**** BEGIN LOGGING AT Wed Jul 22 02:59:59 2015 Jul 22 03:06:51 nerdboy: back Jul 22 03:15:51 this one should not be too distracting => http://www.umich.edu/~eecs381/handouts/CppHeaderFileGuidelines.pdf Jul 22 03:17:20 or this one, as the case may be http://www.umich.edu/~eecs381/handouts/CHeaderFileGuidelines.pdf Jul 22 03:17:31 mostly the same Jul 22 03:18:02 proper ordering is always good Jul 22 03:18:50 http://www.umich.edu/~eecs381/handouts/handouts.html <= found the !index file Jul 22 03:21:47 awesome,thanks Jul 22 03:23:11 I tainted the header file in the API docs on purpose a bit, so I can get some more guidelines from you and alexanderhiam on how I should design around some ideas Jul 22 03:23:20 ei. more of an idea list Jul 22 03:23:55 lots of work to do still, off to some more reading Jul 22 03:24:40 *and doing Jul 22 03:24:59 don't forget "try it and see" Jul 22 03:26:27 hmmm Jul 22 03:27:06 nerdboy: yup, that's the plan Jul 22 03:27:12 also reports Jul 22 03:27:12 neemo: just try the basic C header stuff and apply it to yours Jul 22 03:55:41 * nerdboy full of burritos Jul 22 14:15:09 ebadawy: reading your proposal: Sprint 5 Playing with the current style Jul 22 14:15:29 apaar: fflush where? Jul 22 14:15:52 i don't think you lost a lot of time, basically this week you will finish the BBUI integration, is this correct? Jul 22 14:16:10 Abhishek__ :in the C library pru_read function Jul 22 14:16:39 https://github.com/Apaar/PRU-Bridge/blob/master/C_Lib/Library/libpru_bridge.c#L112 Jul 22 14:16:58 apaar: You do not use fopen for sysfs attributes. Jul 22 14:17:02 Use open() Jul 22 14:17:18 DiegoTc: yeah, what I meant by style back there was about, yours Jul 22 14:17:32 yes, but that was fixed Jul 22 14:17:34 your rock Jul 22 14:17:35 thought that I start my work on top of your style. Jul 22 14:17:47 yeah, I guess so. Jul 22 14:17:47 Abhishek__ :because of the buffering right?? but i needed rewind it was not working for multiple values otherwise Jul 22 14:17:58 have you tried what I did so far Jul 22 14:18:03 I mean testing it? Jul 22 14:18:17 Abhishek__ :https://github.com/Apaar/PRU-Bridge/commit/f4d712f232f293c13d794db25479f56644a04c77 Jul 22 14:19:10 this is were i switched b/w open and fopen Jul 22 14:19:38 i did yesterday found some things ebadawy, but not quite sure if was my internet connections Jul 22 14:19:44 or the web app Jul 22 14:19:51 so I didn't want to publish then Jul 22 14:20:09 basically you're missing the following ebadawy Jul 22 14:20:13 1) integrate BBUI Jul 22 14:20:20 2) Add more tutorials Jul 22 14:20:26 3) home pga eof bone101 Jul 22 14:20:45 4) fixing bugs (all systems have then, we need to find then) Jul 22 14:20:59 5) fixed style things probably jkridner will find Jul 22 14:21:24 do you think I'm missing something ebadawy ? Jul 22 14:21:37 no, I think that is all. Jul 22 14:22:11 so, can we have a nice time schedule for them? Jul 22 14:24:58 apaar: Look here: https://github.com/Apaar/PRU-Bridge/commit/f4d712f232f293c13d794db25479f56644a04c77#diff-4c20186bbb6a4c048ee0d7751b640103L60 , you switch from fopen to fopen ? Jul 22 14:26:02 yep there was some issue with multiple reads which needed rewind Jul 22 14:26:30 rewind needs the file pointer which open does not return Jul 22 14:27:07 Abhishek__ :^ Jul 22 14:27:25 I see that yes ebadawy, you will finish integrating the BBUI this week? Jul 22 14:27:49 apaar: Does changing fopen to open, fread to read, fwrite to write and so on fix it? Jul 22 14:28:27 DiegoTc: not sure, but I will try to do my best for that. Jul 22 14:28:41 Abhishek__ :i switched the other way round... Jul 22 14:28:52 Abhishek__ :if you mean open - Jul 22 14:28:58 to fopen Jul 22 14:29:06 yes it did solve the issue Jul 22 14:29:33 but first need to take with jkridner first to make sure I do it in the right way. Jul 22 14:50:08 yes, jkridner has the final OK, if he doesn't like we're in trouble ebadawy Jul 22 14:52:42 DiegoTc: so the output of this week should be adding jkridner rewrite as a bonecard, right? Jul 22 14:52:50 yes Jul 22 14:53:03 and included number 5 also. Jul 22 14:53:12 and it should be done like scaling it down a little bit to fit with bonecard size? Jul 22 14:53:13 that will be to have the editor on the bonecard shape Jul 22 14:53:17 and not the way you have it Jul 22 14:53:59 I will add 6) have a stop a nd run button at the moment of runningthe code Jul 22 14:54:12 to BBUI? Jul 22 14:55:27 no Jul 22 14:55:46 let's talk of this in some minutes, that's the last challenge in the project ;) Jul 22 14:56:44 * ebadawy curious to know what it is Jul 22 15:01:07 ebadawy: give me some minutes, that need explanation Jul 22 15:01:21 okay Jul 22 15:01:26 well Jul 22 15:01:27 and I'm a little bussy right now, so I will probably not explained it weel Jul 22 15:02:00 np, could be later Jul 22 15:38:31 gm all Jul 22 15:39:15 good morning/evening Jul 22 15:42:26 hi jkridner Jul 22 15:42:32 howdy Jul 22 15:44:03 namaste jkridner Jul 22 15:44:07 back to university; classes resumed; and heavy rains. Jul 22 15:49:08 gm all Jul 22 15:50:34 gm/ge all Jul 22 15:51:07 gm all Jul 22 15:51:27 gm Jul 22 15:53:12 10 min left for the gm? Jul 22 15:59:11 <_av500_> alohi Jul 22 15:59:11 <_av500_> 2 min to 5min to start Jul 22 15:59:42 good morning and good afternoon everybody :) Jul 22 15:59:43 a countdown for the countdown? Jul 22 16:00:30 namaste ankur Jul 22 16:01:11 countdownception? Jul 