**** BEGIN LOGGING AT Fri Mar 16 03:00:02 2018 Mar 16 03:26:04 Hello guys .... I am Adarsh from India ... Mar 16 06:25:00 hi Mar 16 06:25:05 anyone online? Mar 16 06:45:44 I wanted to know if anyone is ready to mentor a project regarding connectivity modules for pocket beagle? Mar 16 06:57:46 sdabhi23: maybe you should describe it in a bit more details .. Mar 16 07:00:02 Hey StephanieLC ! would you be interested in the Serial Terminal project ? Mar 16 07:21:36 zeekhuge I am planning to build sample projects which can be used as cut/paste libraries by the community and hopefully can also be incorporated in bone101 for wifi, ble and gps modules Mar 16 07:23:31 You mean the "Getting started - PBeagle" project ? as mentioned on the Ideas page ? Mar 16 07:25:20 Nope. Actually the "Connected PocketBeagle" project. Though I am also interested in "Getting started - PocketBeagle", but am not able to understand the main goal of that project Mar 16 07:42:22 Hey Wormo ! Mar 16 07:47:08 hey there Mar 16 07:48:41 Wormo: illustris may want to discuss on the Serial Terminal project. Mar 16 07:50:32 Wormo: Hey! I had some questions about the serial terminal project Mar 16 07:50:56 go for it, I'll do what I can to answer Mar 16 07:51:17 Wormo: Btw, is the StephanieLC user not working ? Mar 16 07:51:27 The oldest reference I can find of this project is this post on the mailing list from 2016: Mar 16 07:51:27 https://groups.google.com/forum/#!searchin/beagleboard-gsoc/serial$20terminal|sort:date/beagleboard-gsoc/YEDkXsH49E4/2YqPCQz1AwAJ Mar 16 07:51:27 Andrew Bradford originally seems to have wanted a userspace program that can run on the beaglebone, logging UART and allowing users to attach to the UART terminals over LAN. Mar 16 07:51:32 I mean, that is already in the channel Mar 16 07:52:34 I've already done some tests using UNIX pipes and FIFOs for logging and interacting with the UART devices. This looks like something I would be able to do. I was thinking of doing most of it in C, and using a shell script with cron for periodically splitting the logs and compressing them, like NGINX does with its logs. While telnet is usually used to connect to serial terminal servers, I think using SSH would Mar 16 07:52:34 be a better option. Mar 16 07:52:46 zeekhuge: I don't see that one in channel, I went back to Wormo because having 2 nicks confused jkridner Mar 16 07:53:20 ssh would certainly be a better option illustris Mar 16 07:54:11 And what are your thoughts on writing it in C? I found some needlessly complicated applications from last year using nodejs and a web server Mar 16 07:54:12 have you ever worked on dts files yet? Mar 16 07:55:07 The project description actually suggests C, so that shouldn't be too controversial Mar 16 07:56:32 Haven't done much with dts on BB. I first came across them when working on RISCV processors. I also had to modify the device tree of beaglebone to enable two CANs when i used them for a project Mar 16 07:57:16 that would be relevant for enabling all the serial ports Mar 16 07:59:43 Is there any groundwork you recommend I do before I start writing the proposal? Mar 16 08:00:25 I would look into existing projects that have good serial console support Mar 16 08:00:42 especially libraries that you could use Mar 16 08:02:08 Will do. Thanks! Mar 16 08:02:36 or even apps that are nicely extendible and could be adapted ad needed Mar 16 08:05:13 Some things like non-interactive monitoring of logs can be done by using standard programs like tail. Would I have to re-implement the functionality, or can I just take it for granted that these programs are shipped by default? Mar 16 08:10:31 have you had a chance to use terminal server hardware before btw? Mar 16 08:15:27 illustris: it's important to have some ideas for what UI would be like, and start running it by other developers to see if they would find it convenient Mar 16 08:16:11 Wormo: No, I haven't used one. They're really expensive 😅 Mar 16 08:17:03 Yeah not too surprised Mar 16 08:18:13 it's not like you need to totally copy their features either, but possibly reading manuals of ones with good reviews could be helpful Mar 16 08:18:31 Okay, I'll try to find some manuals online Mar 16 08:18:54 making up a good UI out of thin air is harder than it might seem at first thought :) Mar 16 