**** BEGIN LOGGING AT Thu May 03 03:00:02 2012 May 03 03:13:10 ccache gave me headache, hwo to disable it in angstrom? May 03 03:13:34 ./oebb bitbake virtual/kernel -c compile -f or whatever it will still give me the same kernel modules May 03 04:50:10 hey prpplague, you got any ftdi debug tips? May 03 05:13:44 <_av500_> gm May 03 05:16:02 <_sundar_> gm May 03 05:17:52 gm May 03 05:39:25 howdy jkridner__ May 03 06:50:56 <_tasslehoff_> koen: is this using the "1.7 based DDK" https://lists.yoctoproject.org/pipermail/meta-ti/2012-May/001024.html? I'm hoping that's what the IMGPV in libgles-omap3_4.06.00.01.bb means :) May 03 07:23:06 _tasslehoff_: it is May 03 07:23:33 _tasslehoff_: I was very happy that denix did all the work on that :) May 03 07:29:57 <_tasslehoff_> :) May 03 08:22:14 hi, i have a devkit8000 (said to be a clone of BB). I am trying to run dvsdk filesystem on it. what kernel shall i need? May 03 08:22:35 dvsdk? May 03 08:22:43 and devkit8000 is not a clone May 03 08:22:55 it is similar, but not an exact copy May 03 08:25:26 Hello All. Just trying to install a gui on 12.04. It seems xubuntu-gdm-theme doesn't exist for 12.04 (package not founf error). Any other options? May 03 08:27:21 I am trying to run Root File System from DVSDK (dvsdk_omap3530-evm_4_01_00_09) for OMAP3530 on Devkit8000. May 03 08:27:21 I found that DVSDK (dvsdk_omap3530-evm_4_01_00_09) for OMAP3530 uses linux kernel 2.6.32. I am unable to find a linux kernel patch for devkit8000. May 03 08:27:21 Where can I find a linux kernel patch for 2.6.32 supportin May 03 08:27:37 devkit8000? May 03 08:27:42 I dont know May 03 08:27:52 why dont you ask the people that sell the devkit80000000? May 03 08:28:19 also, this is not the place to get support for DVSDK May 03 08:33:57 DVSDK = Digital Video Software Development Kit May 03 08:34:08 theres no channel on devkit8000 or dvsdk May 03 08:34:48 mak1: correct May 03 08:34:50 may be someone from BB might have used it May 03 08:35:07 mak1: from experience, not many May 03 11:37:15 where is the more in depth hw spec for beaglebone? cant find it :/ May 03 11:38:15 http://beagleboard.org/hardware/design May 03 11:39:15 is there a obius link to that url somwhere? cuz i culdnt find it :( May 03 11:40:20 yes May 03 11:40:24 http://beagleboard.org May 03 11:44:25 oh.. May 03 11:44:47 I know, a most obscure link.... May 03 11:44:57 xD May 03 11:45:16 what the heck is PMIC TPS65217? May 03 11:45:39 its a PMIC May 03 11:45:40 4.0 in the system reference May 03 11:46:08 another obscure link: https://www.google.com/search?q=TPS65217 May 03 11:46:47 yeah i know, i realy need to ask more clarly :P May 03 11:47:44 why is there a PMIC on the board? as i can tell i need externals to achive the use of such stuff right? May 03 11:48:13 the pmic is there because the omap requires it May 03 11:48:54 mru: in what sens does the omap need it? May 03 11:48:58 it needs power May 03 11:49:08 electrical preferably May 03 11:49:13 nooo, i was told it rund on pure willpower! May 03 11:49:23 runs* May 03 11:49:27 it has internal willpower generator May 03 11:49:37 low noise May 03 11:49:49 * av500 uses a cattle prod to achieve the same May 03 11:50:07 or a tazer.... May 03 11:50:10 a cattle prod is the antidote for willpower May 03 11:50:57 :) May 03 11:50:59 isbric: to be honest, the PMIC is there because it's a TI chip and TI likes to sell chips May 03 11:51:14 look carefully, its not connected at all May 03 11:51:39 just like that extra crystal May 03 11:51:59 a soulcrystal on the bone? May 03 11:52:18 man i got lucky to fetch me one of these :P May 03 11:53:05 but honestly, the datasheet is a little bit confusing (probobly got confusing in the process of making it noob friendly imao) May 03 11:53:18 data sheet? May 03 11:53:40 under TPS65217 is says "LiION Single cell battery charger (via expansion*) " May 03 11:53:46 "system reference" May 03 11:54:20 yes, and? May 03 11:54:28 why even mention that if the TPS65217 isnt ment for anything more then regulationg the onboard power. May 03 11:54:56 who said that? May 03 11:55:16 TPS65217 - Plasma antigravaton generator (via expansion*) May 03 11:55:27 and the TPS65217 has a battery charger inside May 03 11:56:12 if it confuses you, don't use it May 03 11:56:12 am i readning the eaglefiles wrong then cus to me it looks like i wount be able to use it on the current design of Rev A5? May 03 11:57:20 ah screw it. May 03 11:57:52 howdy mranostay May 03 12:08:42 isbric, what exactly do you want to do with the TPS65217? May 03 12:13:10 bradfa: screw it May 03 12:13:19 av500, roger! May 03 12:13:27 av500: clockwise or anti-clockwise? May 03 12:13:33 not _roger_ May 03 12:13:48 av500: not screw _roger_? May 03 12:14:15 is it not counterclockwise? May 03 12:14:58 http://oxforddictionaries.com/definition/anticlockwise May 03 12:41:25 xxiao, try bitbake -f -c clean virtual/kernel; bitbake -f -c menuconfig virtual/kernel; bitbake -f -c compile virtual/kernel; bitbake -f -c deploy virtual/kernel May 03 12:49:28 How can I access the cpu temperature on beaglebone is it the same as on a desktop machine? May 03 12:54:39 there's no sensor May 03 12:55:11 if there is, there's no driver :) May 03 13:01:23 jsabeaudry, dunno but maybe check am335x tech ref section 9.3.1.10, bits 8 to 15 are an ADC temperature but don't see any other reference to it May 03 13:01:48 maybe you can read the register directly? May 03 13:03:28 Many MSP430 have internal temp sensor on the ADC, maybe similar? May 03 13:08:31 koen: what happened to the omap3430 temp sensor driver? May 03 13:09:04 * av500 senses a great disturance in the temperature May 03 13:09:05 bradfa, ah great, I just hope the issue I'm having is a termperature issue May 03 13:09:51 jsabeaudry, I couldn't find any table saying what the ADC values correspond to, but I didn't look very hard May 03 13:10:11 jsabeaudry, can you reproduce your issue on more than one bone? May 03 13:10:34 bradfa, yup reproduced on 3 rev A3 bones so far May 03 13:10:38 mru: is was broken in early 34xx, 36xx tends to deny it's there May 03 13:11:09 bradfa, 3 out of 3 so every bone that I tested has the issue May 03 13:11:22 oh, wow May 03 13:11:38 koen: well, I got sane-looking numbers from it... May 03 13:13:19 mru: the non working parts (es 2.x and earlier iirc) will give temperatures in the 80 degrees C range May 03 13:13:26 jsabeaudry, can you share your kernel config? May 03 13:13:29 the working parts give sane numbers May 03 13:13:34 koen, 80 C is nice weather, no? :) May 03 13:13:42 * bradfa from backwards USA May 03 13:13:45 bradfa: in a sauna, sure May 03 13:13:55 mru, hellofa sauna! May 03 