**** BEGIN LOGGING AT Tue Mar 01 02:59:58 2016 Mar 01 08:44:10 Hello Mar 01 08:44:10 can someone guide me for GSOC 2016? Mar 01 09:22:46 hello , can someone help me get going with projects on arduino , autopilot and SPI & Sensor support for Python. Many thanks Mar 01 09:24:34 brateek: please sign up on the mailing list Mar 01 09:24:39 and introduce yourself there Mar 01 09:27:07 sure Mar 01 09:27:49 can anyone give me the architecture of bb blue? Mar 01 09:29:07 similar to black Mar 01 09:33:16 av500: I mean to say the names of the sensors and so on.. Mar 01 09:51:35 Hello Mar 01 09:52:43 Hey!! Mar 01 09:53:29 Can you please tell me do we need to work on fresh projects or existing projects only? Mar 01 10:26:26 Somebody please help me in finding a mentor. Mar 01 10:35:59 can anyone tell me which IMU is the beaglebone blue using? Mar 01 14:16:59 ds2,Just wanted to ask that project demands that the PRU be exposed in the slave mode?I think it is mentioned that it be exposed as a master interface? Mar 01 14:17:40 .. Mar 01 14:17:48 ds2, And I am really sorry for my delayed response.I have my exams going on. Mar 01 15:54:26 jkridner: Can you please tell me which IMU bb blue is using? Mar 01 16:18:50 What is the expected price of the BB Blue? Mar 01 16:19:13 jkridner ^ Mar 01 16:19:42 Not sure yet, but I'd budget less than $80. Mar 01 16:19:56 MPU6150 I think. let me check. Mar 01 16:20:16 I guess it should be MPU9150 Mar 01 16:20:21 MPU-9250 Mar 01 16:20:39 and BMP280 for barometer. Mar 01 16:21:16 Use the Strawson Robotics cape for comparison... few updates from that. Mar 01 16:25:09 Hi Mar 01 16:29:03 I hope this question hasn't been answered somewhere else: but is there some hardware for students that successfull applied for GSoC? Mar 01 16:29:13 Like I'm interesseted about DSP Stuff on the X15 Mar 01 16:30:18 If you are successfully admitted to GSoC - the org will send you all the necessary hardware. Mar 01 16:32:08 sounds good :) Mar 01 16:40:53 jkridner: so basically bb blue = bb black+robotics cape? Mar 01 16:42:07 would an hello world on RPi2 via UART be appropiate? Mar 01 16:42:43 (Bare Metal Programm) Mar 01 16:43:29 Or should it be programm running on an arm linux? Mar 01 16:44:19 oh there is more in the READMe.md in gsoc-application repo Mar 01 16:54:02 evening all Mar 01 16:55:56 hey anujdeshpande Mar 01 16:57:03 hola Abhishek_ ! Mar 01 17:08:33 jkridner, alexanderhiam : I'd like to mentor the "BeagleBone Blue API" project section. Seems in line with my interests. Mar 01 17:10:39 Hello everyone Mar 01 17:12:16 I am very exited for the project mentioned in ideas page. Implement stereo/optical flow using image processing in ardupilot Mar 01 17:13:13 It would be very grateful if someone can review it on the Google group page Mar 01 17:13:24 And give me suggestions fr it Mar 01 17:15:05 It goes with the name - interested in a PRU related project Mar 01 17:16:09 Thank in advance Mar 01 17:19:04 Kartik: I believe that's nerdboy's idea - he's on the US west coast so it's still on the early side, but he should be active here soon Mar 01 17:20:06 Thanks for the reply sir.. I'll be waiting then Mar 01 17:20:22 Kartik: one note - you certainly won't be able to port opencv to the PRU Mar 01 17:21:03 not sure what parts of the image processing make sense to offload to the PRU, but they would likely need to be implemented in some well optimized assembly Mar 01 17:21:31 alexhiam: So is bb blue = bb black + robotics cape? Mar 01 17:21:46 Yes sir.. But data input from pru and later image processing using a Linux based app can work Mar 01 17:21:58 kiran4399: that's my understanding Mar 01 17:23:26 alexhiam: Is it possible to do both bb blue api's and ardupilot interfaces and standalone ros for as a single project for this gsoc? Mar 01 17:23:35 Kartik: but is there any need to port the camera interface to the PRU? Can't the images already be read from the sensor at high rates by the Linux kernel? Seems to me like that would be reinventing the wheel unnecessarily Mar 01 17:24:17 kiran4399: we'd need to see a very well documented plan with a detailed timeline to be convinced it could all be done Mar 01 17:24:36 Right sir. Actually in. On Mar 01 17:25:27 alexhiam: According to you, do you think it can be done? :-P Mar 01 17:25:40 kiran4399: I think the open question now is how much kernel driver work is there to be done - once we have a good plan for that we'll have a better idea of how much time will be left Mar 01 17:25:44 i think the stereo image processing stuff should be doable Mar 01 17:26:08 kiran4399: but yeah, I suspect the ports should be pretty simple once the drivers are all in place Mar 01 17:26:11 optical flow calculations probably not... Mar 01 17:26:37 Kartik: the point of offloading some of the image processing tasks to the PRU is to more or less get the slow stuff done faster Mar 01 17:26:58 Kartik: and it's the processing, not the data retrieval, that's slow Mar 01 17:27:17 alexhiam Just wanted to ask that the idea regarding PRU being exposed as I2C being done in slave mode,it meant that PRU would function as a slave right? Mar 01 17:27:46 nerdboy: what parts of that are you picturing being done in the PRU? Mar 01 17:28:41 chanakya_vc: yeah - thus allowing the BeagleBone w/Linux to be an I2C slave device Mar 01 17:31:26 nerdboy: like the PRU generates a depth map from two images passed to it from the ARM? Mar 01 17:31:57 alexhiam, But then on the ideas page it says that it should serve as the Bbitbang I2C master interface? I am a bit confused. Mar 01 17:32:30 chanakya_vc: oh, well, supporting both would be best Mar 01 17:34:56 * mr_science just had network drop Mar 01 17:35:18 alexhiam: So basically we have to modify the strawson api's and merge it with the bb black kernel to make a bb blue kernel.. right? =) Mar 01 17:38:06 kiran4399: well, the ported Strawson APIs would still live in userland. Step 1 would be to get together kernel drivers for all the devices on the blue. Then port the APIs (packaged in a .deb ideally). A recipe for a blue image with all the drivers included would be a good end goal Mar 01 17:39:35 alexhiam: but when we get together kernel drivers for all devices on the blue.. why should we require those api's? Mar 01 17:40:03 kiran4399: so knowing ahead of time how to build the kernel (https://github.com/beagleboard/linux) and images (https://github.com/beagleboard/image-builder) would be ideal Mar 01 17:41:06 alexhiam: strawson api's could only be used with the robotics cape right? Mar 01 17:41:38 kiran4399: the kernel drivers will provide sysfs entries. Having a port of Strawson C API that uses those sysfs entries instead of the current mmap approach would mean that software that currently uses it with the cape would be easily ported to the blue Mar 01 17:42:26 alexhiam, ds2 , And regarding the issue about how critical timing is in slave mode, I understand that here we are referring to the obvious speed advantage PRU gives over other slave devices? Mar 01 17:43:35 alexhiam: One more question.... do we still need to use the robotics cape on beaglebone blue? Mar 01 17:43:44 chanakya_vc: what do you mean by "over other slave devices"? The question is what are the potential issued with I2C slave mode using Linux Mar 01 17:44:13 kiran4399: no.. the blue is basically the black and the robotics cape on the same board Mar 01 17:44:52 chanakya_vc: *issues* that is, and I mean related to timing specifically Mar 01 17:48:48 alexhiam: as expected.. I got the whole picture.. Mar 01 17:51:42 alexhiam, Okay. Will research a on this. Apart from this,I think that this project would involve modifying the remoteproc? Mar 01 17:52:31 Hello every one Mar 01 17:54:00 my name is Bharat Bhushan from Punjab, India . i have some knowledge of Arduino, raspberry pi. any body can tell me how can i contribute with you. from where i have to start Mar 01 17:54:35 chanakya_vc: yeah, remoteproc would be a good base for the driver Mar 01 17:55:55 alexhiam, So if I got this right,Basically we write the code for bitbanging into the PRU firmware,modify the remote proc?Anyhting that I have missed? Mar 01 17:55:59 Bharat: see https://groups.google.com/forum/#!topic/beagleboard-gsoc/zTcTgXNe0nU and http://elinux.org/BeagleBoard/GSoC/Ideas Mar 01 17:56:57 Hi, I would like to talk about the project and look for a mentor? Mar 01 17:57:19 chanakya_vc: well, "modify the remoteproc" isn't how I would put it... you would use remoteproc to develop a kernel driver that interfaces with the PRU firmware and exposes itself as an I2C device to userland Mar 01 17:59:28 Ok