22 16:01:48 lol neemo Jul 22 16:01:55 seems like we invented seconds/minutes/hours/arbitrarytimemeasures for that very purpose Jul 22 16:02:33 greetings everyone Jul 22 16:02:45 <_av500_> 3min Jul 22 16:03:57 2min Jul 22 16:05:27 *splat* Jul 22 16:05:57 * nerdboy waiting for bad shark movie finale Jul 22 16:06:19 <_av500_> ok Jul 22 16:06:20 <_av500_> time Jul 22 16:06:25 <_av500_> 1) WELCOME everybody Jul 22 16:06:31 hey Jul 22 16:06:51 <_av500_> 2) who's here Jul 22 16:06:53 hiello500_ Jul 22 16:06:54 okay, so we have - arianepaola ankur apaar azizulhakim ebadawy neemo shubhangi rohith Jul 22 16:06:54 'ello Jul 22 16:06:56 hi Jul 22 16:06:59 moin Jul 22 16:07:01 * hello _av500_ Jul 22 16:07:02 hello Jul 22 16:07:06 here Jul 22 16:07:08 yup yup Jul 22 16:07:13 hi all Jul 22 16:07:22 yup Jul 22 16:08:02 * _av500_ has nice irc colors rainbow now Jul 22 16:08:08 <_av500_> ok, all there, very good Jul 22 16:08:11 <_av500_> reports too Jul 22 16:08:31 maybe next year we'll learn how to sound-off for rool-call Jul 22 16:08:40 *roll-call even Jul 22 16:09:17 <_av500_> we have roughly 3.5 weeks left Jul 22 16:09:33 <_av500_> finals come sooner than you think :) Jul 22 16:09:59 3.5 week! that is sooner than I expected!! Jul 22 16:10:06 <_av500_> yes Jul 22 16:10:15 <_av500_> google j:) Jul 22 16:10:19 <_av500_> oops Jul 22 16:10:20 <_av500_> :) Jul 22 16:10:31 <_av500_> 3) status report planning Jul 22 16:10:49 <_av500_> I think everybody did it once, so I need 2 volunteers for today Jul 22 16:11:06 * moto-timo comes in late and looks sheepish Jul 22 16:11:21 hi moto-timo Jul 22 16:11:24 * nerdboy throws a donut at moto-timo Jul 22 16:11:34 <_av500_> shubhangi: azizulhakim: you started, want to continue? Jul 22 16:11:46 moto-timo: greetings Jul 22 16:11:47 okay Jul 22 16:12:05 _av500_: may i ? Jul 22 16:12:13 <_av500_> shubhangi: sec Jul 22 16:12:16 <_av500_> shubhangi: not yet Jul 22 16:12:19 <_av500_> just planning Jul 22 16:12:26 <_av500_> azizulhakim_: you ok for a report today? Jul 22 16:12:56 av500: not today, on a mobile device :( Jul 22 16:12:58 <_av500_> ok Jul 22 16:13:01 <_av500_> apaar: ? Jul 22 16:13:22 <_av500_> neemo: ? Jul 22 16:13:23 _av500_ :i was hoping to get the benchmarks up before... Jul 22 16:13:27 <_av500_> ok Jul 22 16:13:28 <_av500_> neemo: ? Jul 22 16:13:42 * nerdboy not a big fan of phone-irc... Jul 22 16:14:00 _av500_: don't really have much to present, still a lot to do on the API part Jul 22 16:14:05 so I'd recommend against it Jul 22 16:14:13 pass Jul 22 16:14:37 <_av500_> arianepaola: ? Jul 22 16:14:54 I did present two weeks ago, but if you want me to continue where I left off? Jul 22 16:15:00 <_av500_> yes Jul 22 16:15:06 <_av500_> apaar: neemo: next week? Jul 22 16:15:13 <_av500_> shubhangi: and arianepaola today Jul 22 16:15:15 sure :) Jul 22 16:15:34 neemo: next, or week after. Jul 22 16:15:37 <_av500_> next Jul 22 16:15:39 <_av500_> so noted Jul 22 16:15:41 I'll try to get it sorted for next Jul 22 16:15:44 <_av500_> ok Jul 22 16:15:44 :) Jul 22 16:15:58 <_av500_> 4) blocking issues Jul 22 16:16:08 <_av500_> as usual, your time to speak up Jul 22 16:16:15 time :| Jul 22 16:16:16 <_av500_> as said, only 3.5 week + Jul 22 16:16:52 jkridner: was talking with DiegoTc about what should I do with the next week Jul 22 16:17:02 minor floating point stuff, fixing it with nerdboy. solver stuff, but working on that with nerdboy as well Jul 22 16:17:07 <_av500_> I assume you are all doing fine planning with your mentors Jul 22 16:17:17 <_av500_> neemo: flaoting point in the kernel? Jul 22 16:17:22 y Jul 22 16:17:27 _av500_: not stuck but like jumping over one hurdle and encountering another. Vehicle code still gives compilation errors... Jul 22 16:17:39 _av500_ ny recommendation on the flags? Jul 22 16:17:42 although new ones Jul 22 16:17:42 any* Jul 22 16:17:48 we ended up with that I'll work with the rewrite code and integrate what is done with it to be in a bonecard Jul 22 16:18:23 jkridner: just want to confirm with you, that what I'll is scalling it down a little bit to fit with bonecard size, right? Jul 22 16:18:24 no Jul 22 16:18:28 <_av500_> neemo: I thought FP in the kernel as a nogo? Jul 22 16:18:41 ebadawy: think, we should talk about that when your moment, av500 is on the middle of the meeting Jul 22 16:19:06 _av500_: well, I've read multiple conflicting opinions on the FP part. Some say nogo, some say possible but maybe avoid for saving userspace floatingpoint context Jul 22 16:19:13 DiegoTc: but thought it is the blocking issue time Jul 22 16:19:23 _av500_: so really, depends on which camp you listen to, or so it seems. Got any better info? Jul 22 16:19:58 <_av500_> neemo: I was always assuming one should not do it Jul 22 16:20:04 maybe save that for after the mtg... Jul 22 16:20:08 <_av500_> ok Jul 22 16:20:09 yes, but _av500_ gives time to to evryone to talk, this way we don't interrupt others like right now lol Jul 22 16:20:25 nerdboy: _av500_ -> yes, later on that part Jul 22 16:20:28 <_av500_> ok Jul 22 16:20:29 oh... sorry Jul 22 16:20:47 <_av500_> as said, I hope you all do fine planning with your mentors Jul 22 16:21:01 <_av500_> to remove small and medium issues Jul 22 16:21:10 <_av500_> so any big ones today? Jul 22 16:21:23 <_av500_> rohith work with vmayoral? Jul 22 16:21:33 FP in kernel is not advised Jul 22 16:21:43 _av500_: ebadawy needs jkridner confirmation on