13:14:08 koen: I guess I tried it on a working part then May 03 13:14:20 bradfa: 80C is rather cool for a sauna May 03 13:14:34 * bradfa doesn't visit sauna often May 03 13:14:40 * mru neither May 03 13:15:01 jsabeaudry: after reading the datasheet, there must be a sensors, since it implements smart reflex class 2B May 03 13:15:17 "voltage scaling based on die temperature" May 03 13:15:23 so scan the address space May 03 13:16:27 * koen downloads spruh73d May 03 13:16:53 bradfa, http://pastebin.ca/2143429 May 03 13:17:25 right now I have preempt enable but the issue is also reproducible on non-preempt May 03 13:17:54 who's kernel tree? May 03 13:18:43 weird thing happens with pastebin.ca, Firefox can't get to it but wget can... odd May 03 13:18:50 koen's r10d May 03 13:18:52 jsabeaudry: the layout of that sensor looks a lot like omap3, try porting that driver May 03 13:21:59 jsabeaudry, um, I can't build that config on koen's r10d tree... May 03 13:22:21 bradfa, gives you an error? May 03 13:22:23 * bradfa tries again May 03 13:22:28 yes, lots May 03 13:22:56 ah, thumb2... May 03 13:23:00 huh, cant you post diff with config that works? May 03 13:23:07 can* May 03 13:23:12 one sec May 03 13:23:42 bradfa: don't use a broken toolchain :) May 03 13:23:50 koen, :P May 03 13:24:09 jsabeaudry, non PREEMPT kernel on koen's r10d tree I boot: https://raw.github.com/gist/2580060/be5bce61504cfae372425cabb4e76780a5a12d8a/.config May 03 13:24:44 koen, I know you love thumb mode, but I've never been fond of it, I've always had issues building things May 03 13:25:00 I don't like thumb May 03 13:25:03 I do like thumb2 May 03 13:25:11 k, yeah, that May 03 13:25:36 * av500 likes 2 thumbs May 03 13:25:36 you're using the village idiot of compiler versions, so you don't get to complain :) May 03 13:25:49 koen, Can I be the leader of my village? May 03 13:25:50 gcc 4.4 only serves one purpose: make gcc 4.5 look awesome May 03 13:25:50 :) May 03 13:25:56 I do like GCC 4.5 May 03 13:26:17 compilers are overrated May 03 13:26:26 This newfangled 4.7 stuff I still have to check out May 03 13:26:45 I like reading about clang and llvm, never tried though May 03 13:27:56 bradfa, hmmm the configs are completely different, perhaps there is something wrong with the one I use May 03 13:28:27 I didn't compare them. Mine is am335x_evm_defconfig with a few small tweaks May 03 13:28:44 like no freq scaling, no usb gadgets, and enabling of devtmpfs May 03 13:30:04 I installed a big fan see if i can still repro May 03 13:30:27 http://www.bigassfans.com/ May 03 13:30:31 work safe May 03 13:31:03 if the fan doesnt do the trick I'll look into the config thing May 03 13:33:51 Arago kernel repo got patch today for cpsw related things, might be useful? http://arago-project.org/git/projects/?p=linux-am33x.git;a=commit;h=0d0567b6c1a61a374120b985559768077c511a6d May 03 13:34:10 is it possible to bitbake virtual/kernel (after some local hack) without change the PR value? May 03 13:35:10 only after i modified PR then a new kernel was built, otherwise it will fetch from sstate-cache no matter what i tried May 03 13:36:17 bradfa: it looks like that one needs the 2 patches I posted yesterday May 03 13:36:35 koen, and I take it those aren't in Arago yet? May 03 13:37:07 right May 03 13:37:18 wtf May 03 13:37:51 I wish to congrat the people in charge of the hardfloat compilation of the BeagleBoard Ubuntu 12.04 hardfloat version. After 2 years I think it's the first version that I feel confortable with!!! It flies!!! I think this is a very remarcable version!!! May 03 13:38:16 heh May 03 13:38:39 you do know that going from softfp to hardfloat give pretty much no speedup, right? May 03 13:38:49 koen: is it possible to bitbake -f a kernel without modifying PR, thanks May 03 13:39:22 xxiao: yes, but only if you are stupid May 03 13:39:30 if you make a change, bump PR May 03 13:39:35 it's as simple as that May 03 13:40:06 i used to short-cut it, go to tmp/.../git and build, now want to use bitbake for the whole deploy, bumping PR took a long while to build May 03 13:40:13 I supose not in the kernel, but in programs like Codeblocks that I use to program it is three times faster May 03 13:40:16 koen: armv4 soft to armv7 hard does May 03 13:40:18 thanks, just want to make sure PR is the only right way May 03 13:40:28 as far as bitbake is concerned May 03 13:43:26 i do wonder why '-f' can not do a force-rebuild, if PR is unchanged it appears '-f' has no effects on local code changes, bitbake just take them from sstate-cache based on PR? May 03 13:43:34 what's the purpose of -f then May 03 13:44:31 for virtual/kernel i would expect -f can trigger a clean of sstate-cache, or at -c clean time, May 03 13:44:47 sstate works with checksums May 03 13:45:04 so if you add a patch to the SRC_URI, it will rebuild automatically May 03 13:45:24 but -c compile -f works here, though May 03 13:46:30 i updated defconfig(to add a few modules), then a -c clean; -c compile -f never build the modules(vmlinux was built though) May 03 13:46:57 modules are a seperate step May 03 13:47:00 bitbake virtual/kernel does copy the packages/lib/modules/* from sstate, but the new defconfig has no effect May 03 13:47:38 it appears modules build are included in virtual/kernel if i gave no other args to bitbake(e.g. -c) May 03 13:47:48 how to force a module build? thanks May 03 13:48:10 -c listtasks May 03 13:49:07 how can i add a package without using opkg install but using angstrong bitbake? May 03 13:49:44 switch angstrom rpm and use yum May 03 13:49:56 you have to use opkg one way or another to install an ipk May 03 13:50:14 koen: i mean when i create kernel and rootfs May 03 13:50:24 koen: how can i add a package in a rootfs? May 03 13:50:43 cp image.bb my-image.bb ; emacs my-image.bb ; bitbake my-image.bb May 03 13:51:11 mmm May 03 13:51:18 no i need some doc May 03 13:51:54 what i'm doing is MACHINE=beagleboard ./oebb.sh bitbake virtual/kernel May 03 13:52:21 and MACHINE=your_machine ./oebb.sh bitbake systemd-gnome-image May 03 13:52:28 and this create kernel and rootfs May 03 13:52:45 now i can run iamge on bb and using opkg to install some package May 03 13:52:49 cp systemd-gnome-image.bb my-image.bb ; emacs my-image.bb ; bitbake my-image.bb May 03 13:54:22 i don't understand the command that are you using May 03 13:54:40 cp or emacs? May 03 13:55:09 the *.bb May 03 13:55:22 its not a command May 03 13:55:24 what are you doing? copy some file for what? May 03 13:55:24 its a file May 03 13:57:41 av500: i'm using http://www.angstrom-distribution.org/building-angstrom step... May 03 13:58:21 [15:49:08] how can i add a package