08:19:58 I was thinking of something along the lines of screen and tmux, but with a prompt for configuring, listing, attaching to, and displaying logs from individual sessions. Mar 16 08:21:05 That could be a good starting point Mar 16 08:21:29 screen / tmux are quite usable and familar to many developers Mar 16 08:25:19 Does the UI need to be a fancy ncurses interface, or will a simple and functional prompt do? Mar 16 08:26:45 UI for selecting port, or ability to do ncurses on a serial connection? Mar 16 08:29:27 UI for selecting port. Doing ncurses on serial wouldn't be a problem at all Mar 16 08:31:31 echoing the escape sequences from an ncurses UI (or anything that moves the cursor around, prints colors etc) as is will ensure the ncurses UI displayed over serial will appear as is Mar 16 08:34:28 I am partial to CLI interfaces myself, especially if there is tab completion Mar 16 08:36:05 should get some other embedded dev opinions as well Mar 16 08:36:12 on the UI questions Mar 16 08:37:30 probably need to check at another time of day to find them though Mar 16 08:39:36 I made this shell once (https://github.com/illustris/OS-SOP-2-2/tree/master/assignment/traSH), but never got around to adding tab completion. I'd love to give that a try Mar 16 08:44:41 Heh "my hands are typing code" ... did you learn how to write commit messages from xkcd https://xkcd.com/1296/ Mar 16 08:44:59 Or, did xkcd guy learn from you? Mar 16 08:56:06 that repo wasn't meant for version control on a serious project 😅 Mar 16 08:56:21 Just a way for me an d the other guy to keep our work in sync Mar 16 08:56:38 teammate in school project? Mar 16 08:56:51 yeah Mar 16 08:56:58 I did realize it was not a serious project ;) Mar 16 13:19:34 hi Mar 16 13:20:07 i am trying to get my proposal accepted for gsoc, would anyone from beagleboard care to listen to it? Mar 16 13:24:51 i'll just go ahead and say it. I want to make a daemon for linux based machines which provides a real time speech-to-text engine. Why voice? because you cannot carry a keyboard walking down the street. Mar 16 13:25:47 the purpose is to dramatically increase the adoption of voice based input which is 1) easier than a keyboard, 2) almost as fast as a keyboard when done right. Mar 16 13:27:05 the daemon accomplishes this by 1) standardising the speech-to-text to run system-wide and applications can make requests to it. 2) A clean and easy to use API so that it's easy for users and devs to implement voice based interface. Mar 16 13:27:45 doesn’t that exist? Mar 16 13:27:53 But the project stands on the shoulders of giants, ie. the engine doing the actual work mozilla deepspeech, cmu sphinx are viable options. Mar 16 13:28:19 As far as i can tell there are very good engines available which i listed + kaldi and julius Mar 16 13:28:31 but there is hardly any user facing applications Mar 16 13:28:35 how can it be aided by a Beagle? Maybe it lives on the Beagle and makes keyboard/mouse presses for you? Mar 16 13:29:12 you can easily turn a pocketbeagle or other beagle into a keyboard and mouse. Mar 16 13:29:25 you cannot call an embedded device a tiny device if you have to bring a 2 feet keyboard to actually use it Mar 16 13:29:48 ???? Mar 16 13:29:52 why would you? Mar 16 13:30:14 beagles require no such things. Mar 16 13:30:28 do people coming in here and proposing ideas even find out what the Beagles are? Mar 16 13:30:49 im sorry Mar 16 13:30:55 35mm x 55mm powered off of USB. Mar 16 13:30:57 please let me explain Mar 16 13:31:07 can you ask specific questions Mar 16 13:31:17 ill answer them properly this time Mar 16 13:31:56 are you meaning for the automation to be for using a desktop or using your Beagle? Mar 16 13:32:01 thanks for listening to my huge ass explanation too, i was too rude sorry Mar 16 13:32:21 A: i intend it to be ubiquitous Mar 16 13:32:33 What I understood was for automating on your desktop computer. Mar 16 13:32:34 across linux based machines at first Mar 16 13:32:56 no, this has nothing to do with automation Mar 16 13:33:51 and make them easier to develop Mar 16 13:33:51 it is supposed to explore new human-to-computer interfaces beyond keyboard and mouse Mar 16 13:33:58 I consider hands-free control an act of automation. Is that not the case? Mar 16 13:34:24 i believe hands-free only means