something, this is a blocking issue for him Jul 22 16:21:47 _av500_: Will do Jul 22 16:21:51 if you help us on this :) Jul 22 16:21:52 ebadawy: yes Jul 22 16:21:57 portability and long string tugging issues Jul 22 16:22:02 ebadawy: scale it down to fit in a bonecard. Jul 22 16:22:40 _av500_: rohith hasn't met with what he agreed to last week (commiting every day), his contributions are quite poor, and it's right now when i get notified that he is stucked Jul 22 16:22:44 ds2: yeah, that's the impression I've got. but would look into it a bit more, links appreciated. as well as discussion after the mtg Jul 22 16:23:06 _av500_: he hasn't interacted with any of the mentors (3 of us) Jul 22 16:23:14 jkridner: shouhld I make a varilable like 'scale' and mul other varilables, or just changing them? Jul 22 16:23:22 ebadawy: told you, i think i know jkridner dream on bone101 :) Jul 22 16:23:29 <_av500_> vmayoral: :/ Jul 22 16:23:44 _av500_ vmayoral: I sent an email but got no reply :( Jul 22 16:24:01 <_av500_> rohith: there is IRC Jul 22 16:24:02 ebadawy: just change them to fit. Jul 22 16:24:05 <_av500_> there is me here Jul 22 16:24:09 <_av500_> and others Jul 22 16:24:16 ebadawy: doesn't need to dynamically scale... just needs to look nice in that one form-factor. Jul 22 16:24:24 rohith, _av500_: I've put special efforts to be around myself over the last days here at IRC Jul 22 16:24:25 <_av500_> rohith: I dont expect anybody to be stuck on compilation errors Jul 22 16:24:46 jkridner: OK Jul 22 16:25:00 _av500_: Multiple 15+ files are compiled at the same time... Jul 22 16:25:06 <_av500_> so? Jul 22 16:25:22 Each being quite long Jul 22 16:25:35 rohith: ? Jul 22 16:25:39 jkridner: ebadawy only remaining TOP issue will be the stop button Jul 22 16:25:41 can you paste bin it? Jul 22 16:25:43 Still trying to get rid of the errors Jul 22 16:25:46 or just gist it? Jul 22 16:25:53 vmayoral: Okay Jul 22 16:25:59 DiegoTc: I think you should tell my what is it? :) Jul 22 16:26:15 <_av500_> rohith: please reseolve this after the meeting Jul 22 16:26:24 <_av500_> 5) status reports Jul 22 16:26:26 _av500_: Sure :) Jul 22 16:26:35 ebadawy: when yuo run the code with the bonescript library the only way to stop it, is refreshing the page Jul 22 16:26:36 <_av500_> shubhangi: 15 minutes Jul 22 16:26:42 <_av500_> are yours Jul 22 16:26:43 it will execute infinte Jul 22 16:26:44 <_av500_> DiegoTc: later Jul 22 16:26:56 okay Jul 22 16:26:57 sorry _av500_ Jul 22 16:26:59 <_av500_> np Jul 22 16:27:09 so .. Jul 22 16:27:15 <_av500_> *** AGAIN: TALK TO YOUR MENTORS, NOT ONLY ON WEDNESDAYS *** Jul 22 16:27:54 <_av500_> shubhangi: go on :) Jul 22 16:28:40 since one of the main focus of this project was to get remoteproc working on newer kernel i tried to port it 4.1 Jul 22 16:29:55 went on to create a series of patches for 4.1 Jul 22 16:30:41 module compiles and probes fines on 4.1 but still has one issue Jul 22 16:31:01 doesnt probe again after removing ... Jul 22 16:31:26 what does it complain about when probing the second time? Jul 22 16:32:03 however this was taking more time since i have to trace core linux files .. particularly omap device power management which had changed Jul 22 16:32:20 ds2 : pru is registered an omap device Jul 22 16:32:44 removing the omap device was not proper for me Jul 22 16:33:01 did you forget to cleanup/restore state somewhere? Jul 22 16:33:05 i can resolve it .. just postponed .. more pressing issue was vring Jul 22 16:34:09 nerdboy: i didnt get you Jul 22 16:34:13 sorry, keep going... Jul 22 16:34:27 the vrings work :) Jul 22 16:34:42 the rproc finds vring in firmware resource table Jul 22 16:34:42 <_av500_> good Jul 22 16:34:52 registers them with virtio driver Jul 22 16:35:02 oh stuff...hmmm Jul 22 16:35:10 writes resource table to pru memory Jul 22 16:35:23 with the pa of allocated vrings Jul 22 16:35:55 so pru firmware can now fill vring buffer and kick the arm processor Jul 22 16:36:17 a kick - a generic term is implemented in our case with sysevents Jul 22 16:36:41 used to tell other processor that a msg has been written to vring Jul 22 16:36:55 as of now we have 2 vrings .. rx and tx Jul 22 16:37:36 each can have 512 buffers each .. each buffer of 512 bytes .. 56 bytes in each buffer is taken by vring descriptor Jul 22 16:37:53 so 512 - 56 for data Jul 22 16:38:14 rather 512 * ( 512 -56 ) bytes for each direction Jul 22 16:38:59 remaining task : expose a char device to read/write to vrings and test performance Jul 22 16:39:24 get 4.1 patches straight Jul 22 16:40:07 one issue remains if anyone could help Jul 22 16:40:24 i found a work around but i guess its a little hacky Jul 22 16:41:44 is there any way to find the address where a .SECTION is written inside pru memory Jul 22 16:42:00 shubhangi: you creating a linker map? Jul 22 16:42:06 for time was less .. i did this Jul 22 16:42:15 Is that information in the ELF somewhere? Jul 22 16:42:38 https://github.com/shubhi1407/PRU-framework/blob/master/firmware/src/am335x_pru.cmd#L84 Jul 22 16:42:40 Abhishek__: I'm sure it is, but the TI tools provide a .map output that is pretty useful. Jul 22 16:43:16 jkridner: Since the firmware usually will be a elf, I think it makes most sense to extract it from there Jul 22 16:43:24 if possible Jul 22 16:43:46 https://github.com/shubhi1407/PRU-framework/blob/master/firmware/src/am335x_pru.cmd#L84 ... this isnt good right ? Jul 