Mar 01 17:59:38 Thank you Mar 01 18:01:29 alexhiam, So the code for bitbanging would actually be in this driver, right?So I should as of now start investigating the remoteproc driver? Mar 01 18:03:12 alexhiam: the Berkeley guys got 30 fps capture on black Mar 01 18:04:43 processing with opencv brings it down to less than 19 fps Mar 01 18:05:16 chanakya_vc: no.. the bitbanging would be on the PRU. The driver would be to configure it, send data back and forth to userland, etc.. Bitbanging in a kernel driver goes back to the timing question Mar 01 18:05:37 that's with a load of optimization tricks Mar 01 18:06:30 nerdboy: so, streaming stereo images on the ARM to shared memory, PRU processes them and streams a depth map to shared memory? Mar 01 18:07:02 sounds reasonable Mar 01 18:07:36 afaict these guys were only using cpu/fpu and software optimazation Mar 01 18:08:03 nerdboy: would fixed point be suitable? I don't really know anything about stereo processing... Mar 01 18:08:37 needs a bit of analysis like any good problem... Mar 01 18:09:01 I guess that just comes down to accuracy/resolution Mar 01 18:09:23 fixed point certainly would work, I guess the question is would it lead to crashes :P Mar 01 18:10:08 what is BeagleBone Blue APIs ? Mar 01 18:10:11 http://www.eecs.berkeley.edu/Pubs/TechRpts/2014/EECS-2014-117.pdf Mar 01 18:11:04 alexhiam, Thanks.Will investigate and get back to you.Apart from this anything else that you would suggest I read about? Mar 01 18:11:49 chanakya_vc: having a solid understanding of all the serial protocols you would want to implement is number one Mar 01 18:13:08 Bharat: the BeagleBone Blue is a BeagleBone Black and this cape merged onto one board: www.strawsondesign.com/#!board-features Mar 01 18:13:26 Bharat: the devs of that cape made a C API for it: https://github.com/StrawsonDesign/Robotics_Cape_Installer Mar 01 18:15:20 alexhiam, Okay so I will read about SPI,I2C and UART. And also read about bitbanging in a kernel driver. Mar 01 18:15:40 alexhiam, And get back to you. Mar 01 18:16:03 http://arxiv.org/pdf/1412.6153.pdf <= robot avoidance version using stereo image processnig Mar 01 18:16:43 alexhiam: i will read it, Mar 01 18:17:40 that could make a nice autopilot... Mar 01 18:17:50 bradfa what do you think about the project SPI Flash emulator, and if you want to be a mentor Mar 01 18:18:49 nerdboy: awesome. That looks like a good starting point for sure Mar 01 18:21:00 alexhiam, Just one last question,would you be the mentor of this project? Mar 01 18:21:41 chanakya_vc: co-mentor possibly. Depends on the other projects as well Mar 01 18:22:18 alexhiam, Okay.Thanks for your help!! Mar 01 18:26:12 alexhiam: we can talk about the project? I have a few questions. Mar 01 18:27:05 the lack of response from other mentors Mar 01 18:28:26 * nerdboy heads for espresso machine... Mar 01 18:28:40 somebody could put a beaglebone on *that* Mar 01 18:29:07 pmezydlo: well if you are going to be mean, i'm sure they are going to lack responses :) Mar 01 18:29:07 pmezydlo: the spi flash emulator is my idea, so I think it's a good one :) Mar 01 18:30:20 pmezydlo: what lack of response of other potential mentors are you asking about? Mar 01 18:31:11 pmezydlo: keep in mind that the people in here are in many time zones - everyone's not always at their keyboards waiting for someone to ask a question... patience is good on irc ;) Mar 01 18:31:48 pmezydlo: and many of the people who are volunteering to be mentors have full time jobs and families, so please be patient Mar 01 18:32:16 pmezydlo (and everyone else who's new to irc): https://workaround.org/getting-help-on-irc/ Mar 01 18:34:12 * alexhiam goes afk for a bit... Mar 01 18:34:24 i've written my ideas about the project on the mailing list. I would like somebody to comment on it Mar 01 18:38:58 pmezydlo: I am responding to you now Mar 01 18:41:55 bradfa: thanks, most complicated is the same driver on pc whether it is necessary Mar 01 18:46:07 \clear Mar 01 18:51:09 pmezydlo: if you find that you are not getting a response via email to something sent to the gsoc beagleboard list, please don't hesitate to ask on IRC for some attention or simply reply to your last mail on the list so that people are more likely to see it (some people get a LOT of email) Mar 01 