without the use of hands Mar 16 13:34:26 k. do you have a target programming language or such? Mar 16 13:34:47 the daemon probably in C Mar 16 13:35:22 i am not writing the actual speech-to-text translation part Mar 16 13:35:36 only a daemon/service so that it becomes more standardised Mar 16 13:35:39 my hope was to propose using a Beagle to replace the keyboard on a normal desktop and see if that could fit in your plans. Mar 16 13:36:01 Q. how can beagle replace a keyboard? Mar 16 13:36:08 mostly because I think voice controls largely exist, but aren’t really usable. Mar 16 13:36:20 agreed. i plan to change that Mar 16 13:36:39 the Beagle can look like a USB HID device to the host, emulating a keyboard and mouse. Mar 16 13:36:57 there are several examples of this in Python. Mar 16 13:36:58 and my idea is built on the assumption that today speech-to-text is there in terms of quality, but is hard to develop for Mar 16 13:37:28 so no real applications take advantage of voice input Mar 16 13:37:57 Q. and how does beagle know what i want to type? Mar 16 13:39:08 this one seems popular https://github.com/qlyoung/keyboard-gadget Mar 16 13:39:57 I like Phil’s stuff: https://github.com/ppolstra/usb-impersonator Mar 16 13:40:12 figuring out what to send could be tricky. Mar 16 13:40:49 saying letters would be easy. words could also be automated. keyboard shortcuts would probably help a lot. Mar 16 13:41:22 https://github.com/ppolstra/UDeck/blob/master/udeckHid.py Mar 16 13:41:47 ah i get the automation part now Mar 16 13:42:19 most all applications accept keyboard input, so swapping a keyboard would make integration easier in some way. Mar 16 13:43:16 aah, interesting Mar 16 13:44:38 i was proposing something else entirely, applications like having voice control to browse the web and do tasks on computers(small or big) Mar 16 13:45:59 i don't think i'll be able to build the applications, but if i can build the foundation to make subsequent work easier, that would be pretty good i guess Mar 16 13:46:42 Is beagle only based on ARM? what about the Risc V hype? Mar 16 13:46:55 can you find something that exists today to provide a starting integration? Mar 16 13:47:07 it is ARM today. Mar 16 13:47:18 and PRUs. Mar 16 13:49:06 ithink linux has been ported to risc v Mar 16 13:49:30 it has, but the SoCs still have a way to go. Mar 16 13:49:39 ohhh Mar 16 13:50:07 there were some great demos at ELC Mar 16 13:50:57 ELC? Mar 16 13:51:28 Embedded Linux Conference which finishes yesterday Mar 16 13:51:36 many videos on youtube. Mar 16 13:52:17 cool Mar 16 13:52:44 back on topic, would beagle be interested in speech-recognition and the like? Mar 16 13:53:31 i could use help Mar 16 14:14:22 We could use some help attracting good mentors.... Mar 16 14:14:39 I don't know which direction I'll focus my mentoring right now. Mar 16 14:14:49 Seems our mentors are quiet this year. Mar 16 14:14:56 with a few exceptions... Mar 16 14:15:22 they are largely looking for more low-level large-value projects to draw them in. Mar 16 14:15:36 these userspace things aren't so attractive for our core mentoring community. Mar 16 14:15:55 ds2 could probably provide better emphasis. Mar 16 14:36:10 @nerdboy i wanted to dicuss the different kind of sensors and what will be preferred since there are many cheaper/expensive sensors how do i contact stephanieLc to discuss this ? kindly guide me i want to go over this before the final application date Mar 16 14:44:32 any GSoCers want to fix create.arduino.cc experience? Mar 16 14:54:36 jkridner: tried that with beaglebone black yesterday, experienced the issues Mar 16 14:55:24 would like to fix those Mar 16 14:55:28 indeed. I sat down with Massimo and Luca right after the announcement and we are working on plans to improve the experience. Mar 16 14:55:51 ravikp7: are you still a student this year? Mar 16 14:56:13 jkridner: yes Mar 16 14:56:34 cool! nice to have the experience already working with you. are you already working on proposals? Mar 16 14:57:34 The create server that runs on your host is written in Go, FYI, so node-beagle-boot would need some help if it was going to bootstrap boards. Mar 16 14:58:41 jkridner: started gathering info for etcher integration project, not sure if that project would be sufficient for 12 weeks Mar 16 15:04:34 