22 16:43:56 http://elinux.org/Executable_and_Linkable_Format_%28ELF%29 Jul 22 16:45:37 from the elf .. i have that address since the rproc driver uses that info to copy the table to pru mem ... what i can do is send this address to pru Jul 22 16:47:00 _av500_ : i guess that would be the most of it ... covering major developements .. leaving out minor fixes here and there Jul 22 16:47:40 <_av500_> thanks! Jul 22 16:47:51 <_av500_> sounds good to me Jul 22 16:47:52 I remember last year that the pru-remoteproc driver even in 3.8.13 could not be loaded once unloaded without a reboot Jul 22 16:47:57 <_av500_> arianepaola_: you're up Jul 22 16:48:19 thanks _av500_ Jul 22 16:48:27 _av500_: sorry for making it quick .. i have to rush to the railway station Jul 22 16:48:48 <_av500_> shubhangi np Jul 22 16:49:31 so in the last meeting, I explained how the build system worked, how the application was built and the initial features Jul 22 16:50:10 detecting the beagle, connecting to it via ssh, using scp directly from the app Jul 22 16:51:07 now you can call the cloud 9 ide, which is on the beagle image also from the app, as jkridner suggested Jul 22 16:52:00 arianepaola: I'm a bit frustrated that the issue list hasn't been fleshed out. Jul 22 16:52:01 for the build part, there are still some minor issues during automation using appveyor and travis ci Jul 22 16:52:11 arianepaola: you've had the build working for ages now. Jul 22 16:52:27 yes jkridner Jul 22 16:52:53 I am working now on flashing the downloaded images from the application Jul 22 16:54:19 the build is being resolved, from this Friday on I will have a mac for testing and development, so this will help to resolve the bugs there Jul 22 16:55:17 although nwjs is cross platform and you can build automatically the packages, which is working, there are dependencies that require more than basic nodejs code, e.g. they need special code compilation on each platform Jul 22 16:55:50 arianepaola: no commits since Jul 7. Jul 22 16:55:52 :( Jul 22 16:56:20 jkridner, my code is in several virtual machines, I will take time to commit everything this weekend at latest Jul 22 16:56:31 sorry for that Jul 22 16:56:58 shouldn't be more than a day or two. can we say by Friday? Jul 22 16:57:21 you can have "junk branches", but it is helpful for me to see that you are coding. Jul 22 16:57:27 I think you are trying too hard to make your commits perfect... Jul 22 16:57:39 moto-timo: :-( Jul 22 16:58:06 jkridner: ok Jul 22 16:58:08 make a WIP branch? Jul 22 16:58:18 wip? Jul 22 16:58:22 work in progress Jul 22 16:58:22 vip? Jul 22 16:58:25 haha Jul 22 16:58:25 work in progress Jul 22 16:58:26 ok Jul 22 16:58:29 work-in-progress Jul 22 16:58:29 work in progress Jul 22 16:58:36 jinx! Jul 22 16:58:56 * nerdboy throws another donut at moto-timo Jul 22 16:59:01 :) Jul 22 17:00:36 so for flashing, as you know dd comes on linux and generally linux does not make problems Jul 22 17:01:06 what did take some time was to resolve the elevation of privileges on windows Jul 22 17:01:31 bundling a dd executable? Jul 22 17:01:38 when doing a spawn of the process Jul 22 17:01:45 nerdboy: win32 disk imager Jul 22 17:02:02 yeah, the ubuntu ppa thing Jul 22 17:02:23 * nerdboy hasn't actually touched windows in a long time... Jul 22 17:03:34 so next steps are to finish the work on flashing, also doing the testing on mac Jul 22 17:04:18 after that updating the emmc flash of the board from the updated image Jul 22 17:04:58 have you looked at vvu's BBBlfs? Jul 22 17:05:31 it would be nice to build on prior GSoC work Jul 22 17:06:06 * jkridner holds breath Jul 22 17:06:10 moto-timo: I have pinged vvu to let him know that I am going to use it on mac, so it would be great to have some tips from him Jul 22 17:06:36 arianepaola: you are actually using BBBlfs itself to init the bootloading, yes? Jul 22 17:07:10 sounds like another topic for issues... Jul 22 17:07:27 jkridner: not yet, I have so far been writing the images Jul 22 17:08:15 nerdboy: +1000 Jul 22 17:10:48 can i buy another espresso with that? Jul 22 17:12:25 mmmmm... espresso. Jul 22 17:12:30 did I hear 4 nays at the beginning to present and volunteered to present? I just don't understand if being sarcastic is appropriate here.... Jul 22 17:12:53 ignoring comments Jul 22 17:13:15 always, it is silent. Jul 22 17:13:33 ah, crickets Jul 22 17:13:34 things should be productive :-) Jul 22 17:13:47 thank you for volunteering arianepaola Jul 22 17:13:57 yes, thanks arianepaola. Jul 22 17:14:14 <_av500_> arianepaola: thanks Jul 22 17:14:49 <_av500_> apaar and neemo are up next week Jul 22 17:15:02 cool :) Jul 22 17:17:19 * nerdboy proud of the good job from both presenters today Jul 22 17:17:24 <_av500_> +111 Jul 22 17:17:31 <_av500_> 6) the end Jul 22 17:17:32 <_av500_> :) Jul 22 17:18:08 :-D Jul 22 17:18:51 * nerdboy been to way too many mtgs and can't hold back the comedy/sarcasm Jul 22 17:19:21 i could claim it all started recently but that would be a lie... Jul 22 17:20:38 moto-timo: arianepaola i still have active the osx branch Jul 22 17:21:03 https://github.com/ungureanuvladvictor/BBBlfs/tree/osx Jul 22 17:21:23 be sure to follow the brew instructions for libusb Jul 22 17:21:35 _av500_: will do Jul 22 17:21:44 alexanderhiam / neemo: gimme 10 minutes to make more coffee Jul 22 17:21:44 <_av500_> ok Jul 22 17:21:47 <_av500_> thx Jul 22 17:21:52 the usage instructions