**** BEGIN LOGGING AT Tue Mar 15 02:59:58 2016 Mar 15 05:08:26 hmmm Mar 15 05:19:36 Hello everyone! I wrote the script to add support for https and error handling for loading Bonescript.js Mar 15 05:20:04 Since I don't have beagleboard, can someone please check if the script is working fine? Mar 15 05:20:05 https://github.com/raghavjajodia/bonescript Mar 15 06:24:00 Hey jkridner, Are you there? Mar 15 09:46:35 !any pmezydlo Mar 15 11:31:24 Hi everybody! I am interested in applying for GSOC under your organization. Mar 15 11:31:39 I have selected a project, worked out the timeline as well Mar 15 11:32:00 I also found a mentor who is willing to guide me during summers Mar 15 11:32:41 He is not one from the list of people listed on Ideas page Mar 15 11:33:01 Should I ask him to apply to be a mentor first? Mar 15 11:33:27 And, since I have already talked to him beforehand, will my application be considered as biased? Mar 15 11:33:51 I do not want to be disqualified for wrong reasons Mar 15 11:35:24 Please clarify. Thanks. Mar 15 11:57:15 bitsnbytes: yes, if you have someone who would like to be a mentor for beagleboard.org gsoc, they need to apply and be approved before they can officially be a mentor Mar 15 11:59:48 a (college) senior(who has cracked GSOC before) told me that if I bring another mentor on board with my project, my application may be disqualified because the mentor would be biased about my application(due to our previous talk) Mar 15 12:00:15 Is this true? Mar 15 12:03:43 bitsnbytes: I don't know. If your mentor wants to mentor, they need to sign up to be a mentor. I'm not sure about what kinds of things can disqualify someone, sorry. jkridner can likely help, here. Mar 15 12:06:26 bitsnbytes: regardless of any of your concerns, have you shared your project idea with the mailing list? Mar 15 12:06:47 bitsnbytes: if you haven't introduced yourself and your project idea on the mailing list, that's probably a good next step for you. Mar 15 12:19:13 Thanks! Mar 15 12:55:15 raghavjajodia: I'm here now. Mar 15 12:57:24 Checking if I can get @beagleslackbot to speak up Mar 15 12:57:39 if you want to use Slack: https://beagleboard.slack.com/messages/gsoc/ Mar 15 12:59:34 a bias based on conversation about the project shouldn't be an issue. if there was a previous relationship, that most likely would. of course, others should be allowed to engage the prospective mentor and propose similar projects. Mar 15 13:34:00 morning Mar 15 13:36:22 jkridner: Requesting a Slack invite Mar 15 13:47:04 sent Mar 15 13:50:24 oh cool, there's two way sync it seems Mar 15 13:50:41 yeah, it is Mar 15 13:51:23 I just restarted the bot to connect it to #beagle as well Mar 15 13:51:40 it might take a second to connect, not sure. Mar 15 13:56:02 got an error to debug: kumarabhishekkakkar@gmail.com Mar 15 13:56:14 ooops, sorry wrong pastebin. :( Mar 15 13:56:22 error: Received error event from IRC prefix=holmes.freenode.net, server=holmes.freenode.net, command=err_notregistered, rawCommand=451, commandType=error, args=[*, You have not registered] Mar 15 14:05:06 /msg NickServ VERIFY REGISTER beagleslackbot udfunnjqrnau Mar 15 14:05:29 oops, accidental space Mar 15 14:13:11 do you guys prefer gdocs for proposal drafts or better not? Mar 15 14:13:31 Testing @beagleslackbot for https://beagleboard.slack.com/messages/gsoc/ Mar 15 14:13:42 @hpfmn: I don't how the interface is for you guys this year Mar 15 14:14:01 * Abhishek_ is on Slack Mar 15 14:14:12 hpfmn: likely depends on the mentor. I'd say Google is trying to encourage Gdocs for draft proposals, so we should probably lean that direction as well. Mar 15 14:15:01 you can also do Markdown docs on GitHub but GDocs means that commenting on your proposals is very simple and can be very targeted. Mar 15 14:22:20 if the bot is really objectionable, I can leave it off this channel. There is a #beagle-support channel I can try to push more noise too. Mar 15 14:22:24 er, to Mar 15 14:30:00 @ apart from my proposal i was alsothinking of making a