without using opkg install but using angstrong bitbake? May 03 13:58:27 you asked and got an answer May 03 13:58:43 ok, but i want understand the passage... May 03 13:58:50 sure May 03 13:59:03 look at the files involved May 03 13:59:07 look inside them too May 03 13:59:42 there is more folder...one generare is build that have tmp-angrstomv2012... May 03 13:59:59 and its contains kernel image uboot rootfs.bz2 ecc.. May 03 14:00:39 and there is conf script source git...that i think is the 'bitbake file to generate the tmp-angstrom file May 03 14:02:14 so i have to find where i have image.bb? May 03 14:02:17 koen: compile_kernelmodules got all modules built under git, it's not put into packages/ and not in deploy, -c deploy did not do that either, anything i'm missing? May 03 14:02:59 -c package_write does the package step May 03 14:03:04 and -c deploy the deploy step May 03 14:03:51 thanks! May 03 14:08:05 hmm virtual/kernel -c package_write or deploy still only affects the kernel, not the modules after compile_kernelmodules May 03 14:12:38 http://patches.openembedded.org/patch/16821/ related topic here May 03 14:29:08 openembedded is not green...a simple module addition caused massive re-baking over long time, a waste of electricity, either by PR change or -f May 03 14:29:38 intel is investing in it May 03 14:29:45 that should have given you a clue :) May 03 14:30:48 * xxiao is waiting for 1000 beagleboards cook the recipes with lower power May 03 14:31:23 And 1000 wall-warts to power them May 03 14:34:09 i do hope bake virtual/kernel contains in-kernel modules automatically though, the adding of module is more involving than i expected, bump PR is just too slow for this May 03 14:35:20 * av500 edits a linux.config file and recompiles to "add a module" May 03 14:35:50 av500: yeah i normally compile the kernel directly, bypassing bitbake, May 03 14:35:51 well, and edits another file to get that module copied over to the image May 03 14:36:06 so many files May 03 14:36:13 just started to try the bitbake route since last night, what a journey May 03 14:36:26 I can turn it into a dbus call to make it more unix like May 03 14:36:46 make it a systemd job ! May 03 14:36:52 koen: using a python tool? May 03 14:36:56 the final result: -c clean; -c cleansstate; bitbake virtualkernel; this will update modules and uImage etc, take a little less time than bumping PR May 03 14:37:17 Well well, now approaching close to 3x to my usual maximal run time with a big fan blowing on the am335x May 03 14:38:31 i spent 5 hours to save 10 minutes, back to the old way, compile kernel modules etc without bitbake May 03 14:39:22 xxiao, join the club of people who couldnt tame the oe/bb beast May 03 14:45:47 Ah, the temperature sensor is off and thermal shutdown are off by default? May 03 14:46:12 so thermal meltdown is on by default? May 03 14:46:37 am335x has a built in thermal sensor? May 03 14:47:01 Mojito, checkout 9.3.1.10 in the trm May 03 14:47:13 there is an adc reading some temperature May 03 14:47:36 Hmm, is there a driver that provides access to that? May 03 14:48:23 Apparently not, but I can't even turn on the sensor using devmem2 May 03 14:48:34 I have a 'general system health' report in my app that I would like to add that to, if I could May 03 14:49:54 I made note of the trm section number for future reference, and will wait for some API to appear May 03 14:56:58 if you want to display a logo @beagle startup apart from what is already there... what do I need to change? May 03 14:59:28 https://www.google.com/search?q=linux+kernel+splash May 03 15:00:13 great May 03 15:02:17 av500: what about the orange screen that come up right when the power is switched on with an XM? May 03 15:02:35 thats in uboot May 03 15:02:57 but it could be just dss default color to orange May 03 15:03:02 not some bitmap May 03 15:03:12 but there were some patches for splash in uboot May 03 15:03:25 theres also an omap3 dss driver in uboot git afaik May 03 15:03:52 http://android.serverbox.ch/?p=70 May 03 15:03:56 will try it out, thx May 03 15:05:08 Sweet jesus, I resurrected a zombie beaglebone! May 03 15:14:51 Question of the day, are the PMIC voltages adjusted properly when running cpufreq-set -f 720000 May 03 15:18:23 av500: i also get on screen something like beagleboard-evm login: _ flashing_ is there a way to get rid of that aswell? May 03 15:18:59 thats probably framebuffer console May 03 15:19:03 disable it in the kernel May 03 15:19:07 fbcon something May 03 15:19:22 hmmm May 03 15:19:41 its pretty useless anyway if you have serial console May 03 15:20:02 true May 03 15:20:34 it's only to impress people with movie style "scrolling source code on terminal" May 03 15:21:00 heh May 03 15:35:52 Well, looks like the beaglebone not only overheats at 720mhz but also at 500 May 03 15:36:47 unsafe at any speed? May 03 15:37:05 keep it on the track May 03 15:37:20 av500, maybe not under 50% cpu usage, all my tests are done over 50% May 03 15:39:17 thinking of it, at under 10% cpu usage it stays alive fine for 16+ hours May 03 15:56:24 Looks like temperature is not the whole answer for the gp_timer crashes May 03 15:59:13 anyone aware of some light-weight web-based top/vmstat monitor program? i.e. draw nice graph on cpu usage, not the snmp/big-guns May 03 16:02:31 rrdtool, but it will require a bit of glue May 03 16:04:16 there are ready-made packages based on rrdtool for various data sources May 03 16:05:17 jay6981: i'm thinking about grab the top/vmstat/etc and feed it to a js graph module and use busybox http to present it May 03 16:06:13 or feed the data to rrdtool, have rrdtool make the PNG and serve it with your httpd May 03 16:06:44 want to have live realtime display... May 03 16:07:59 jay6981: feed realtime data to something like this: http://square.github.com/cubism/ May 03 16:09:15 yeah, looks neat. rrdtool might be better for long term stats gathering/visualization May 03 16:11:24 Wow, just found out that the "ls" command does not work in zombie state May 03 16:12:00 zombie state? May 03 16:12:29 thats northern of texas i think May 03 16:12:37 and a bit to the west May 03 16:12:45 utah? May 03 16:12:55 dunno, thats what a texan told me once May 03 16:13:19 mru, after prolonged use of the beaglebone, the led stop flashing but i'm still connected by ssh, a lot of the stuff dont work anymore but some stuff still works, it's not clear how to reach this state May 03 16:13:31 the led = the heartbeat led May 03 16:13:49 sounds like your kernel is half-dead May 03 16:13:58 not a state I recommend staying in May 03 16:14:02 mru, yup exactly May 03 16:14:13 but no oops