are outdated Jul 22 17:21:54 <_av500_> bye everybody, need to go Jul 22 17:21:54 urghh Jul 22 17:22:03 bye _av500_ Jul 22 17:22:03 vvu: thanks, brew is like macports, right? Jul 22 17:22:06 yes Jul 22 17:22:11 you have macports installed ? Jul 22 17:22:49 I use macports, but use brew if you can because that is what is on travis-ci Jul 22 17:22:49 nerdboy: roger Jul 22 17:23:01 hendersa: hi Jul 22 17:23:11 brew and macports do not play well together Jul 22 17:23:12 vvu: a friend will lend me a mac this weekend for a few weeks, I had used macports though before Jul 22 17:23:22 brew is easier :) Jul 22 17:23:29 vvu: +1 Jul 22 17:23:42 ankur: Hey ankur. Jul 22 17:23:45 +1 for homebrew Jul 22 17:23:54 hendersa: did you get my mail? Jul 22 17:23:57 * karki_ was busy with interviews :'( Jul 22 17:23:59 an offtopic question: anyone from here has experience with decoding convolution codes using a viterbi decoder ? Jul 22 17:24:14 ankur: I did and I answered it. I commented on your HAL issue on github, as well. Jul 22 17:24:28 vmayoral: Seems like I have to add more files to the vehicle folder to get the code compiling Jul 22 17:24:49 ronith: homebrew it is then :-) Jul 22 17:25:11 * nerdboy pushed a static makefile example from redis-ipc to beagle fork Jul 22 17:25:22 arianepaola: It's very easy to install and update Jul 22 17:25:24 https://github.com/sarnold/BeagleSat/blob/master/API/Makefile.example <= neemo take a look at that Jul 22 17:26:29 arianepaola: Good choice Jul 22 17:26:40 vmayoral: ping Jul 22 17:26:50 ds2: /me looking Jul 22 17:26:54 not that Jul 22 17:26:57 * neemo looking Jul 22 17:27:02 that Jul 22 17:27:22 also ds2: feel free to chime in on the float in kernel discussion, which should happen soon I think Jul 22 17:29:14 neemo: short summary - don't. Jul 22 17:29:24 ebadawy: still ehre? Jul 22 17:29:32 yup Jul 22 17:29:44 neemo: there are too many subtle ways to trip things up with FP in the kernel. Jul 22 17:30:03 up until ARMv7, there are different FP's in the arm world Jul 22 17:30:17 couple that with other arch that your code mightwant to be ported to.... Jul 22 17:30:41 along with potentially creating nonRT blocks due to added savings code Jul 22 17:30:57 ds2: I know, but current setup goes towards => do all the processing in kernel code, It'll be faster/better/(not tested in yet though) and I've read it can be done, albeit painfully Jul 22 17:30:58 and to top it off, doing that in less then 3weeks is a bit too much Jul 22 17:31:19 neemo: why do you think you NEED floating point? Jul 22 17:31:23 so currently at the, should I look for alternatives, or try to figure out how to pass the thorns Jul 22 17:31:27 hendersa: https://github.com/ankurayadav/demoapps-bbbandroid/blob/master/app07-can/src/com/bbbandroid/can/HardwareTask.java#L47 Jul 22 17:31:40 my java code is getting hang here Jul 22 17:31:40 nerdboy: alexanderhiam ^ FP stuff Jul 22 17:31:46 https://github.com/ankurayadav/demoapps-bbbandroid/blob/master/app07-can/jni/jni_wrapper.c#L478 Jul 22 17:31:57 neemo: explain what you are you doing that needs floating point? are you covering some insanely large dynamic range? Jul 22 17:32:07 like over 300dB? Jul 22 17:32:10 ds2: well, I don't strictly think i NEED FP in Kernel Jul 22 17:32:14 no Jul 22 17:32:22 then explain why Jul 22 17:32:32 ebadawy: so how I was telling you, if you run a code using the bonescript library right now Jul 22 17:32:38 it runs infintelu Jul 22 17:32:42 cuz there are simple techniques Jul 22 17:32:52 infinite Jul 22 17:32:57 but it depends on what the reason you are needing it Jul 22 17:33:00 I should use FP, since the values I'll be getting will be vary in size quite a bit in nT, or so I expect Jul 22 17:33:04 * rohith searches for vmayoral Jul 22 17:33:15 the only way of stopping is, refreshing the page Jul 22 17:33:25 the idea, is to have a stop button Jul 22 17:33:26 :) Jul 22 17:33:31 neemo: how much do they vary? aren't you ultimately limited by your sensor resolution, say 16bit or less? Jul 22 17:33:50 neemo: consider working with scaled _FIXED_ point values Jul 22 17:33:52 jkridner: have something else to add to my simple explanation of the run button Jul 22 17:33:54 though as nerdboy mentioned before, I can go for fixed point math I guess Jul 22 17:34:02 run and stop button Jul 22 17:34:07 yeah scaled fixed is one of the alternative roads Jul 22 17:34:15 i.e. multiple everything by 64K and store values in a 32bit integer Jul 22 17:34:19 I see that jkridner have added the disconnect, that won't do the job? Jul 22 17:34:22 and yes, 16bit is the resolution of both sensors, so that would be the limit for now Jul 22 17:34:37 floating point is really only needed if you have huge dynamic ranges to cover Jul 22 17:35:07 i'm talking about the bonescript library Jul 22 17:35:25 yeah, I know Jul 22 17:35:25 ds2: I'll be doing multiple divisions, multis with matrices Jul 22 17:35:28 I have had good luck doing sensor math using fixed point....picking proper scaling values is where the work is at Jul 22 17:35:48 ok, so scaling with FIXED is a possible alternative Jul 22 17:35:55 neemo: worse case - 64bit integers scaled by 32bit Jul 22 17:35:56 that will work in kernel space at least Jul 22 17:36:09 it doesn't you can tried it Jul 22 17:36:16 division can be a little tricky but not impossible Jul 22 17:36:32 look at the ARM CMSIS-DSP lib for ideas on the math Jul 22 17:36:43 