#qi-hardware IRC log for Thursday, 2010-12-02

qi-bot[commit] Xiangfu Liu: new package: moc, a console audio player http://qi-hw.com/p/openwrt-packages/d072c9100:50
qi-bot[commit] kyak: delete accidental config.full_system.rej http://qi-hw.com/p/openwrt-xburst/bba48ac02:01
xiangfukyak: thanks. I am thinking first talk with Mirko. :)02:04
kyakxiangfu: hm.. do you mean that this rej file could have been there for a reason?02:06
kyak.. i thought it accidentaly got to commit02:06
xiangfukyak: no. let him know. because the commit is from Mirko. if I modify someone's commit. I would like let him know. :)02:07
xiangfukyak: yes. sure.02:07
kyakoh, ok!02:07
kyakxiangfu: btw, do you have any problems building fbterm/lynx?02:07
xiangfukyak: no, not at all. there is daily build in build host. all fbterm/lynx compile fine.02:08
xiangfukyak: by the way. thanks for your fix about nightsky configure file .02:08
kyakxiangfu: np!02:14
kyakmirko: hey! how's your problem with fbterm/lynx?02:14
kyakmirko: xiangfu confirms that they build fine in a daily build on build host :)02:15
mirkokyak: they do not02:16
mirkonot for me02:16
mirkohttp://pastebin.com/x5Gt3CNj02:16
mirkomaybe it's a problem on my side, howeve ri do not have any idea what that could be02:17
xiangfukyak: do you know if the libmms patented ?? can we include the libmms in BNN openwrt image?02:18
mirkoxiangfu: http://msdn.microsoft.com/en-us/library/cc234711%28PROT.10%29.aspx :)02:19
kyakxiangfu: have no idea about libmmss..02:21
xiangfumirko:  hmm.. I just found the 'mpd' build with mms. what ever the "BUILD_PATENTED" set.02:21
xiangfu"Patents. Microsoft has patents that may cover your implementations of the technologies described in the Open Specifications. "02:22
xiangfumirko: I just send out one patch email to openwrt list. there are two duplicate line  about enable-aac in mpd Makefile.02:23
xiangfumirko: then I found the libmms.02:23
kyakmirko: your are building against 0.9.3202:23
kyakyou should enable locale in 0.9.3202:23
kyakfor 0.9.30.1 we have in in toolchain/uClibc/config-0.9.30.1/mipsel :02:24
kyakUCLIBC_HAS_LOCALE=y02:27
kyakUCLIBC_HAS_XLOCALE=y02:27
kyak# UCLIBC_PREGENERATED_LOCALE_DATA is not set02:27
kyak# UCLIBC_HAS_GLIBC_DIGIT_GROUPING is not set02:27
kyakit might be a little differenet (more options?) in uClibc 0.9.32.. you can find it by running menuconfig in build dir of uClibc02:28
xiangfukyak: why this "UCLIBC_HAS_LOCALE" configure make fbterm not compile. (need learning :)02:31
wolfspraulxiangfu: libmms should be OK for now02:32
kyakxiangfu: fbterm relies on several iconv functions, if i remember correctly. we had the patch for fbterm once, when we didn't have locale support.. Now that we have this local support, that patch is not necessary and it is..02:33
wolfspraulI think we should only exclude patented technologies if we positively know someone is enforcing something.02:33
kyak..dropped.. let me find it02:33
kyakxiangfu: this commit 2d7e398567856d8953b9f64576928573aae2c991 in openwrt-packages02:34
kyakxiangfu: and i'm not sure about lynx, but it is something similar.. lynx (just like any other ncurses app) is not able to display utf-8 without locale support from uClibc02:34
xiangfukyak: thanks.02:40
xiangfuwolfspraul: ok. then when openwrt fix the AAC. I will include the mpd in full_system02:41
kristianpaulmpd :)05:54
kristianpaulah, is not needed eny more to run script..  command in openwrt in order to update feeds?05:59
kyakhuh?06:00
kristianpaulhttp://en.qi-hardware.com/wiki/Building_Software_Image06:02
kristianpaulscripts/feeds install  -a ...06:03
kristianpaulis not more the the this page06:03
kyakis make package/symlinks there?06:08