short video about what it is all about and wanted to make a prototype code to show my skills (code will be done almost till the end of proposal submission). so my question is when does beaglebone start shortlisting students for gsoc Mar 15 14:30:39 kartik: there isn't a shortlisting process. all students are viable up to the announcement. Mar 15 14:31:53 av500 might correct me, but I believe it is true and all mentors should be aware that there can be no commitments or guidance ahead of the announcement. we simply ask all the students to build the connections with mentors they can and do the best proposal they can. Mar 15 14:32:57 indeed Mar 15 14:33:12 kartik: no shortlist Mar 15 14:33:34 kartik: what counts is the content of your proposal Mar 15 14:35:03 thanks for the info. i should probably focus more on making a good proposal than a test code to show my ability of completing the project Mar 15 14:39:57 is that right? Mar 15 14:50:31 kartik: depends some on your potential mentors.... I personally prefer to see some code. Mar 15 14:51:20 I think we're all pushing for seeing some more stuff on github ahead of time this year Mar 15 14:51:28 thanks for the help Mar 15 14:51:29 showing code as well as documentation Mar 15 14:53:03 thanks for the help Mar 15 14:54:48 * jkridner would even prefer README.md proposals to gdocs! Mar 15 14:55:00 though I am supportive of gdocs. Mar 15 14:55:02 +1 Mar 15 14:55:29 Github Summer of Code ? ;-) Mar 15 14:55:38 I feel like gdocs are more prone to disappear and not be helpful references for future students than github repos Mar 15 14:56:33 yeah, gdocs are more about collaborating on a one-time-snapshot proposal, rather than a living document. Mar 15 14:56:41 * jkridner wonders if we should provide a template. Mar 15 14:56:49 * alexhiam may start refusing to look at any application draft that's not on github... Mar 15 15:02:47 No I wouldn't be that hard. As I said, it's more convenient to place comments inline in a GDoc, not sure about markdowns Mar 15 15:05:09 how do I log onto the slack? Mar 15 15:14:52 wait, jkridner, what's the slack for? Mar 15 15:15:41 logging + to try to draw some people in who are reluctant to use IRC. Mar 15 15:16:00 * jkridner looks at pdp7 Mar 15 15:17:18 can you enable some oauths or something? I'm not seeing a way to log in without an @beagleboard.org account Mar 15 15:17:37 ah, I think I have to invite otherwise.... not sure how to make it more open. Mar 15 15:17:59 There's a heroku app where one can enter an email and request an invite Mar 15 15:18:13 yeah, that's an appeal of gitter for me, it just uses github oauth Mar 15 15:18:43 I guess slack can't be totally open to registration because their pricing is based on team size, isn't it? Mar 15 15:19:16 yeah and scrollback is limited to 10000 messages, and I see us hitting that number before the GSoC ends... Mar 15 15:19:52 jkridner: have you looked at setting something self-hosted like rocket.chat up? that way you could have chat.beagleboard.org Mar 15 15:20:15 why not #beagle? Mar 15 15:20:16 alexhiam: no.... source repo? bridges to IRC? Mar 15 15:20:23 I set one up recently to try it out: https://chat.graycat.io/home/ Mar 15 15:20:24 av500: did add slack to #beagle Mar 15 15:21:14 no existing irc bridge that I know of, but I started writing a plugin for one of the python irc bot frameworks, would be simple enough Mar 15 15:22:22 jkridner: https://github.com/RocketChat/Rocket.Chat Mar 15 15:23:46 a bit of a pain to spin up on an already live server along side other stuff, but easy enough to spin it up in a digital ocean droplet and point at that with a dns record Mar 15 15:25:05 * alexhiam has started using letsencrypt.org and now has subdomains all over the place all with ssl certs, for free! Mar 15 15:27:05 but even with gitter there could be open beagle and beagle-gsoc rooms, private internal rooms, and support rooms for every beagleboard.org repo Mar 15 15:33:22 alexhiam: is there a gitter to freenode bridge? Mar 15 15:33:34 I know you get get to