well, multi/div, and trigonometric funcs are a must Jul 22 17:36:55 other stuff... I think I can work around Jul 22 17:37:06 ankur: Is that a blocking read on that socket? Jul 22 17:37:06 why do you need trig? Jul 22 17:37:06 DiegoTc: yeah, you are right, just tried it. Jul 22 17:37:21 that's basically Jul 22 17:37:22 yes Jul 22 17:37:26 neemo: are you generating euler angles from the sensors? Jul 22 17:37:36 i remember i did it worked, but have to check Jul 22 17:37:42 I don't remember what i did Jul 22 17:37:48 its working fine when compiled directly in c program Jul 22 17:37:50 jajja Jul 22 17:37:52 crazy solutions Jul 22 17:37:54 ankur: You should perform a select() on the file descriptor and then read if there is data present. Jul 22 17:38:11 ds2: in the simplest form for fixing, I can precompute the conversions (equations) between my intermediate values and my final correction values Jul 22 17:38:31 ds2: but those equations are bound to have some trig functions expressed within Jul 22 17:38:36 ankur: select() will let you peek at the fd and see if there is data available. Jul 22 17:38:38 hendersa: ok so with read timeout Jul 22 17:38:43 ankur: Yes. Jul 22 17:38:54 ok thanks Jul 22 17:39:00 neemo: not always... there might be a complex representation that lets you do it with matrix math Jul 22 17:39:03 ebadawy: but let's worry of the integration first Jul 22 17:39:04 mmmm, so you are saying that I should give it a hight prioirity Jul 22 17:39:13 first the BBUI Jul 22 17:39:13 ankur: Treat the CAN bus just like a network socket. Jul 22 17:39:27 I think that I can finish up with BBUI this week Jul 22 17:39:30 neemo: from the stuff i have see, only final reductions need the trig and that can be done in userspace Jul 22 17:39:41 hendersa: one more thing I am testing can with vcan Jul 22 17:39:45 ds2: so the part that's supposed to work live in kernel is fitting (which is pure matrix math) and the actual conversion between what the matrix gives out (intermediates) and final correction params (ones I like to have) Jul 22 17:39:54 cool, I will be asking you on that, are you going to push on jkridner rep this week also? Jul 22 17:39:56 ds2: that's for the get my correction params Jul 22 17:40:13 can i connect two beagles directly over can Jul 22 17:40:31 sure. as soon as I finish up with it, I'll make the patch ready Jul 22 17:40:39 ds2: to do corrections, I'll need some trig as well, since that's how the raw -> corrected data points are expressed Jul 22 17:41:20 neemo: hmmm... what about preconvert the trig to something else? Jul 22 17:41:33 ds2: I could do those to parts in userspace, but the idea was to do it in kernel as much as possible. Could drop that maybe? <= and ^ nerdboy & alexanderhiam Jul 22 17:42:03 many of these things have polar and non polar representations... the polar ones tend to want trig. Jul 22 17:42:09 ankur: CAN can be multi-master, but I have not tried it that way. I tend to leave the BBB as the master for whatever bus I am using to generate the clock and have everything else as a slave. Jul 22 17:42:30 neemo: catch me later... I need to go right now. be happy to discuss more Jul 22 17:42:42 (1042 here, need to get day started) Jul 22 17:42:44 ds2: I can look into it, not sure what I'll get. the current setup is just expressing one axis as reference and calibrating others based on that one, in this case the X axis Jul 22 17:42:51 ankur: You can try connecting two BBBs on the same CAN bus and have them talk, but the troubleshooting involved may eat up more of your time than it is worth. Jul 22 17:43:02 ds2: sure, thanks for the chat Jul 22 17:43:16 happy to carry on later Jul 22 17:43:18 hendersa: any hardware you suggest for testing can or vcan testing is fine? Jul 22 17:43:23 ds2: also, have a good day :D Jul 22 17:43:47 neemo: do the terms euler angles and quaterions ring bells? Jul 22 17:44:38 ankur: Work with vcan for now, since we're just providing a simple interface to people using their own CAN code for OBD stuff. As long as your socket comms work, you've given people the low-level code that they can build on. Jul 22 17:45:21 ok Jul 22 17:48:23 ankur: I'm going to get back to work, but if you have any questions, send me a mail and I'll get you pointed in the right direction. Jul 22 17:48:30 nerdboy: caffeine supply on-board? Jul 22 17:48:53 hendersa: sure :) Jul 22 17:49:03 ankur: Keep up the good work! Jul 22 17:49:17 :D Jul 22 17:51:14 * nerdboy reading backscroll... Jul 22 17:52:12 should work fine Jul 22 17:52:54 you can even use vfp for that i think Jul 22 17:53:06 nerdboy: sorry for the text wall :) Jul 22 17:53:21 vfp -> virtual floating point? Jul 22 17:53:53 vvu: any feedback on done features? I need some feedback in case we want to change some approach or something Jul 22 17:55:50 vector processor on arm Jul 22 17:56:15 but that would be an optimization for later... Jul 22 17:56:55 yeah, understood Jul 22 17:57:01 * neemo adds vfp to bucket list Jul 22 17:57:46 but what's the current status on FP in kernel then? Should I avoid it all together? Jul 22 17:58:02 what ds2 said Jul 22 17:58:09 fixed point math? Jul 22 17:58:37 yup Jul 22 17:58:38 I've did reading on it before, going to catch up on that Jul 22 17:58:55 I've even found some pretty good trig solutions, if i remember correctly Jul 22 17:59:18 * nerdboy was thinking of python for quick user-space framework Jul 22 17:59:28 but