or anything on the serial console May 03 16:14:36 well I'd love to diagnose the problem, so I hope to learn something in that state May 03 16:15:02 and your gp timers aren't firing? May 03 16:15:22 i May 03 16:15:30 j May 03 16:15:31 jay6981, nope, cat /proc/interrupts always shoes the same number for gp_timer May 03 16:15:40 issues related? May 03 16:16:27 often the dma stops working properly, sometimes it also kills the ethernet May 03 16:17:00 jsabeaudry: on a single bone? or several? May 03 16:17:04 many commands do nothing (have to Ctrl-C out of them) : top, htop, ls, g++, etc... May 03 16:17:16 av500, 3 out of 3 that I have tested May 03 16:17:27 does the clock still increment? May 03 16:17:34 jay6981, how do I check? May 03 16:17:43 type date? May 03 16:18:00 date doesnt work: like top, htop, g++ May 03 16:18:10 if I cat /proc/uptime the first number goes up May 03 16:18:15 but not the second May 03 16:18:27 could be asking a stupid question now, but is there a way to set a lot of output pins to a logical at the same time ? I mean in for example an arduino you can treat a port as a byte value and by writing value 255 to it you set all pins high. Is something similar possible on beaglebone or do I really have to set the 8 pins high 1 after the other (which is still extremely fast of course) ? May 03 16:18:38 sure May 03 16:19:04 gpio pins are hooked io registers May 03 16:19:08 hooked to* May 03 16:19:26 I know how to set 1 pin high May 03 16:19:49 but setting 8 of them is 8 times setting 1 high or also achieveable in 1 command? May 03 16:20:13 sure, write the value to the register May 03 16:20:28 (assuming kernel) May 03 16:20:28 you can write 8 bytes at once with a single machine instruction May 03 16:20:38 probably more with simd May 03 16:20:48 1 register is 32bits? May 03 16:20:52 ah the dd command still works but not hexdump May 03 16:21:19 32bit is the native word size so most ops are done that way May 03 16:21:37 jay6981: I guess he is using sysfs May 03 16:21:47 yeah May 03 16:22:06 where each gpio is treated separate May 03 16:22:11 I use the virtual value file for for example gpio38 May 03 16:22:17 yes indeed May 03 16:22:26 educa: so from user space the answer is no May 03 16:22:40 unless you patch the sysfs stuff to allow that May 03 16:23:05 ok, but with direct register access it would be possible then if I understand right? May 03 16:23:12 in your case, a simple I2C port expander might be simpler May 03 16:23:28 educa: yes May 03 16:23:48 and is accessing a pin through sysfs slower then direct register access? May 03 16:23:58 slower May 03 16:24:01 yes May 03 16:24:09 you have to go through many software layers to affect the change to the register May 03 16:24:21 hmmmm May 03 16:24:36 one price you pay by running a general purpose OS May 03 16:24:38 then I guess I learned something, but I learned the slow version May 03 16:25:04 can I from gcc directly call this register or is linux going to protect me from doing that ? May 03 16:25:28 it'll stop you May 03 16:25:43 ok so unless you hack kernel its a no go May 03 16:25:53 thats probably why ncbas does it all from PRUSS May 03 16:25:53 you could go into uboot and poke the registers yourself just to see it work May 03 16:26:24 there's also that linux uio stuff that might be adaptable May 03 16:30:14 educa: but yes, you will have to delve into kernel May 03 16:39:15 * jkridner__ is on the search for the best way to poll for GPIO interrupts from userspace. May 03 16:39:59 poll? May 03 16:40:02 http://pod.tst.eu/http://cvs.schmorp.de/libev/ev.pod has some glorious words on using epoll: Epoll is truly the train wreck among event poll mechanisms, a frankenpoll, cobbled together in a hurry, no thought to design or interaction with others. Oh, the pain, will it ever stop... May 03 16:40:31 educa: yes, polling for interrupts is en vogue in 2012 May 03 16:41:04 my problem with poll is it doesn't allow me to avoid entire redesign of my event loop, which is based on libev. May 03 16:41:26 POLLPRI is not supported by libev... May 03 16:41:49 why the writers of gpio didn't make the device simply work like input events, I have no idea. May 03 16:42:13 maybe because that would have just worked? May 03 16:42:48 jkridner__: send a patch? May 03 16:42:54 https://www.ridgerun.com/developer/wiki/index.php/Gpio-int-test.c provides a nice poll example, but I can't use it. May 03 16:43:16 I'm considering it, but I was hoping to get something working first. May 03 16:43:29 why not use it? May 03 16:43:46 because libev doesn't support POLLPRI. May 03 16:43:58 and my event loop is build with libev. May 03 16:44:09 send a patch? May 03 16:48:25 jkridner__: is it just a matter of setting that flag? May 03 16:48:49 lol: http://comments.gmane.org/gmane.comp.lib.ev/1562 May 03 16:48:59 I don't know, but the documentation clearly states that it is not supported. May 03 16:49:22 and that stops you? May 03 16:49:37 what mext? red traffic lights? May 03 16:49:43 next* May 03 16:49:49 well, I need to see if epoll supports POLLPRI... May 03 16:50:04 what is wrong about that response is that epoll *is* Linux specific. May 03 16:50:33 well, I bet libev is widely used on HPUX, AIX, Solaris etc... May 03 16:50:49 so adding linux specific stuff must be a burden.... May 03 16:50:58 yet even to find people to support it.... May 03 16:51:45 i guess linux doesn't have kqueue May 03 16:52:11 and? May 03 17:11:16 jkridner__: why not edit your board file and set up the gpio with gpio-keys, then you can have the desired input event May 03 17:11:43 I guess he wants it more generic May 03 17:11:49 because I'm trying to make a dynamic environment. May 03 17:11:52 yup. May 03 17:12:05 I don't think I can insmod gpio-keys w/o platform data. May 03 17:12:14 use DT May 03 17:12:44 can I use DT to provide the platform data and still unload the gpio-keys driver? May 03 17:13:00 for that matter, it would be nice to be able to unload gpio-leds May 03 17:14:47 Hi All, does anyone have some tips to get oss support from the alsa drivers under Angstrom on the Beaglebone? I can't find the /dev/dsp folder nor the right kernel module to load May 03 17:14:53 I almost have epoll working on GPIO events---that'll hopefully work with libev/node.js. May 03 17:15:09 for some reason, the read only returns the first time, so I probably have to issue a seek. May 03 17:15:12 is there a way to install it or do I have to compile a custom kernel? May 03 17:18:04 discuss: http://www.hotbot.com/ May 03 17:19:05 the page has more white than I recall from last time I looked at it, ~1997 May 03 17:21:32 jkridner__: yes, you need to issue a seek to "reset" the poll May 03 17:22:16 jkridner__: oh, you are using epoll, I've never used that May 03 17:22:21 Jelmer_: OSS emulation is most likely not configured May 03 17:22:41 or maybe you need to load a module May 03 17:22:41 can we have oss with alsa emulation instead? May 03 17:22:50 jkridner__: but with poll you definitely need a seek after each poll completes May 03 17:22:50 seems to work similarly, as far as I have encountered so far, despite the warnings. May 03 17:23:52 i think you need to modprobe the snd-pcm-oss module, but that fails because its not there May 03 17:23:59 now how do i get it there? May 03 17:25:48 sakoman_: oddly, my polltest.c code didn't have a seek and kept reading new values. May 03 17:25:54 i just need the /dev/dsp and /dev/mixer interfaces for my application, i don't care who emulates whom ;) May 03 17:26:00 Jelmer_: do you have aoss? May 03 17:26:51 yes May 03 17:27:00 aoss May 03 17:27:16 thats the new alsa way of doing oss May 03 17:29:17 ooh, that got me a little further indeed May 03 17:29:50 still got /dev/dsp: Invalid argument but thats already better than file not found May 03 17:30:00 thanks av500 ! May 03 17:30:23 de rien May 03 17:32:09 k, I'm happy with my epolltest May 03 17:34:36 koen: I guess you dropped snd-pcm-oss? or is that still in modules.tgz? May 03 17:41:48 fyi: https://gist.github.com/2587524 May 03 17:42:27 av500: dunno May 03 17:42:45 av500: I haven't usethe OSS interface in ages May 03 17:43:37 koen: but you might now if it being built? May 03 17:43:53 zcat /proc/config.gz May 03 17:45:31 CONFIG_SND_PCM_OSS=y May 03 17:45:45 * av500 just remembered he had a ssh'able beagle at home May 03 17:47:41 hmm, and /dev/dsp May 03 17:55:01 Anyone else getting a 500 on this https://github.com/koenkooi/linux/branches May 03 17:55:49 ah a couple of refreshes did the trick May 03 18:00:40 github is often flaky like that for me on the branches/ links May 03 18:01:41 I think tubes get a little clogged from time to time May 03 18:07:00 jkridner, the sample you're using is missing the needed lseek in the loop. The you get a value each time. May 03 18:07:27 I have an lseek in the loop May 03 18:07:45 Then what is not working? May 03 18:07:50 gpios are seekable? May 03 18:07:53 it is working. May 03 18:08:03 I was just showing off the final working test program. May 03 18:08:26 av500: yes, I'm seeking time. May 03 18:08:33 Ah, must have missed something. Just saw your troubles. May 03 18:09:00 the last comment was that I needed a seek, so I added it and posted the results along with the working output. May 03 18:09:26 you can see that the value is toggling between 1 and 0... May 03 18:09:49 i could tell locally it was happing at the 1 second periodic rate at which I have the input toggling. May 03 18:10:02 Yup, I should have read the complete backlog of several hours not being online. May 03 18:10:38 I haven't bothered to save the commands to set the GPIO to the right mode. May 03 18:10:50 echo in > direction May 03 18:10:55 echo both > edge May 03 18:11:07 quit using the raw GPIO interfaces for LEDs! May 03 18:11:27 who said anything about LEDs? May 03 18:11:28 use a proper driver so you don't accidentally fry the ball drivers! May 03 18:11:39 that was a general complain May 03 18:11:42 :) May 03 18:11:43 :) May 03 18:12:01 letting sw folks configure pinmuxes and gpio direction is downright dangerous May 03 18:12:11 it is laughable how many people do it in videos that they post to the web. May 03 18:12:23 indeed. May 03 18:12:26 *nod* May 03 18:12:47 at least series resistors on the board might have been a bit nice, but it would mess up some interfacing. May 03 18:12:51 its the HW guys that have to make sure nothing bad can happen May 03 18:13:14 okay, let's forbid all inputs ;) May 03 18:13:34 and put optoisolators on all the outputs May 03 18:13:38 in my test, I didn't have a resistor handy and I directly tied P8_3 to P8_5. :-o May 03 18:13:54 makes me nervous just sitting here. May 03 18:15:41 jsabeaudry: in fact, 500ing now on my repo trying to get a link for somebody :( May 03 18:16:18 mdp, ya i think github doesnt like branches ironically May 03 18:17:01 hardcore forking action works well though May 03 18:17:58 mdp: do you know of anyone hooking a CC2500 (directly) onto any of the chips you work on? May 03 18:22:03 ds2, cc2591 in this one: http://www.sparkfun.com/tutorials/334 May 03 18:22:06 is that close enough? May 03 18:22:25 let me see what the cc2591 is May 03 18:24:02 mdp: there are children present :P May 03 18:24:09 nm, I figured it was in the same family without looking May 03 18:24:20 mdp: this one is just a radio... doesn't have the demod blocks. was tryign to find out how much bypassing I need extra on the SPI lines May 03 18:24:24 it's just a range extender companion May 03 18:24:27 yeah May 03 18:24:34 yeah May 03 18:25:09 bah, 3703 has no sgx May 03 18:25:16 nest cannot run android - fail May 03 18:25:28 guess I'll just throw it into the entire mix of other variables May 03 18:25:37 mdp: so that is the infamous thermostat? May 03 18:25:42 av500: GB should be runable May 03 18:26:58 hmm, 64MB only May 03 18:27:04 will be tough for GB even May 03 18:27:41 depends on what patches you pull in May 03 18:28:01 IIRC, I think the at91sam folks have patches to make it run okay May 03 18:28:11 I'm sure they were warned that they would fail as a thermostat if they didn't plan for an ICS upgrade path May 03 18:28:16 can I patch /dev/mem and make it bigger? May 03 18:28:23 but we still take their money May 03 18:28:39 food, bb May 03 18:29:20 ds2, I might have some info but it'll take a few minutes of digging May 03 19:53:24 mranostay: that is a fancy thermostat, yes May 03 20:03:21 rcn-ee: is sound supposed to work in 3.2? May 03 20:03:38 rcn-ee: I get an error in xbmc, but I haven't looked beyond that :) May 03 20:04:25 * prpplague pokes koen in the eye with a sharp stick May 03 20:04:37 that's not very nice May 03 20:04:49 koen: you had a chance to try the new version of fb-test? May 03 20:04:55 sound on the beagle is a little bit of a mess in 3.2, you need right combo of module/built-in and the right settings for alsa, along with turnning on the right channel... May 03 20:05:11 i kinda gave up figuring that one out. May 03 20:05:14 I sent patches to fix the module messup a long, long time ago May 03 20:05:17 they were ignored May 03 20:05:30 * mru is not motivated to forward-port and resend May 03 20:05:37 yeap, they were, i pulled them in.. then they re-wrote them again. ;) May 03 20:05:54 prpplague: not yet, stuck in documentation land May 03 20:06:01 koen: ahh ok May 03 20:07:06 