gitter by IRC, but I want to use the existing rooms. Mar 15 15:34:01 alexhiam: beagleboard.org now has letsencrypt.org certs, so https://beagleboard.org is live. Mar 15 15:34:09 nice! Mar 15 15:35:14 jkridner: looks like this might do the trick: https://sameroom.io/ Mar 15 15:35:56 that video though! Mar 15 15:39:36 [alexanderhiam] test message... Mar 15 15:39:47 jkridner: looks like it works^ Mar 15 15:39:58 that's from https://gitter.im/graycatlabs/PyBBIO/beagle-gsoc Mar 15 15:40:31 looks like the nickserv auth didn't work though... Mar 15 15:41:33 ugh, only 30 messages/day for free :( Mar 15 15:42:23 wut? Mar 15 15:42:30 sameroom.io Mar 15 15:44:23 nope, I'm not buying Mar 15 15:44:36 I have a bridge half written though... Mar 15 15:50:28 ugh, no one's written a Python library for the gitter API yet :( Mar 15 15:51:12 alexhiam: k, when you get one written, I'll create chat.beagleboard.org. I'm not sure what repo/gitter to associate with which channel (or what the pushback will be on #beagle). Mar 15 15:51:39 * jkridner was hoping to get pdp7 to participate since he's required to use Slack at his full-time job. Mar 15 15:51:40 jkridner: can you use your own domain with gitter? Mar 15 15:52:00 alexhiam: oh, I don't know. I was just thinking of doing an Mar 15 15:52:18 guess that could be done on http://beagleboard.org/chat instead. Mar 15 15:52:26 if it isn't a different server. Mar 15 15:52:26 gotcha Mar 15 15:52:59 I expect much pushback :P Mar 15 15:54:20 I think having the bridge in place will be essential Mar 15 15:56:41 hi Mar 15 16:25:52 jkridner: why are people reluctant to use irc? just cause the freenode web chat is not working? Mar 15 16:26:47 freenode webchat works fine... no? Mar 15 16:26:55 bradfa: you are so smart! :) Mar 15 16:54:00 alexhiam: I should build the bb blue kernel from 4.1 right? Mar 15 16:55:23 alexhiam: also... what about the device tree overlay for bb blue? should I make it from bb black or is it available somewhere? Mar 15 16:57:29 Hey m_w ! Mar 15 16:57:47 How are you ? Mar 15 17:00:08 kiran4399: I imagine there's probably an overlay for the robotics cape Mar 15 17:01:23 I think ideally the BeagleBone Blue would run the same exact kernel as the black (and green, enhanced, etc.), and there would be some nice documentation describing how to load drivers and start services to use the different blue features Mar 15 17:03:19 alexhiam: so.. do you want me to use the same overlay for the robotics cape? Mar 15 17:03:57 I don't know that it should have one overlay... Mar 15 17:04:21 e.g. someone might want to use the IMU and not use the PRU PWM driver Mar 15 17:05:01 and the IMU and baro wouldn't need overlays anyways, as you would just need to make sure I2C is enabled then modprobe the driver Mar 15 17:06:04 alexhiam: wait.. wait.. why do IMU and baro don't need any overlays?? they are onboard...right?? Mar 15 17:06:10 alexhiam, can you tell if the iio driver in linux is based on spi driver or not ? Mar 15 17:06:39 alexhiam: or is it that.. even though the device is onboard.. it is interfaced by i2c?? Mar 15 17:07:12 kiran4399: yeah, as long as the i2c bus it's on is enabled then you just need to load the module Mar 15 17:07:18 I mean, the IIO devices are communicated over SPI, so the iio driver will need to use the spi driver to actually talk to the device, isnt it ? Mar 15 17:07:41 ZeekHuge: I would assume so. I have no experience with iio though... Mar 15 17:08:16 alexhiam: then why are there kernel drivers for such devices?? can't you just modprobe them from userspace? Mar 15 17:08:18 aa... ohk. Mar 15 17:09:35 kiran4399: modprobe loads the kernel driver... Mar 15 17:14:54 ZeekHuge: good Mar 15 17:15:54 iio has spi, i2c, and platform bus devices Mar 15 17:19:45 the mainline got more support for invensense IMUs recently Mar 15 17:21:49 well should be merged soon Mar 15 17:24:05 http://git.kernel.org/cgit/linux/kernel/git/next/linux-next.git/log/drivers/iio/imu/inv_mpu6050 Mar 15 17:25:02 I tested the driver with the MPU-9250 Mar 15 17:25:30 