kyakif it is, then it's the same as feeds update &06:08
kristianpaulit is06:09
kristianpaulah ok :)06:09
kristianpauli just wanted try moc06:09
kyakit's marked as BORKEN06:10
kyakwon't appear in menuconfig06:10
kyakso you would have to modify the Makefile a bit06:10
kristianpaulah well i must un broke it manually then06:10
kyakand see why xiangfu decided to mark it so :)06:11
kristianpaulthanks for remenber that06:11
kristianpaulwell is quite new06:11
kristianpaulhey adamwang :)06:11
adamwangkristianpaul, hihi06:12
kyakwatching moc screenshots.. not bad at all06:31
kyaki bet it can support utf-8 if built against ncursesw06:32
wpwrak_mocp is great. it's what i use after the death of xmms.06:41
kyakthere was time, i used ncmpc06:44
kyaki don't listen a lot of music now :)06:45
zeari used mocp for some time, but then i switched to mplayer :)06:51
kyakit's like you were travelling on horse and now you are travelling on elephant :)06:54
zearwhy? All i need to do is mplayer /music/collection/*ogg06:55
kyaksure, sure! i mean, mplayer is so powerfull as it seems to me :)06:55
zeari'm not a music maniac, when i feel a need to listen to, i simply launch that couple of tracks i have in one dir06:56
zearso mplayer is perfect for it06:56
zearand mocp is too bloated06:56
kyakare yo uaware of < > keys that allow you to skep to the next track? :)06:59
wpwrak_(mocp) i want the playlist. all the supposed xmms replacements with a gui are way too greedy when it comes to screen real estate07:14
xiangfuwpwrak_: (mocp) there is something wrong with moc package in openwrt-package.git. I am working on that now :)07:24
kyakxiangfu: btw, don't want to disturb you, but i checked the usbboot output yesterday.. indeed it's a continous output to terminal. But as soon as i pipe it into something like | grep SUCCESS, the ouput turns into some kind of..07:28
kyak..buffer07:28
kyakmaybe the output of usbboot needs to be fflushed in the usbboot itself07:28
kyaki mean, need to fflush() each time after printf()07:29
kyaki never noticed such behaviour in other software (think of tcpdump)07:29
xiangfukyak: thanks for the info.07:36
xiangfudo you think we should add fflush() each time after printf()? I don't much care about the delay. :)07:38
kyaki'm not even sure this will help. but yes, i would like to have a smooth indication, not that buffered-skipping output :)07:39
kyaki know that some progress bars tend to display 0% to you - then, bam, it's 50%! and then you don't really know when it will end :)07:41
xiangfukyak: ok. if I have time  I will fix that.07:47
viricFor music I use exactly mplayer like zear07:58
kyakxiangfu: thanks!07:58
xiangfukyak: also I already cleanup the output message. disable some of them.07:59
xiangfukyak: if I fix the nerase 16 4080 0 0 problem. I will release a new usbboot.08:00
xiangfukyak: sometimes the 'nerase 16 4080 0 0" erase the bootloader which is not right. but for now I can not produce that. but from the nand erase code. seems there is something wrong deal with bad block.08:01
xiangfuviric: I am using the mpc/mpd08:05
viricI still don't know those.08:09
viricIs anyone running the SIMD mplayer?08:09
xiangfuin moc package. it always using the host library path . /usr/lib ... etc.08:16
viricxiangfu: 'host', you mean, the build system?08:33
viricsome mix host/guest, build/host, and all that08:33
kyakxiangfu: i hope cleaning up the output won't break the reflash_ben.sh :) the progress bar depends on specific messages output  by usbboot08:37
kyakxiangfu: weird bug erasing the bootloader.. never happened to me08:38
viricwhat bug?08:38
kyaki, however, usually reflash all three uboot/kernel/rootfs08:38