jkridner: if we use node-beagle-boot as a node module in etcher to boot board into mass storage then etcher integration is going to be any easy task Mar 16 15:06:27 jkridner: if you can suggest me some more tasks with it like the webusb implementation or this arduino create fixes for bb, then I can make a proposal that suffices 12 week work Mar 16 15:08:04 In addition to etcher, the electron app should include features like showing the u-boot console as an option (click to view console during boot), provide a console after boot and act as a network proxy. Mar 16 15:08:23 I think with that, 12 weeks should be mostly filled. Mar 16 15:11:06 Pac23: Wormo = Stephanie Mar 16 15:12:13 I added capture of netconsole, I think we need to look at if we should capture the interface at the kernel, both network and serial, to provide interfaces within the Electron app as well as to act as a network proxy without host dependencies. Mar 16 15:12:21 Hi Wormo ! Mar 16 15:12:33 Wormo: any of these project proposals capturing your interest? Mar 16 15:12:47 btw, had a great time at SCaLE... what a great show! Mar 16 15:35:47 @zeekhuge thanks Mar 16 15:36:36 jkridner: electron app? you mean adding u-boot console to the etcher (resin.io's) app or the app that I created last year Mar 16 15:36:56 @wormo what kind of sensors do you want the student to use ? cheap ones like the dht11 or the expensive one's like the bosch 680 Mar 16 15:37:13 the app you created last year. Mar 16 15:37:31 also i have written the proposal and shared the draft and would like to know your comments and how it can be improved. Mar 16 15:38:29 jkridner: ok, so submitting a pull request to etcher for node-beagle-boot integration is still also a part? Mar 16 15:53:14 jkridner: watched your elc presentation, you talked about getting the kernel boot serial logs right in there like u-boot, just looking more into it.. Mar 16 15:53:37 have to say, enjoyed the presentation Mar 16 15:53:39 ravikp7: yeah Mar 16 15:54:19 now I got how you used it as pocket-beagle tester Mar 16 15:57:31 jkridner: (act as network proxy) means we should be looking into routing all networking from board through usb? Mar 16 15:58:07 if you got some relevant links for these things, make sure to share that, btw I'll looking at info from the presentation slides Mar 16 16:39:55 jkridner[m]: apparently wormo and i don't have ze google-y accounts any more Mar 16 16:40:51 mentors need a new invite every year Mar 16 16:41:08 (AFAIK) Mar 16 16:41:21 what’s that? do you mean for the summerofcode.withgoogle.com site? Mar 16 16:41:44 yup Mar 16 16:43:32 i’ll try to log in and send invites this weekend, if that’s the process. Mar 16 16:44:07 jkridner[m]: are you on windows? Mar 16 16:45:05 you're sending some odd wide character for a "'" (i�@Yll - i@Yll) Mar 16 17:05:36 iphone riot.im. Mar 16 18:14:09 can i get some quotes regarding the PyPRUSS project to include in my proposal Mar 16 18:19:03 jkridner: abhishek_: zeekhuge: ^ Mar 16 18:20:00 muneeb17: can you summarize the objective in just one sentence? Mar 16 18:20:41 in your own words Mar 16 18:21:27 Update the PyPRUSS API(currently using the UIO interface) to remoteproc and the examples(currently in PASM) to pru-gcc Mar 16 18:22:26 PyPRUSS is a python based binding/API to control/communicate with the PRUs Mar 17 00:37:48 jkridner[m]: Saw your discussion with riscii_ about HID emulation. What do you think of a BB powered macro keyboard? Mar 17 01:01:17 not sure how sustainable/reusable that’d be on its own without voice recognition. Mar 17 01:28:01 Is there HID emulation support for BB right now? I could only find a couple of inconclusive threads on mailing lists Mar 17 02:25:51 voice recognition... maybe we could combine parts of the offline smart speaker idea with that Mar 17 02:30:24 ds2: I replied to your comments on the mailing list, is there anything else i should consider/include in my Smart speaker project project? Mar 17 02:33:03 ds2: any other suggestions? Mar 17 02:33:39 apoorvtintin: be convincing... I am not the only person looking at it Mar 17 02:34:27 you gloss over the trigger word detection... some more justitification on why pocketsphinx is sufficient...most existing implementations have dedicated trigger word detection code **** ENDING LOGGING AT Sat Mar 17 03:00:00 2018