alexhiam: OH!! sorry.. Mar 15 17:25:36 and full support from there should be relatively straightforward Mar 15 17:26:11 how from from mainline is the newest beagle kernel? Mar 15 17:28:31 m_w: not too, too bad. Look at http://GitHub.com/beagleboard/linux? There is a 4.4 branch. Mar 15 17:29:26 yeah but how many patches are applied on 4.4? Mar 15 17:29:37 I think there are quite a few Mar 15 17:30:22 it would be nice if they could get the rest of the DT overlay stuff mainline Mar 15 17:30:52 I think that other boards could benefit from it Mar 15 17:33:42 it has been in the works for quite a long time Mar 15 18:26:05 ohk so m_w , that means the iio driver communicates data over the spi driver to the iio devices. right ? Mar 15 18:42:07 yup Mar 15 18:43:43 So, cant we code a new driver, that would implement the same interface as of the SPI driver, but would communicate that data to the PRU rather than the SPI subsystem directly ? Mar 15 18:44:28 that would allow us to use the mainline IIO drivers and hence a large number of devices could be supported then. Mar 15 18:48:11 Hi I am interested to participate in GSOC 2016 Mar 15 18:48:27 Interested in first idea listed on gsoc ideas page Mar 15 18:48:49 Linux kernel support for embedded devices and interfaces Mar 15 18:49:33 mdp : What are the prerequisites should I show you to improve my chances to get accepted ? Mar 15 18:50:46 What features can I consider to include in my proposal ? Mar 15 18:52:17 bradfa: hi, Don't worry, I do realize that the recrutation is still in progress and that Google is still to decide who is going to participate in GSoC. Mar 15 18:53:14 bradfa: What do you think about my application? Mar 15 18:56:14 Hi punitvara_, I recommend coming up with an idea in the scope of that first project and discussing here and the list Mar 15 18:56:17 pmezydlo: I think you should make your application a little bit more specific to implementing a mainline-able SPI slave infrastructure and less about just SPI flash. Overall, though, you seem to be very enthusiastic about the project, which is a good thing, you just need to keep the GSoC focus on software and not hardware. Mar 15 18:56:46 That first one is just a vague general idea for improving any gap in existing drivers or other kernel support... Mar 15 18:58:23 punitvara_: it will probably help to look at some of http://beagleboard.org/project listings as often you'll find projects working around gaps in the kernel support you might improve upon. That can be a source for specific ideas you may propose. Mar 15 18:58:48 ZeekHuge: this is the idea I guess Mar 15 19:00:14 bradfa: thanks for the answer and i will correct it Mar 15 19:01:53 punitvara_: in the meantime, complete the hello world application as described on the ideas page and send a pull request..that will demonstrate to the mentors that you have basic competency for taking on a project. Mar 15 19:09:55 Ohk Thank you Mar 15 19:10:57 I have upstreamed some patches to linux kernel as well Mar 15 19:12:32 mdp: For improvements i.e. I should look into TODO pages in staging directory . Can please tell me which possible directories should I focus to improve feature for BBB ? Mar 15 19:15:01 jkridner: Thanks for your reply! I guess that setTargetAddress is a client side function-call. Mar 15 19:15:52 I couldnt find this functions implementation at github: jkridner/bonescript Mar 15 19:18:21 bradfa: should i remove the schematic cape? Mar 15 19:19:33 Thats why I uploaded this single file in github. Can you please share the link of this file in github repo? ( bonescript.js having setTargetAddress function) Mar 15 19:33:53 ohk m_w, so this is it : https://docs.google.com/document/d/15dfkOS3a2TO5SAIsbBdYWpAipQwbNE6PukJxP_1sRcw/edit?usp=sharing Mar 15 19:39:01 pmezydlo: the schematic probably isn't necessary for a gsoc application Mar 15 19:51:15 ZeekHuge: I will check out the doc when I get back from my meeting Mar 15 19:51:31 ohk m_w :) Mar 15 19:51:47 if you have a second Mar 15 19:52:02 m_w, : just chech the title