wmf freedom baseline :-)22:32
wolfspra1land I cannot imagine that such law would apply to the NanoNote, it's similar to the patents I think where if you go around and read read read, worry worry worry, you will always find a reason to see a threat22:34
wolfspra1lbut in reality these things are written, meant and applied to very specific cases22:34
wolfspra1lwhere someone thinks money can easily be made :-)22:34
--- Thu Nov 18 201000:00

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