18:51:22 @mentors I think it would be cool to 'open source' some of the applications by previous participants. Help the present batch to approach projects with the right attitude ? Mar 01 18:51:48 anujdeshpande: like take existing submissions from melange and make them public? Mar 01 18:52:20 bradfa: melange afaik only shows an excerpt of the selected application proposal Mar 01 18:53:02 anujdeshpande: ok, do you have the existing applications? or would this be something jkridner would need to help with? Mar 01 18:53:54 bradfa: i think jkridner would have to figure something out. especially since we have now moved away from melange Mar 01 18:54:17 I don't know if we can still go back and access the entire application proposal Mar 01 18:55:01 Even if the mentors who have been around for a bit, recollect one or two applications which were outstanding, we could drop an email/ping on IRC to ask if the former-students still have it Mar 01 19:02:19 stupid proxy... Mar 01 21:44:34 I don't understand that basic requirement on the page - do I have to do these things before applying for GSoC or after I'm accepted by google Mar 01 21:45:22 hpfmn: which requirements page? Mar 01 21:45:40 alexhiam: http://elinux.org/BeagleBoard/GSoC/Ideas#General_requirements Mar 01 21:45:52 I mean the first bullet point says "once accepted" Mar 01 21:46:06 but some things sound like they should happen before - like point 8 Mar 01 21:46:12 yeah, that's BeagleBoard.org's requirements for a gsoc project Mar 01 21:46:28 yeah, number 8 comes before acceptance Mar 01 21:46:37 I guess it's not in chronological order... Mar 01 21:47:40 btw. is point 6 like scrum? Mar 01 21:49:01 we don't do agile - more of just a status report before our weekly irc meeting so we all know where everyone is at, and so we can see right away if a student is getting stuck/falling behind/etc. Mar 01 21:57:35 Hey ! i was trying to work with PRU-framework and i tried the make process but the make file at location (on my BBB) /usr/src/linux-headers-4.1.17-ti-rt-r48/Makefile has got some serious errors. Mar 01 21:58:21 ZeekHuge, what errors? btw, did you install "linux-headers-`uname -r`" ? Mar 01 22:02:09 rcn-ee, http://pastebin.com/JhSjLf7B this is the error, when i try to make remoteproc drivers. Mar 01 22:02:40 yes i already updated the linux-headers Mar 01 22:03:41 encoding? Mar 01 22:04:09 ZeekHuge, run dos2unix on ~/Projects/PRUProjects/PRU-framework/drivers/remoteproc Mar 01 22:04:57 i tried to cat the makefile at that location, but it wont show anything and nano showed repeating this --> "^@" Mar 01 22:04:59 ohk Mar 01 22:09:04 ZeekHuge, looking at: https://github.com/shubhi1407/PRU-framework/blob/master/drivers/remoteproc/Makefile and your error, re-git-clone... Mar 01 22:09:46 ohk, btw, that "dos2unix" didnt solve it. Mar 01 22:10:36 it was a guess, but now i found what project your cloned.. Mar 01 22:18:12 rcn-ee, still that same error. Mar 01 22:19:41 but why i am not able to see the contents of /usr/src/linux-headers-4.1.17-ti-rt-r48/Makefile ? Mar 01 22:23:47 ZeekHuge, works here: https://paste.debian.net/410796/ (from: cat /usr/src/linux-headers-4.1.18-ti-rt-r49/Makefile | pastebinit) Mar 01 22:25:14 ZeekHuge, works as advertised: https://paste.debian.net/410798/ Mar 01 22:29:50 hmmmmmmm Mar 01 22:30:00 what's the problem with remoteproc on PRU? Mar 01 22:30:16 ohk, While upgrading BBB, it got stuck on some npm package, ahh for 2-3 hours, so i removed the package. Can that be the reason ? Mar 01 22:30:24 blah node Mar 01 22:30:30 * ds2 goes back into the hole Mar 01 22:31:15 ZeekHuge: seems like the linux-headers package must have got corrupt, try removing and installing it again Mar 01 22:31:45 ZeekHuge, ^ and it'll take time to uncompress, linux-header has lots of files... Mar 01 22:33:34 ohk, will do. But just one question (it may be a really foolish question ), why am i not able to see the contents of the Makefile at /usr/src/linux-headers-4.1.17-ti-rt-r48/Makefile ? Mar 01 22:33:57 ZeekHuge, corruption, could explain it.. Mar 01 22:34:31 just one file ? Ohk I'll reinstall it . Mar 01 22:34:37 Thank you :) Mar 01 22:35:47 is there any difference between qemu-arm and qemu-arm-static? Mar 01 22:36:02 hpfmn, yeapers... Mar 01 22:36:52 hmm sadly there is no package in archlinux