Mar 15 19:52:08 *check Mar 15 19:52:23 for the idea. Mar 15 19:53:24 alexhiam, can you please see if the title for the idea looks appropriate ? https://docs.google.com/document/d/15dfkOS3a2TO5SAIsbBdYWpAipQwbNE6PukJxP_1sRcw/edit?usp=sharing Mar 15 20:04:11 jkridner: there is a friend of mine interested to work on Bone101, I believe he can make a great work, he is currently investigating the code and will confirm to you tomorrow. Mar 15 20:16:23 Hey bradfa ! can you please go through this and give your feedback on this :) Mar 15 20:18:48 alexhiam, If you get some time, please review this https://groups.google.com/forum/#!topic/beagleboard-gsoc/IC-NeFHp4T4 Mar 15 20:19:48 Hey ds2 it would be a great help if you could provide your feedback on this https://groups.google.com/forum/#!topic/beagleboard-gsoc/IC-NeFHp4T4 :) Mar 15 20:32:45 Hey nerdboy, can you please review this idea on the ml :) https://groups.google.com/forum/#!topic/beagleboard-gsoc/IC-NeFHp4T4 Mar 15 20:45:50 Hey Abhishek_ ! Can you please provide your feedback on this : https://groups.google.com/forum/#!topic/beagleboard-gsoc/IC-NeFHp4T4 Mar 15 20:55:10 sounds good to me Mar 15 20:55:40 can't really comment on any specific caveats Mar 15 21:00:10 sent it to stephanie since she just signeddup today... Mar 15 21:00:23 *signed up even Mar 15 21:11:05 jkridner: does mtg time stay stuck on same GMT offset? Mar 15 21:11:53 nerdboy: we've been doing US daylight savings. :( Mar 15 21:12:23 last week was PST Mar 15 21:12:30 not anymore... Mar 15 21:13:23 so does it move with daylight-stupid-time offset? Mar 15 21:14:09 * nerdboy bets stephanie hoping for 1000 pdt Mar 15 21:15:43 or do you need it to be lunchtime east coast? Mar 15 21:17:50 hi Mar 15 21:18:07 did you have feedback for ZeekHuge ? Mar 15 21:18:17 Wormo: ^^ Mar 15 21:19:46 using PRU to handle SPI would be a great choice for talking to lepton video-over-SPI camera for instance Mar 15 21:19:49 https://groupgets.com/manufacturers/flir/products/flir-lepton Mar 15 21:26:41 nerdboy: I think we were trying to align with PDT plus suitable times for India. Mar 15 21:27:17 well, that was kinda my point, albeit indirectly Mar 15 21:27:51 does India have clock switching stupidity like US? Mar 15 21:27:52 just because most of the US does stupid time doesn't mean everyone else does Mar 15 21:28:30 since we started last week with an offset (from gmt) we could choose to keep it that way Mar 15 21:28:48 or move with stupid time... Mar 15 21:29:04 * nerdboy just throwing it out in the open Mar 15 21:29:12 * Wormo votes for GMT Mar 15 21:42:09 Hello Mar 15 21:42:38 For the project "Port/implement MAV (drone) optical flow or stereo image processing to PRUs", what cameras can I use Mar 15 21:42:51 and what would be my parameters for chhosing one ? Mar 15 21:47:04 cameras would mostly be usb or maybe spi (like the flir lepton) Mar 15 21:47:46 you'll be limited in framerate/resolution by a few things, depending Mar 15 21:48:20 you get to do the analysis as an exercise for the application... Mar 15 21:55:59 Okayy Mar 15 21:57:57 And when we take the 2 usb cameras, we'd put a usb hub to connect them to the BBB, right? Mar 15 22:03:37 yup, only one host port Mar 15 22:04:08 unless you add cameras to bbmini imu cape Mar 15 22:04:30 or use mcspi bus for two leptons Mar 15 22:05:47 yashoza: google will show you some papers using bbb for image capture/processing Mar 15 22:06:45 at least one has analysis of opencv/other performance Mar 15 22:07:01 ewwwwwwwwwwww USB cameras Mar 15 22:07:20 yeah, all too common sadly... Mar 15 22:07:44 Yes..i have looked at a few papers and will use them as reference Mar 15 22:07:47 there are easier ways of hooking up cameras Mar 15 22:07:53 PRUSS is nice for that Mar 15 22:08:47 but in the project, the idea is to use PRUSS for getting the depth map from the captured images Mar 15 22:08:59 so thats what we are offloading to the pruss Mar 15 22:09:10 how big are your algs for figuring out depth maps? Mar 15 22:09:27 1G? 1M? 1K? 1byte? Mar 15 22:09:55 