any1 happen to know if there is a gnu pascal to run on the angstrom os that came with my bone rev5 ? May 03 20:08:07 pascal, yuck May 03 20:08:13 hasn't that been deleted yet? May 03 20:08:24 rm -rf /pascal May 03 20:09:03 nice to see pascal has fans here. Did you know that pascal sometimes even outputs more optimised asm then c? May 03 20:09:09 (NOT ALWAYS of course) :) May 03 20:09:20 that's a stupid statement May 03 20:09:39 C and pascal as languages sit at roughly the same level May 03 20:09:54 how efficient code you get depends entirely on the compiler May 03 20:10:04 pascal is 4x more verbose :-D May 03 20:10:05 your "yuck" was also nog that nice May 03 20:10:16 nog=not May 03 20:10:24 it was aimed at pascal, not you May 03 20:10:25 I just meant it, pascal isn't that bad May 03 20:10:36 I know mru May 03 20:10:41 absolutely no hard feelings May 03 20:10:42 ;) May 03 20:10:44 the linux pascal compiler sucks BTW May 03 20:10:45 for anything more elaborate than "hello world", pascal is actually quite horrible May 03 20:10:51 not exactly a ringing endorsement :) May 03 20:11:05 XorA: "the"? May 03 20:11:10 * XorA has earned many $$$ May 03 20:11:16 mru: there is more than one? May 03 20:11:50 XorA: there is one? May 03 20:12:09 problem is all my life I code in delphi on windows, so used to pascal. I know basics of ansi C, but now it looks like I will write on the beaglebone in C and since I want to have a running program which is also listening on a socket for some data, it looks like I'll have a lot to learn, most likely about threading..... May 03 20:12:41 or learn to use select/poll/epoll May 03 20:13:02 ic, well anyway lots to learn :) May 03 20:13:10 pascal or no pascal, programming beaglebone/linux will be nothing like delphi on windows May 03 20:13:24 mru , of course its totally different May 03 20:13:49 * XorA remembers the linux delphi port May 03 20:13:56 lazarus May 03 20:13:56 koen, i just pushed mru's audio patch fixes i had to the shared tree, it majorly fixes the module loading, you'll still have to set the correct audio channel thou.. May 03 20:14:38 Release Notes: This release adds an Objective-Pascal dialect (supported on all Mac OS X and iOS targets), many Delphi compatibility mode improvements, and various new ARM code generator features. May 03 20:14:46 sounds like it cross compiles :-D May 03 20:15:25 USB sound works ok on the BBone. No module stuff to fool with May 03 20:20:37 <_av500_> for some reason delphi does not want to die May 03 20:20:55 ;) May 03 20:21:08 pyscript is by delphi? May 03 20:21:20 <_av500_> i know a delphi dev that went android and his company had a hard time to replace him May 03 20:22:51 <_av500_> delphi is the cobol of the PC era May 03 20:23:17 * XorA used to make money from Delphi May 03 20:23:21 many years ago May 03 20:23:28 isn't Delphi like the remote ancestor of VB? May 03 20:23:51 <_av500_> like javascript is the cobol of the web 2.0 era May 03 20:24:05 I thought Java was the new COBOL May 03 20:24:11 it is May 03 20:24:14 javascript != java May 03 20:24:23 <_av500_> no really? May 03 20:24:24 java is the new enterprise cobol May 03 20:24:33 everything has a cobol May 03 20:25:15 * _av500_ eats the kids' leftover school sandwiches May 03 20:25:20 I used to work in a COBOL shop. And people criticize Java for being too wordy... May 03 20:25:28 mru: nah, cobol devs still make more ;) May 03 20:25:35 i thought delphi was a response to VB May 03 20:25:48 :) May 03 20:26:04 <_av500_> delphi was borlands reponse to "huh, a gui" May 03 20:26:07 C++Builder was the bastard child of Delphi :-D May 03 20:26:14 an amusing mix of Pascal and C May 03 20:26:17 I don't remember which came first of delphi and vb, but I don't think they are directly related May 03 20:26:32 <_av500_> vb came later May 03 20:26:33 VB and Delphi pretty much came same time May 03 20:26:45 <_av500_> or that May 03 20:27:00 I was using Deplhi 1 and VB 2 at same time I think May 03 20:27:01 either way, no real programmer would touch either of them with a 10-foot pole May 03 20:27:11 visual foxpro and visual objects May 03 20:27:18 VB started on windows 3.1 May 03 20:27:19 so XorA must either have a 20-foot pole or he's not a real programmer May 03 20:27:32 delphi was always more powerfull then vb May 03 20:27:41 <_av500_> mru: he is a brogrammer May 03 20:27:44 but its more used by very specific customers May 03 20:28:09 no-one has complained of my pole size anyway ;-) May 03 20:28:37 but when it comes to languages I program what people pay $$$ for May 03 20:28:42 Knowledge of dead programming languages can come in handy if you are a contractor who gets called in to fix very old systems. An employment person once told me to leave the "MUMPS" in my resume... May 03 20:28:58 so Ive done VB, Delphi, Javasctips, Java, Perl May 03 20:29:01 maybe a stupid question, but I'm currently having fun (not) editing a .c file using putty on the beaglebone itself. Is there a way (doable for beginner like me) to get access to my /home/root like if it were a physical drive on my windows pc ? May 03 20:29:11 <_av500_> sure May 03 20:29:18 <_av500_> oh, windows May 03 20:29:21 <_av500_> well, samba May 03 20:29:26 no programmer would touch it with a 10M pole ;) May 03 20:29:26 opkg install samba :-D May 03 20:29:46 anyone booted the latest Angstrom img on beaglebone? The original img works fine but the new one stalls... May 03 20:29:50 samba lets you access a remote Windows "share" May 03 20:30:03 <_av500_> it also serves one May 03 20:30:34 <_av500_> educa: you can also export a drive on windows and mount it with -t cifs May 03 20:30:43 <_av500_> thats probably easier May 03 20:31:03 <_av500_> instand backup as well May 03 20:31:04 * XorA wonders if he can get Pascal for his zx81 May 03 20:31:10 <_av500_> instant May 03 20:31:12 ic, would be easy if I knew what to type May 03 20:31:25 <_av500_> type your account May 03 20:31:29 av500 what do you mean ? I type this in the beagle? May 03 20:31:31 <_av500_> type your account # and pin May 03 20:31:35 <_av500_> no here May 03 20:31:45 * ogra_ gets pen and paper May 03 20:31:55 * _av500_ opens log file May 03 20:32:00 ogra_: make mine a tall cold one? May 03 20:32:37 XorA, which ? your pin or your accuont # ? :P May 03 20:33:19 ogra_: the beers you can buy now they got cheaper May 03 20:33:39 <_av500_> educa: mount -t cifs windows_ip/share /some_random_folder May 03 20:33:43 <_av500_> something like that May 03 20:33:48 <_av500_> google knows May 03 20:33:54 XorA, oh, no prob, yeah, indeed May 03 20:34:07 /windows_ip/share /folder May 03 20:34:11 //windows_ip/share /folder May 03 20:34:16 <_av500_> /////// May 03 20:34:28 goddam irc nicking me // May 03 20:34:42 /\/\/\/ May 03 20:55:24 ds2: I went back and looked a bit at cc2500..definitely haven't seen anybody use that one with these parts yet…but I only see the tiniest of tiny tiny fractions of designs May 03 20:59:59 koen: what version do you have on your bone? because my "zcat /proc/config.gz" gives "# CONFIG_SND_MIXER_OSS is not set # CONFIG_SND_PCM_OSS is not set # CONFIG_SND_SEQUENCER_OSS is not set" May 03 21:00:34 mine is: Linux version 3.2.0+ (koen@dominion) (gcc version 4.5.4 20111126 (prerelease) (GCC) ) #1 Fri Feb 10 09:53:04 CET 2012 May 03 21:01:08 oh that was for av500, sorry koen May 03 21:05:59 <_av500_> Linux beagleboard 3.0.17+ #1 Fri Jan 13 14:39:56 CET 2012 armv7l GNU/Linux May 03 21:06:05 nice - ' Large open rooms, detached garage, needs paint carpet and a big dumpster.' May 03 21:07:35 <_av500_> you like big dumpsters? May 03 21:08:39 Russ, "previously featured on Hoarders" May 03 21:12:35 hmm so i guess its not included anymore in the newer builds May 03 21:20:08 <_av500_> blame koen May 03 21:23:46 I do like an honest listing May 03 21:55:04 mdp: okya, thanks. was the CC2400 more common? May 03 21:57:15 I haven't seen that either…several designs are using the 802.15.4 stuff and zigbee for access to the smart energy profiles May 03 21:59:14 I've got a couple cc2400 modules coming for my own personal eval..so in theory I'll have a project with those eventually :) May 03 22:01:02 supposely, the cc2500 is to replace the cc2400 May 03 22:01:09 cc2400 is listed as NRND May 03 22:01:22 outside of our processors, I see a lot of people in mcu-land doing cc110l/cc1100 and cc2400 for cost sensitive stuff May 03 22:02:04 there is a nice bt sniffer designed based around the cc2400 May 03 22:02:21 yeah, I got some cheap modules from circuitspecialists May 03 22:02:24 would be nice to get that working on say the bone with a CC2500 (actually a RF2500T) May 03 22:02:35 cheap?!?!? how can you get cheaper then RF2500T's? May 03 22:02:50 want to compare against the 110l May 03 22:03:05 110L? is that a nordic or? May 03 22:06:15 that's the "value line" xcvr May 03 22:06:24 not 2.4ghz May 03 22:06:35 ah. didn't recogonize it w/o the CC prefix May 03 22:07:54 oops..gotta go from work to "chores" now May 03 22:07:55 bbl May 03 22:50:34 mmh try to compile py-openzwave (went fine for the c++ part) but now im stuck with cython her http://pastebin.com/vxr7n7NX --- where does cython grab this strange logcall without _loglevel? source is at http://code.google.com/p/open-zwave/ May 04 00:44:20 is anyone aware of an arm platfrom with a reasonable amount of cores/ram (embedded isn't necessary, I'm going to produce funtoo stages for the beagle, and need a native environment) May 04 00:44:29 the panda is almost big enough May 04 00:44:32 twice the ram would probably do it May 04 00:45:18 why not cross-compile? May 04 00:45:25 mru, takes longer? :) May 04 00:45:34 wtf May 04 00:45:56 trying to cross compile from a 386SX-16? May 04 00:46:19 but if you insist, try to get your hands on some calxeda kit May 04 00:46:21 mru, the panda (the latest gen arms) are faster than cross-comiling May 04 00:46:23 ok May 04 00:46:49 the panda is in no way faster than cross-compiling on even a semi-sluggish pc May 04 00:47:25 oh please, don't tell me you've been compiling under qemu May 04 00:49:06 yeah :/ May 04 00:49:33 I am so sorry May 04 00:49:46 nobody should have to endure such pain May 04 00:49:55 ouch May 04 00:50:07 jay6981: ever heard of scarebox? May 04 00:50:22 heh no May 04 00:50:30 good for you May 04 00:50:42 the name reminds me of dune for some reason May 04 00:50:51 that thing he had to stick his hand into May 04 01:03:07 trelane: I'd have to ask prp^2 if any pandas had more than 1GB RAM May 04 01:03:24 btw, I'm always more responsive here than private messages. May 04 01:03:26 crap sorry that autotargeted private message May 04 01:03:30 yeah just realized it May 04 01:03:41 if there is one, I'm going to get a funtoo build more current than last october May 04 01:03:54 even if they could just plug a cable in, stick a drive on it, and plug in ethernet so we could get to it May 04 01:03:55 * mru still says cross-compile May 04 01:04:08 mru, metro doesn't support cross compiling May 04 01:04:14 wtf is metro? May 04 01:04:16 so it's qemu (sucks as you so accurately point out) May 04 01:04:18 and why can't it be fixed May 04 01:04:27 mru, a stage building and qa system for Gentoo/Funtoo May 04 01:04:36 mru, I'm assuming it will but probably not in the near future :) May 04 01:05:15 people like you taking the cowardly approach sure won't speed it up May 04 01:05:23 go ahead, file some bug reports May 04 01:05:25 troll some devs May 04 01:05:37 maybe even send some patches May 04 01:05:54 mru, the dev is a very good friend of mine, we're aware of the problem and time to fix is long due to upstream gentoo issues May 04 01:06:00 mru, problem is political not technical :) May 04 01:06:13 then shoot the president May 04 01:06:31 I think the Secret Service would either complain or buy me a hooker May 04 01:06:32 trelane, instead of using metro in qemu, why not use a chroot with a different os on the panda? May 04 01:06:47 rcn-ee, that's the goal, but I don't have a panda May 04 01:06:49 rcn-ee: I don't see what good that would do May 04 01:06:52 was hoping to get access to one to do so :) May 04 01:07:03 it wouldn't be qemu. ;) May 04 01:07:10 rcn-ee, concur :) May 04 01:07:14 was hoping for a bit more ram :) May 04 01:07:20 but why chroot? May 04 01:07:31 mru, really I wouldn't need a chroot except for QA/testing May 04 01:07:39 metro takes care of that :) May 04 01:07:55 sounds like you just need a board. ;) May 04 01:08:04 rcn-ee, with a bit more ram, hence my question :) May 04 01:08:06 ooh May 04 01:08:10 question didnt get here because I was an idiot! May 04 01:08:11 hang on May 04 01:08:20 are there any pandaboards sitting around with more than 1gb of ram? May 04 01:08:22 was the question May 04 01:08:27 rcn-ee, that was the direction I was going :) May 04 01:08:45 the omap4 can address 2GB May 04 01:08:49 * jkridner__ reads http://www.funtoo.org/wiki/Metro May 04 01:08:54 ah.. not that i have seen... just stick a fast usb disk and add swap, sure it's slower then ram. ;) May 04 01:09:07 no idea if such chips exist though May 04 01:09:09 jkridner__, between that and a script to build binpackages of everything else with standard portage :) May 04 01:09:14 swap... eeew May 04 01:09:37 trelane: maybe you can cross-compile with distcc May 04 01:09:37 jkridner__, we could automate it