kyakviric: see the message from xiangfu above08:39
viricAh, the range 16 4080... I reported the problem.08:39
viricVery annoying08:39
viricxiangfu: you cannot reproduce it? I can try again back at home...08:40
kyakviric: why are you erasing the whole NAND?08:40
kyakis 4080 the whole NAND?08:40
viricyes08:40
kyak16 is the end of uboot right?08:40
viricWell, all but the bootloader and the kernel08:40
kyakah, ok08:40
viric16 is the start of the rootfs08:40
kyakwhat is you do 16 4060?08:42
kyaki.e. leave some space a little bit (for bad blocks?)08:42
xiangfuviric: I cann't reproduce that. but from the source code. : http://projects.qi-hardware.com/index.php/p/xburst-tools/source/tree/master/usbboot/xburst_stage2/nandflash_4740.c#L33908:42
kyak*what if you do08:42
xiangfuviric: when there is a bad block. it's will skip the block, but not run i++, which mean :08:42
viricI see08:43
kyakseems that you need bad blocks to reproduce that :)08:43
xiangfu1. if you want erase 10 blocks.  2 there is one bad block in the 10-blocks. 3. it will erase 11 blocks.08:43
viricah...08:43
viricIn fact, the command said something like it overrun08:44
wolfspraulalmost every 2 GB nand chip should have a few bad blocks08:44
viricwould it overrun, if it asks for somethign beyond the last block?08:45
kyakwolfspraul: is it tested at production line?08:45
xiangfukyak: hmm..  we just need know, if we pass the >4096 blocks address, what the nand chip do.08:45
kyakxiangfu: right08:45
kyakit would bite it's tail08:45
kyaki.e. start from block 008:46
kyaki.e. reflash the bootloader :)08:46
xiangfuwe need to make sure.08:46
viricSo08:46
viricI tried more than once the nerase 16 408008:46
viricAnd it always erased the bootloader08:46
viricI can't remember if it also erased the kernel08:46
kyakviric: you can detect the number of bad blocks gradually decreasing this 4080 :)08:47
xiangfuviric: can you try this stage2: http://downloads.qi-hardware.com/people/xiangfu/tmp/xburst_stage2.bin08:52
xiangfuviric: by : sudo usbboot -2 ./xburst_stage2.bin -c "boot;nerase 16 4080 0 0"08:52
xiangfuviric: 1. reflash the bootloader to your nanonote. 2. sudo usbboot -2 ./xburst_stage2.bin -c "boot;nerase 16 4080 0 0"08:53
virichm08:53
Last message repeated 1 time(s).08:53
viricI'll try.08:53
xiangfuviric: what is the last message when you do 'nerase 16 4080 0 0'08:53
viricFor me it's easier to build the stage2. That isn't commited anywhere08:53
viric?08:53
xiangfuviric: here is "Force erase, no bad block infomation!"08:53
xiangfuviric: not yet commit.08:54
viricok08:54
viricwhy would I need to reflash the bootloader? Does it matter? I have a good bootloader already :)08:55
viric(can you remind me the letter in the mailing list? :)08:55
xiangfuviric: you just said the nerase your bootloader, so I just guess there is no bootloader in your nanonote. sorry.08:55
xiangfuviric: yes. sure.08:55
viricxiangfu: ah, when I noticed that the bootloadear was erased, I flashed  it anew.08:56
xiangfuviric: this stage2 only delete the line 339, http://projects.qi-hardware.com/index.php/p/xburst-tools/source/tree/master/usbboot/xburst_stage2/nandflash_4740.c#L33908:57
viriccur += ppb; ?09:00
xiangfuyes09:00
viricok09:00
viricI'll try only once at home09:01
viric(in some hours)09:01
kyakviric: right, never accept binary files from strangers ;)09:03
viricnever :)09:05
xiangfukyak: viric here is the compile log of moc. it started. then give a  Segmentation fault09:05
virichmm I did not have any interest in running ;)09:06
viricdo you ask for help getting it to work?09:06
viricin running *it*09:06
kyakstrace might give some hints09:06