i dont have an estimate as i havent started on that..sorry Mar 15 22:10:00 ds2: what other ways? Mar 15 22:10:11 but they arent that big, could the pruss be used for data capture? Mar 15 22:10:14 s/other/easier/ Mar 15 22:10:18 nerdboy: my favorite is PRUSS Mar 15 22:10:31 using what exactly? Mar 15 22:10:34 yazhoza: what size though? you have very limited space on the PRU Mar 15 22:10:39 using the PRUSS IO pins Mar 15 22:10:44 works with parallel cameras Mar 15 22:11:15 example camera? Mar 15 22:11:18 only catch is you are a bit resolution limited; barely got it to work with PCLK of about 27MHz Mar 15 22:11:38 in my case it is a TVP5160AM taking composite video and digitizing it Mar 15 22:11:43 so around 320x240? Mar 15 22:11:49 no reason why it can't work with the parallel omni vision cameras Mar 15 22:11:58 no, 720x480 IIRC Mar 15 22:12:10 oh okay Mar 15 22:12:15 if the data is sinked in the PRUSS, it is a little easier Mar 15 22:12:20 that would beat usb... Mar 15 22:12:40 the issue was getting data out of the PRUSS so I reduced it to mono (Y only, not YUV) Mar 15 22:13:22 were you doing anything besides passing data? Mar 15 22:13:34 *anything in the prus Mar 15 22:13:51 well, 1 PRU was passing data, the other was generating an overlay on the video Mar 15 22:14:03 so, no...just passing data; in 2 directions :D Mar 15 22:14:34 the killer is the possible stall when writing outside of the PRU Mar 15 22:14:44 well, if that's the architecture i guess you're on the hook too... Mar 15 22:14:53 if the PRU had enough register memory to buffer 1 line, then it'd be easier Mar 15 22:15:01 on the hook for? Mar 15 22:15:08 helping? Mar 15 22:15:20 I am signed up as a mentor :D Mar 15 22:15:22 * nerdboy added the drone vision project Mar 15 22:15:49 I'd find doing it on the lepton to be more interesting :D Mar 15 22:16:11 IMO, there is not enough space on the PRUSS for interesting image processing but I am not an alg person Mar 15 22:16:55 I'm guessing it would be great if we could run the img processing alg on 1 pru and use the other one for data capture Mar 15 22:17:08 you need to define how big are the algs Mar 15 22:17:09 yeah, hard to tell for me without some analysis/testing Mar 15 22:17:24 and the resulting image could be passed on the the arm core Mar 15 22:17:27 hmm Mar 15 22:17:28 otherwise, you can tell me you are creating a unicon from the BBB Mar 15 22:17:34 unicorn Mar 15 22:17:52 haha Mar 15 22:17:53 I'd personally process it on the SGX Mar 15 22:18:00 more memory, rarely used in embedded apps Mar 15 22:18:05 i could apply for that the next GSoC :D Mar 15 22:18:09 might be enough to apply the initial ffts and pass that data to arm side Mar 15 22:18:13 ohh..okayy Mar 15 22:18:35 nerdboy: I doubt if there is enough IRAM or DRAM to do an FFT Mar 15 22:19:23 needs analysis, but i think you've got a better feel than i do right now... Mar 15 22:19:54 I been looking at doing a SDR on there and things are coming up very very tight Mar 15 22:19:55 yashoza: yes, don't forget you have sgx to use opencv/other on Mar 15 22:22:50 Okayy..I actually havent used it before..so yupp..i'll go dig in on that in some time Mar 15 22:23:28 ds2 : can i get any links to the work you've done? Mar 15 22:24:29 yashoza: I been behind on writing things up but - http://www.hy-research.com/be-wear.html ... it is mostly an overview Mar 15 22:24:40 the SGX and SDR parts hasn't really been written up Mar 15 22:26:13 Alright! Mar 15 22:26:42 with the PRUs, you really need to keep in mind code space and ram usage Mar 15 22:27:22 I was just looking at the Lepton camera by FLIR, and looks like it would make for a really interesting hook up :) Mar 15 22:27:34 yes..i understand! Mar 15 22:27:46 where are you based? Mar 15 22:28:27 India Mar 15 22:28:46 wonder how hard is it to source that there Mar 15 22:29:09 i'd be very interested in helping a project using the lepton :D Mar 15 22:29:24 I can check that up..most sites do ship stuff here Mar 15 22:29:48 but for that to work, there needs to be real value