Mar 01 22:37:35 hpfmn, https://www.archlinux.org/packages/?name=qemu-arch-extra Mar 01 22:38:31 yeah that ships qemu-arm but not qemu-arm-static Mar 01 22:39:00 'static' package was built statically... Mar 01 22:39:33 that's the only difference.. ;) Mar 01 22:39:40 ohh well Mar 01 22:41:06 so it doesn't matter for me ;) Mar 01 22:48:19 will static packages get by ESD protection? Mar 01 22:49:37 it's qemu, even the static version doesn't have enough protection to protect against random bit changes... Mar 01 22:53:34 the google group is not usable as a mailing list, isn't it? Mar 01 22:54:10 hpfmn, it works perfectly fine, if you don't use it as "google group" just use email: beagleboard@googlegroups.com ;0 Mar 01 22:54:47 rcn-ee: but where do i subsscribe? Mar 01 22:55:21 rcn-ee, That worked ! thank you :) Mar 01 22:56:38 hpfmn, https://groups.google.com/forum/#!forum/beagleboard/join Mar 01 23:08:11 for some reason that gives me a never ending "loading" screen but you can subscribe by emailing beagleboard+subscribe@googlegroups.com Mar 01 23:09:45 but thank you rcn-ee for your kind help :) Mar 01 23:43:29 do we have a wiki page/other with bb dev tools/image descriptions? Mar 01 23:44:09 as in "do we want students poking around on their own or should we point them to specifics?" Mar 01 23:44:58 * nerdboy would recommend either machinekit/ubuntu+rcn kernels or build it with oe Mar 01 23:45:37 including meta-uav and other cool stuff Mar 01 23:46:49 we *could* think about a ci build/deploy setup and package up the work so it's easy to push/test Mar 01 23:47:14 * nerdboy used to hacking recipes Mar 01 23:48:18 package feeds, yada yada... Mar 01 23:59:39 nerdboy: ci for all (buildable) projects this year would be pretty sweat Mar 01 23:59:45 sweet even Mar 02 00:00:51 individual repos are easy too, i was thinking the whole OS build Mar 02 00:01:25 but yeah, students will write tests this year and run them via travis-ci Mar 02 00:01:41 oh yes they will... Mar 02 00:02:30 i think they default to ruby but you can .yaml your own "make blah" Mar 02 00:03:17 don't even need yaml, you can just use bash command Mar 02 00:03:31 oh, I'm thinking jenkins... Mar 02 00:03:43 https://github.com/sarnold/cyclo/blob/master/.travis.yml Mar 02 00:04:19 travis-ci != FOSS? Mar 02 00:04:21 if we want notifies we'll need a #beagle-gsoc-ci channel... Mar 02 00:04:53 i was just playing with github service-y stuff Mar 02 00:05:02 got a better one? Mar 02 00:05:21 I just got an instance of jenkins setup, seems pretty nice Mar 02 00:05:29 probably easiest/best documented for students... Mar 02 00:05:43 * nerdboy had to flog jenkins in previous job Mar 02 00:05:58 not really a good feel there yet Mar 02 00:06:23 we should have students push to their github repos and a single beagleboard hosted ci platform that does runs daily builds of each Mar 02 00:06:23 i meant a better public/foss service for github Mar 02 00:06:36 ci for github you mean? Mar 02 00:06:46 build/test on push is better feedback Mar 02 00:07:18 beagleboard one could do nightly OS/integration builds tho Mar 02 00:08:06 I've been meaning to check out drone.io... Mar 02 00:08:16 except "nightly" would be daytime for a lot of people... Mar 02 00:08:18 though self hosted is much better imo Mar 02 00:08:40 (I may have just set myself up to manage it by saying that...) Mar 02 00:09:13 doesn't rcn-ee use jenkins? Mar 02 00:11:52 one potential issue I see with building on push is that it could discourage students to hold back their pushes Mar 02 00:26:30 or encourage them to push in a buildable state Mar 02 00:26:42 plus it doesn't build every branch Mar 02 00:26:54 so encourage branching, etc Mar 02 00:27:16 true. We should encourage good branch management Mar 02 00:27:24 push/communication frequency should be scored at midterm Mar 02 00:27:49 pushing design/docs/reqs early would be good... Mar 02 00:28:19 software dependencies, etc Mar 02 00:30:52 we could go as far as to require repos be created during the application period with a README and labeled issues for everything on the timelines... Mar 02 00:31:18 * nerdboy likes that Mar 02 00:41:28 * nerdboy also wondering how many vendors will use 2-factor auth with the new "smart" credit/debit cards... **** ENDING LOGGING AT Wed Mar 02 02:59:58 2016