viricfor a segfault? :) that would be good chance09:06
viricgdb helps more09:06
kyakeverything is stripped in openwrt09:07
kyakwould require a complete rebuild to make use of gdb09:07
kyakstrace gave me useful hints couple of times09:07
viricdebug info in mpc only could be enough09:07
kyakdebug info in a single package didn't help me when i tryed debuggin stardict :)09:08
xiangfutoo bad, I just run the "nerase 16 4080 0 0", don't have rootfs any more. :(09:08
kyakxiangfu: btw, should we perhaps enable debug and gdb in config.minimal? or maybe create a config.debug?09:10
wolfspraulkyak: we should document how to zoom in on particular bugs with gdb09:10
viricxiangfu: that is what it should do. Erase the rootfs. :)09:10
wolfspraulyes! :-) that's just what I wanted to write09:10
kyakmight be usefull in such cases09:10
wolfspraulconfig.debug or so09:10
viricyou are crazy. You did not have means to debug?: )09:10
kyakwe do. don't you see, we already agreed to have config.debug? :)09:11
virichehe09:11
kyakit's easier than you think09:12
viricok09:12
kyakjust enabling couple of options in menuconfig09:12
viricyou use uclibc, right?09:12
kyakwe do, however, this is adjustable09:12
viricnice09:13
kyakare you basing on uclibc, too?09:14
viricnow I'm using glibc09:14
viricI can try to switch...09:14
kyakare you suffering of spped decrease or something? is it even noticeable?09:14
viricI'm not comparing. I don't know. :)09:15
kyakoh ok09:15
viricI don't notice anything special.09:15
kyakit would be initersting to build against glibc09:15
kyaki remember someone developed simple benchmark in lua for Ben09:15
viricif you have any benchmark I should try... I can09:16
kyakmm09:16
kyaksomewhere in the mailing list09:16
kyakwould be very intetsrting to compare indeed09:16
kyak(and with jlime, too)09:16
viriclet's start a contest :)09:17
kyak...and start by finding this mail :)09:18
xiangfukyak: see the line 109 : http://pastebin.ca/200860209:18
xiangfuit try to using HOST lib path. I  just don't know how to set it to $(STAGE_DIR)/lib ...09:19
xiangfuI try to export the "LD_LIBRARY_PATH" before configure. but it's still output like "/lib/x86_64-linux-gnu  ..." not openwrt path.09:20
kyakxiangfu: hm... need to have a look at configure script09:21
kyakxiangfu: i'll try it later and let you know09:21
wolfspraulkyak: can you commit an initial config.debug, same concept as config.minimal, just with the debugging essentials included as well09:24
kyakwolfspraul: sure, will do it09:25
kyak(when i have it ;)09:26
kyakcan't seem to find that benchmark...09:28
viricwas it about memset or memcpy?09:32
kyaki see some messages about that,too.. not sure09:34
kyakit was something different09:34
viricI did not search. I simply have something in mind about that.09:35
viricAs if I had read it.09:35
virickyak: can you build 'binutils' for your nanonote in openwrt, with uclibc? It's not trivial to link (it wants libutil from glibc).09:35
kyakyes, it's ported09:37
viricgreat. thank you .)09:38
viric:)09:38
kyakxiangfu: seems that the default library search path for moc can be overridden by lt_cv_sys_dlsearch_path09:40
kyaklike CONFIGURE_VARS+=lt_cv_sys_dlsearch_path=$(STAGING_DIR)/usr/lib09:41
xiangfukyak: can we modify the "configure.in" ?09:42
kyaki think so.. i'd prefer the CONFIGURE_VARS way (no need for patch)09:43
xiangfukyak: re-compiling .09:45
kyakviric: can't find it -\ i guess it was my imagination09:52
kyakxiangfu: have to go now. good luck to you :)09:52
xiangfukyak, same error.09:53
xiangfukyak: http://pastebin.ca/200862509:54
qi-bot[commit] Werner Almesberger: usrp/fscan: results are now quite stable, so 20 instead of 100 runs are enough http://qi-hw.com/p/ben-wpan/60d78fd10:25