for the community Mar 15 22:31:12 Yes..and i through this project, i wanted to build a framework that would help future beginners in getting started with bbb+image processing Mar 15 22:31:33 from the basics of img processing..till object detection or so Mar 15 22:31:44 object detection? Mar 15 22:31:46 wouldnt that be considerable help to the community ? Mar 15 22:32:06 what frame rate? Mar 15 22:32:16 how is that different from telling folks use Open CV? Mar 15 22:33:10 * nerdboy in a lovely mtg... Mar 15 22:33:30 nerdboy: so getting your session to beep continiously would be bad? :D Mar 15 22:33:55 ^^ funny guy Mar 15 22:33:57 with depth map generation offloaded to the PRU, it would free up more resources for other sorts of opencv processing as well, such as object detection, which could then be overlayed on the PRU generated depth map. Mar 15 22:34:31 yashoza: what is your education background? Mar 15 22:35:23 with regards to the frame rate, that was still on the optimization list..getting it upwards of 30 fps would be great Mar 15 22:35:44 I'm pursuing electrical and electronics engineering Mar 15 22:35:48 ..third year Mar 15 22:36:46 I have a good background in opencv as i used it for a bunch of competitions Mar 15 22:37:23 but fundamentally, you are a EE/CE guy not an alg (image processing/math) guy? Mar 15 22:37:45 s/guy/person/; using it in the generic sense. appologies. Mar 15 22:38:23 Not an alg guy..but I have a good experience in opencv Mar 15 22:38:29 Yes..not a problem :) Mar 15 22:40:08 opencv has a lot of overhead/crap Mar 15 22:40:22 I don't bother with it to keep things simple Mar 15 22:40:28 having said that, the algs are still not that simple :( Mar 15 22:40:59 It might be helpful to explain (in a convincing way) how you plan to squeeze the code into the memory space available in the time frame of GSoC Mar 15 22:41:20 okayy Mar 15 22:42:20 i think doing a smaller task before the submission deadline and pushing it on github would be a good idea? Mar 15 22:42:32 wouldn't hurt Mar 15 22:42:49 something that could prove i could finish the project successfully! hmmm Mar 15 22:42:58 don't necessarily think of it as GSoC specific... do it because you want to do it and are interested Mar 15 22:43:08 if GSoC happens, great.... if it doesn't.... Mar 15 22:43:41 yess :) Mar 15 22:44:34 i initially though getting gsoc was important so that i could get a good mentorship..but i think i'd get that either ways :D Mar 15 22:44:45 yep Mar 15 22:45:40 great! Mar 15 22:46:24 btw, any pointers to a good hello world task i can start working on? Mar 15 22:46:47 look on the archives...there is a hello world for arm and pull task Mar 15 22:46:58 I don't have the URL off hand Mar 15 22:49:08 the cross-compilation task ? i am almost done with that Mar 15 22:49:09 i wanted to start with something in the direction of our project Mar 15 23:01:58 Thanks for the help ds2 and nerdboy! I'll go read up about things that we discussed today, and lets continue the discussion tomm! Mar 15 23:03:53 Wormo has some embedded image-y stuff Mar 15 23:04:06 experience even Mar 15 23:20:00 bradfa: saying API you mean: parameters, files that appear in linux and allow you to communicate applications with the driver? Mar 15 23:30:58 ds2: you mean like some loop optimizations and what-not? Mar 15 23:31:18 or are you hand-optimizing stuff? Mar 15 23:32:57 * nerdboy would say use the compiler for most things Mar 15 23:36:03 nerdboy: loop optimization? Mar 15 23:36:23 (sorry, got many things happening in parellel; lost my context) Mar 15 23:37:03 i mean gcc optimizations vs hand-optimizing code Mar 15 23:37:15 what about it? Mar 15 23:37:18 graphite, vector stuff, etc Mar 15 23:37:29 also -Os for size Mar 15 23:37:40 gcc doesn't work on the PRU yet, AFAIK Mar 15 23:48:13 anything available in the cgt stuff? Mar 15 23:48:40 http://hubicka.blogspot.com/2014/04/linktime-optimization-in-gcc-2-firefox.html <= interesting data on size vs optimization for firefox **** ENDING LOGGING AT Wed Mar 16 02:59:58 2016