and totally eliminate OE :) May 04 01:10:24 mru, and yet I'm willing to send jkridner green sheets of paper with dead white guys on it May 04 01:10:25 I think mru builds Gentoo fine w/o OE using cross-compilation. May 04 01:10:28 so I was hoping to go that route :) May 04 01:10:47 jkridner__: yes, but I don't build perl... May 04 01:10:48 I'm not sure which problem Metro solves that otherwise needs OE, thus reading. May 04 01:10:54 ah. May 04 01:11:09 I can live without perl on the dev boards May 04 01:11:21 jkridner__, you don't need OE is a HUGE fix :) funtoo/gentoo has much better QA, and on and on :) May 04 01:11:28 and a few packages need some help to cross-compile properly May 04 01:11:52 yeah, some tricky apps are why I gave up on my own Gentoo builds as koen was soooooo much faster at building apps w/ OE. May 04 01:11:54 mostly setting some variables to skip autoconf tests that fuck up May 04 01:12:14 jkridner__, you'll remember my vivid description of OE, and while your points are valid, I think losing OE is a huge plus in life :) May 04 01:13:06 well, needing to go back to native builds isn't a step in the right direction. May 04 01:13:35 jkridner__, my goal is to provide portage binpkgs May 04 01:13:42 so those native builds are only necessary once :) May 04 01:13:44 perhaps some way to do native invocations in testing make sense for QA. May 04 01:14:17 but, for the distro to advance, others need to be able to replicate. May 04 01:14:19 jkridner__, portage supports using binary packages and basically assuming that that package is taking care of pkg_build and pkg_install May 04 01:14:28 that's what metro and the scripts are good for May 04 01:14:34 simply opensource May 04 01:14:46 metro's on github, any other scripts can be stuck there too :) May 04 01:14:56 but if you go the path of native builds, then the problem of cross-compiling isn't solved. May 04 01:15:00 Funtoo will fix in our tree and submit build bugs upstream :) May 04 01:15:23 perhaps I'm missing something, but I'm not 100% sure why it needs to be? May 04 01:15:25 fixing the .ebuilds to support cross-compiling seems like a better fix. May 04 01:15:49 my concern there would be the sheer quantities of arches May 04 01:16:12 amd64_armel i386_armel i386_sparc and on and on for every source and every target that GCC supports May 04 01:16:16 doesn't gentoo/funtoo already support a huge # of archs? May 04 01:16:18 that's a HUGE insane amount of QA May 04 01:16:21 yes May 04 01:16:35 but fixing every gcc hiccup is not trivial May 04 01:16:36 do you mean as cross-hosts? May 04 01:16:42 right May 04 01:17:10 because we shouldn't assume that arm(whatever) is the only target for cross-compiling May 04 01:17:18 in most cases, fixing one cross-build pair fixes them all May 04 01:17:20 true, but fix i686->armhf/armel and you'd solve 90% of build platforms. May 04 01:17:43 sometimes the fix depends on things like the word size of the target May 04 01:17:47 and, i agree with mru, most of the issues are in telling autotools about the target, not something host specific. May 04 01:17:50 not exactly rocket surgery May 04 01:18:14 mru, tell that to North Korea, they have a lot of trouble iwth Rocket Surgery, and dedicate like 90% of hte GDP to it! :) May 04 01:18:25 well, perhaps not rocket science, but I'm not so convinced it isn't rocket surgery. May 04 01:18:26 their gdp isn't very high though May 04 01:18:35 true enough May 04 01:18:54 * trelane points to NASA which doesn't even have a rocket right now May 04 01:19:25 they had a saturn v sitting on the ground when I visited last year May 04 01:19:30 again though, my goal wouldn't be to fix every quirk of autotools May 04 01:19:35 mru, and they can't actually build one now May 04 01:19:38 knowledge loss sucks May 04 01:19:46 of course they could if they wanted May 04 01:19:53 sure, a few old tapes have been lost May 04 01:20:02 Warner Von Braun, he wrote a book once called "I aim for the stars", he should have noted that usually he hit London May 04 01:21:02 anyway, cross-compiling will always be a good thing to have working May 04 01:21:09 since some machines will always be faster than others May 04 01:21:42 agreed, but is solving that problem worth continuing with the suffering of OE? May 04 01:21:53 what has OE got to do with anything? May 04 01:22:19 most cross-compiling problems are easily fixed by patching the package source May 04 01:22:28 I compile against oe generate libs... it's not that painful once I figured out how May 04 01:22:32 right but that's not my goal for today May 04 01:22:56 my goal would be to provide an alternative simpler than OE (which metro+portage is) May 04 01:23:21 if it relies on native building, it's not an alternative May 04 01:26:07 I think that's short-sighted given the speed with which ARM is growing and the speed/performance is increasing May 04 01:32:14 I just don't understand how given the power in the pandas we still have a flight involving stops to get to ARM May 04 01:32:40 what fight? May 04 01:32:43 it just works May 04 01:33:37 ds2, that's kind of my point (why crosscompile at this point?) May 04 01:33:54 cross compiling is no fight May 04 01:33:59 it just works May 04 01:34:15 unless you are using one of those totally completely absurd pieces of code that insist on using libtools and/or autoconf May 04 01:34:20 ds2, but that's neither the argument I'm getting nor my experience with OE/Angstrom May 04 01:34:40 ds2, which is pretty much 80% of packages (or more?) May 04 01:34:57 then don't use those crap May 04 01:35:43 I'm trying to fix it so people dont have to, I just find the basic rational "we should fix every package using libtools/autoconf to support cross-compiling for our specific case" as a waste of time and energy May 04 01:36:00 use of libtools/autoconf is a waste of time. May 04 01:36:08 remove that sort of crap. May 04 01:36:13 agreed ,but I can't convince thousands of others of that :) May 04 01:36:22 that is irrelevant. May 04 01:41:46 and besides, native compile works just fine. been doing that on the Beagle Classic years ago May 04 01:44:23 ds2, that's my point, I'm just going to abstract all the problems of cross-compiling for the small cost of using a beagle as a build env, or using pre-built packages (Which is essentially the situation with OE anyway except you can use x86 and encounter LOTS MOAR PROBLEMS (TM) May 04 01:44:55 and with that hanging parenthesis, I'm leaving to watch TV for a bit while things compile) May 04 01:44:57 there is not a lot of problems with either May 04 02:35:04 * heathkid has arrived! :) **** ENDING LOGGING AT Fri May 04 02:59:59 2012