qi-bot[commit] Werner Almesberger: usrp: adjust outlier rejection and signal range for improved antennas http://qi-hw.com/p/ben-wpan/c2b489d10:25
qi-bot[commit] Werner Almesberger: atusb: rearrange I/Os for new layout http://qi-hw.com/p/ben-wpan/36ea2be10:25
qi-bot[commit] Werner Almesberger: atusb.brd: new layout (in progress) http://qi-hw.com/p/ben-wpan/ecaab6010:25
bartbesso what's ash equivalent of .bashrc?11:50
viricthat reminds me of...11:51
bartbesnvm found it11:51
bartbes.profile11:51
viricSomehow, 'dropbear' called shell (bash) does not source /etc/bashrc when running commands like "ssh root@nano ls"11:51
viricbartbes: 'profile' is for interactive shells. .bashrc is for non-interactive shells.11:51
bartbes.bashrc is for bash11:51
bartbes:P11:51
viricno11:52
viric.bashrc is the rc script for bash for non-interactive shells11:52
viric.bash_profile is the rc script for bash for interactive shells11:52
viricwhich have /etc mates: /etc/bashrc and /etc/profile11:52
viric(usually)11:52
bartbes.profile != .bash_profile11:53
bartbeshmm anyway, I just need an env var set11:54
bartbesit didn't work btw11:55
viric:D11:55
viricwhy you run ash?11:56
bartbesbecause it's the default?11:56
viricI hate ash, because it confuses me making me think it's bash. :)11:56
bartbesit's the one is busybox11:56
viriccan't you put bash?11:56
bartbeshmm but that one isn't in busybox11:56
bartbesand wolfspra1l said that ash is *very* resource efficient11:56
bartbesanyway, I guess ash isn't executed as login shell, and that's why it ignores .profile11:57
bartbesit is..11:57
viricbash is not efficient at all11:57
viricI have to choose whether to sacrifice my time learning ash, or sacrifice cpu cycles due to bash inefficiency :)11:58
bartbes/etc/profile is where it's at ;)12:00
bartbeswhat would you need to learn about ash?12:00
viricit does not like some 'bashisms' I write12:00
viricwpwrak_: may it be that the SD works faster than the nand then?16:32
virickyak: thought of any benchmark?16:32
viricgrr16:54
viricI'm trying to build a *single applet* of busybox16:54
viricAnd I don't know how to parse its '.config' for it to do only the applet I want16:55
viricah.. 'allnoconfig' may help16:58
kristianpaulwardriving :)17:14
kristianpaulnew word to my dictionary17:14
wpwrak_viric: (sd > nand) i remember than someone said so a while ago17:46
wpwrak_viric: dunno how much of this is error correction and such, though. the "naked" NAND ought to be faster, given that its bus is twice as wide and it's an internal bus.17:48
qi-bot[commit] Werner Almesberger: atusb.brd: more layout work (on-going) http://qi-hw.com/p/ben-wpan/20c051719:05
qi-bot[commit] Werner Almesberger: atusb: added bottom ground zones (on-going) http://qi-hw.com/p/ben-wpan/8a3a42819:05
qi-bot[commit] Werner Almesberger: atusb/atusb.brd: added ground zones on front layer as well http://qi-hw.com/p/ben-wpan/c7d485a19:05
qi-bot[commit] Werner Almesberger: atusb: move traces to coincide with thermal relief http://qi-hw.com/p/ben-wpan/a641eb721:31
qi-bot[commit] Werner Almesberger: atusb: bumped version number to 20101202 http://qi-hw.com/p/ben-wpan/fb2694121:31
qi-bot[commit] Werner Almesberger: atusb/cam: updated CAM process for new board http://qi-hw.com/p/ben-wpan/876e12b21:31
qi-bot[commit] Werner Almesberger: atusb: add automatic board revision detection; clean up feed line http://qi-hw.com/p/ben-wpan/b8877e022:06
roh   ~.22:27
qi-bot[commit] Xiangfu Liu: add -Vall to alsamixer, for display MIC stuff http://qi-hw.com/p/openwrt-xburst/df2cf6722:36
--- Fri Dec 3 201000:00

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