ok, I'll move onto that, and stop fiddling with compiling my LKM with float support Jul 22 17:59:46 the whole quaternion thing is already done Jul 22 18:00:24 * nerdboy would assume the nav stuff already has that Jul 22 18:00:53 python framework? Jul 22 18:01:25 multiple reference frames/coordinate transformations, etc Jul 22 18:01:56 python for first cut at userspace sensor code Jul 22 18:02:42 should work Jul 22 18:02:57 spi should have interrupt pin for timing, etc, and would be the bottleneck for data acq anyway Jul 22 18:03:19 but we're aiming for a C driver in the end, no? Jul 22 18:03:55 with sysfs hooks Jul 22 18:05:06 only if necessary for performance/other reqs Jul 22 18:05:21 bbio python interface is already there Jul 22 18:05:58 makes sense Jul 22 18:06:07 cython would be option as well Jul 22 18:06:21 *next performant option Jul 22 18:07:14 time is running out and python is your best bet for cranking out useful code i think Jul 22 18:07:24 * neemo bucket list +1 Jul 22 18:07:33 yeah, makes sense Jul 22 18:07:44 I've parsed through the docs you sent me this morning Jul 22 18:08:01 and yeah, I'll start working on tidying up the API and splitting it properly Jul 22 18:08:28 tomorrow morning is the scheduled window for that, then I'll ping you and alexanderhiam for more feedback what I should change/keep Jul 22 18:09:00 also, I'll have to edit the Makefile, to work with splitting. Jul 22 18:09:28 nerdboy: What cues was I supposed to pick up from the Makefile.example you linked? Jul 22 18:11:50 pretty much all of it Jul 22 18:12:14 not all of it is needed right away Jul 22 18:12:54 ok Jul 22 18:13:06 so I should splice the one I have to be more like the one you linked? Jul 22 18:13:09 even that one isn't perfect but it works for manual native/cross and shows some mak.inc stuff for tools like doxygen, etc Jul 22 18:13:20 yeah Jul 22 18:13:42 technically the one I have up right now, was set up on the fly from a C/P I found somewhere Jul 22 18:13:53 just to have src/obj/bin separation Jul 22 18:14:21 the only part I really edited to work was the make doxy instructions to make building the doxygen files a breeze Jul 22 18:14:32 there are a lot of suspect makefiles out there... Jul 22 18:14:54 few people really use it well anymore Jul 22 18:15:44 I agree with that Jul 22 18:15:57 not that I'm an expert or anything, I get by Jul 22 18:16:17 * neemo should order that book which is sitting in my wishlist to fix that Jul 22 18:16:21 i prefer not maintaining static makefiles if i can help it Jul 22 18:16:55 yeah Jul 22 18:16:59 just try swapping out the bits that you need now Jul 22 18:17:10 Ok, so got orders, fix stuff, work on API, report back tomorrow? Jul 22 18:17:20 header/source separation Jul 22 18:17:23 catch alexanderhiam here as well hopefully Jul 22 18:17:26 yup Jul 22 18:17:29 that the work on API part Jul 22 18:17:31 separation Jul 22 18:17:38 and tidying up the functions to make sense Jul 22 18:17:39 maybe he got sucked into a mtg... Jul 22 18:17:55 nerdboy: I should probably avoid returining the triplet right? Jul 22 18:18:03 triplet struct* Jul 22 18:18:06 try pushing some stuff on a WIP branch today Jul 22 18:18:16 yo Jul 22 18:18:25 * alexanderhiam had to play sysadmin for a bit, back now Jul 22 18:18:41 +1 to python for device drivers Jul 22 18:18:44 It looked very cool from a usability point of view, but it's probably terible as a return idea Jul 22 18:18:52 alexanderhiam: 'ello Jul 22 18:18:57 there's just so little time at this point Jul 22 18:18:58 boss pinging me, bbiaf Jul 22 18:19:10 sysadmin is cool, <3 Linux Jul 22 18:19:30 * neemo is prepping for a sysadmin interview in the near future Jul 22 18:20:07 if we go with Python you can use PyBBIO and I could potentially merge in device drivers Jul 22 18:20:14 alexanderhiam: well, it's not like I'm gonna abruptly stop working on BeagleSat when the GSoC deadline hits Jul 22 18:20:34 I mean, something presentable should be up by the end of GSoC, because that's the point of the program Jul 22 18:20:45 well your pass/fail status depends on the work you do before it ends Jul 22 18:21:12 but I did envision working on it afterwards as well Jul 22 18:21:37 true that, I'll do my best to be well on the pass side Jul 22 18:22:42 alexanderhiam: comments on the (maybe not sane) ideas in the API header Jul 22 18:22:59 more API ideas, then header currently, but hey Jul 22 18:23:15 they're in the /// @todo tags Jul 22 18:26:12 * neemo cries a bit while re-reading all the errors made in this IRC Jul 22 18:26:23 have you done any python before? Jul 22 18:26:27 * neemo 's English sucks at 2 am Jul 22 18:26:34 yes Jul 22 18:26:54 not a huge amount, but I know my way around it Jul 22 18:26:58 I'm leaning towards doing all the userspace stuff in Python Jul 22 18:27:35 not this type of stuff, but open to working on it Jul 22 18:28:20 plenty of sensor examples, they're mostly specific to each device Jul 22 18:28:39 yeah, makes sense Jul 22 18:29:02 saw some code, tested some for these sensors while I was getting readouts Jul 22 18:29:51 some basic test code for a few sensors on github, except it was written against rpi.gpio, etc Jul 22 18:30:06 so current setup is C API, processing in kernel with fixed point math, and getting readouts via Python? Jul 22 18:30:10 *i mean in one of my repos Jul 22 18:30:29 ton of sensor on github including all the Adafruit stuff Jul 22 18:30:29 * neemo does mental stretches to stay awake and focused Jul 22 18:30:37 *sensor shit even Jul 22 18:30:41 yup, went through those Jul 22 18:30:42 pybbio has eveything you need: https://github.com/graycatlabs/PyBBIO Jul 22 18:30:54 thanks alexanderhiam Jul 22 18:31:08 including some device drivers: https://github.com/graycatlabs/PyBBIO/tree/master/bbio/libraries Jul 22 18:31:23 test code would use same-ish interfaces on bbb too Jul 22 18:31:48 https://github.com/VCTLabs/pi-sensor-test Jul 22 18:32:36 sounds like a plan Jul 22 18:32:56 the code is gonna be a mish-mash though, but I'll work on making it right Jul 22 18:33:19 nerdboy: alexanderhiam you guys gonna be here in the evening hours (your time)? Jul 22 18:33:35 what do you mean it's gonna be a mish-mash? Jul 22 18:34:22 yeah, i would not put that in the design plan... Jul 22 18:34:49 alexanderhiam: well, it should be a mish mash Jul 22 18:35:08 the API is messy now, but will work on that from tomorrow Jul 22 18:35:14 should => shouldn't Jul 22 18:35:21 -.- Jul 22 18:35:28 using the bbio interface to talk to your magnetometer? Jul 22 18:35:50 and I'll go through BBIO tomorrow as well to see how to interface it to the magnetometer Jul 22 18:36:01 haven't used that yet, but I guess it should be straightforward Jul 22 18:36:14 no Jul 22 18:36:14 ? Jul 22 18:36:29 go ahead and clean up your c interface as long as it's fresh-ish Jul 22 18:36:47 i would switch to python soon tho... Jul 22 18:37:06 ok Jul 22 18:37:27 should be quite a bit more productive if you don't worry about python optimization, etc Jul 22 18:37:35 well, do we want to drop the C alltogether for now? Jul 22 18:37:48 good question Jul 22 18:38:11 could just park it on its own branch for now Jul 22 18:38:26 and do the API in python Jul 22 18:38:29 well, I'd vote for no drop, since I've invested time in it and want to take it through Jul 22 18:38:43 time in learning how to do it that is Jul 22 18:38:55 would not hurt to update it as you go if time is there... Jul 22 18:39:06 well, it doesn't really make sense to do device drivers in Python if the API is in C Jul 22 18:39:07 but if you think python would be more productive for GSoC, I'm ok with that Jul 22 18:39:28 it'll be easy to move parts to C later on Jul 22 18:39:37 for the project you're probably right Jul 22 18:39:47 alexanderhiam: yeah, that's why I thought about it a bit mish-mashy Jul 22 18:39:56 park what you have for now Jul 22 18:41:21 so, step 1: PyBBIO device drivers (try to write them like the others so they can be merged). 2: create a BeagleSat Python package that uses those drivers and provides a nice API Jul 22 18:41:34 the filtering could be done in Python for now as well Jul 22 18:42:06 python can interface to other stuff later Jul 22 18:42:58 later/last step might be to check SciPy for nonlinear solvers Jul 22 18:43:07 good thinking Jul 22 18:43:19 there should be some Jul 22 18:43:37 * nerdboy has earlier c/matlab package in back pocket if needed Jul 22 18:44:02 * neemo bucket list +1 more Jul 22 18:44:24 well, I'm not really happy with leaving C behind Jul 22 18:44:31 you're not throwing everything out, you can use the same object model and API concepts in Python Jul 22 18:44:35 but I guess I can come back to it later Jul 22 18:44:38 just remember to sleep/eat/bathe once in a while between coding sessions... Jul 22 18:44:42 yeah Jul 22 18:44:46 Python's implemented in C, so you're not leaving it behind! Jul 22 18:44:55 cython Jul 22 18:45:01 cython ftw Jul 22 18:45:04 *for later Jul 22 18:45:17 no intentions of leaving it unfinished anyway Jul 22 18:45:33 alright, then pyBBIO first thing in the morning Jul 22 18:45:55 use the python interface and get some data passed around Jul 22 18:45:56 after that see for a basic BeagleSat Python interface? Jul 22 18:46:46 that's the plan, right? Jul 22 18:46:50 yup Jul 22 18:46:55 ok Jul 22 18:47:12 there's plenty of code in PyBBIO you can use as reference Jul 22 18:47:12 I'm gonna leave you gents and go count some sheep Jul 22 18:47:18 will do Jul 22 18:47:29 the device driver should be simple, just use the registers you need Jul 22 18:47:32 drop a link or two so I can have my reference points Jul 22 18:47:39 if it's not too much to ask :) Jul 22 18:48:22 * neemo accepts link even when not awake (wink) (wink) Jul 22 18:48:43 anyway, got my orders, and I'm off to bed Jul 22 18:48:49 gn Jul 22 18:49:16 BTW I'll try to force the solver with Mathematica, once the darn thing finishes downloading Jul 22 18:49:32 * neemo looking at for two weeks now Jul 22 18:49:40 * neemo looking at it for two weeks now Jul 22 18:49:41 * nerdboy still not sure how to "condition" people that "try it and see" won't make anything disappear into a black hole... Jul 22 18:49:58 most likely not anyway Jul 22 18:50:10 lol Jul 22 18:50:19 nor will it break the keyboard Jul 22 18:50:20 well, I'm on the trying path Jul 22 18:50:25 just need the tools :) Jul 22 18:50:47 gn all Jul 22 18:50:52 can be a little more careful when you're wiring things... Jul 22 18:51:52 * nerdboy would not worry about mathematica or touch it probably Jul 22 18:52:30 the universe will provide a solver at the appropriate time and not before Jul 22 18:55:42 * nerdboy shaking his head and wagging his finger Jul 22 22:25:48 back Jul 22 22:41:24 massage? Jul 22 22:41:52 * nerdboy has a stiff neck already... **** ENDING LOGGING AT Thu Jul 23 02:59:59 2015