**** BEGIN LOGGING AT Mon May 09 02:59:59 2016 May 09 03:52:36 mdp: please see email just sent about SPI flash emulator project May 09 03:53:55 I think we need to modify the scope and/or implementation of the project May 09 13:23:28 pmezydlo: did you see my last email and understand the problem? May 09 13:28:16 m_w:so I've seen it May 09 13:29:42 m_w:I don't know what I think about it May 09 13:30:25 pmezydlo: well firstly don't panic we will figure something out May 09 13:31:04 m_w:spi slave under certain conditions can run May 09 13:31:35 yes but not as a generic protocol framework May 09 13:32:13 it cannot emulate SPI flash or many other devices that use register maps May 09 13:33:09 because of the transmit restrictions, unless you see something that I don't May 09 13:34:44 in my opinion, which driver isn't needed May 09 13:35:48 ??? May 09 13:38:07 slave driver which don't send data isn't needed May 09 13:38:54 FPGA is a good idea May 09 13:40:44 well FPGA also adds a layer of complexity May 09 13:40:56 are you familar with FPGAs? May 09 13:41:56 m_w yes, i have FPGA and I've done several projects May 09 13:42:31 well lets wait for some feedback from the other mentors May 09 13:43:31 hey jic23 May 09 13:44:45 hi. Having boring day (fpga debugging so long round trip on tests) May 09 13:44:53 Thought I'd drop by. May 09 13:45:04 What's the issue with the spi slave flash memory thingy? May 09 13:45:16 Just not enough time to build the messages? May 09 13:45:36 well there was an oversight in the project spec May 09 13:46:05 the McSPI controller has some restrictions that disallow generic protocol emulation. May 09 13:46:25 I'd assumed we were looking at bit banging on that one... May 09 13:46:48 never read the proposal :( May 09 13:47:10 yeah it talks of using the slave support from mcspi May 09 13:47:34 . When McSPI is selected by an external master (active May 09 13:47:34 signal on the SPIEN port assigned to channel 0), the transmitter register May 09 13:47:34 content of channel0 is always loaded in shift register whether it has been May 09 13:47:34 updated or not. The transmitter register should be loaded before McSPI May 09 13:47:37 is selected by a master. May 09 13:48:00 chicken egg problem May 09 13:48:00 hmm.. fairly standard for a hardware implementation I think... May 09 13:48:50 can't do offloading to hardware quick enough any other way. May 09 13:49:14 don't think that the hardware will work either way May 09 13:49:15 so it's expected a request / then gap / then reply? May 09 13:50:13 because the transmitted data is dependent on the first few bits of data in the transaction May 09 13:50:14 (expecting - gah - where does my ability to type English go the moment I start using IRC) May 09 13:50:22 Sure. May 09 13:50:53 Why can't you bit bang it? (not using the serializer) but instead just relying on high speed gpi reading. May 09 13:51:27 this was one of my options that I proposed May 09 13:51:29 Should have a few clock cycles to work out what the reply is. May 09 13:51:59 the other was a more flexible slave controller in FPGA May 09 13:52:31 hmm. To make it flexible enough and quick enough you aren't going to be far off just implementing a flash chip! May 09 13:52:50 which I suppose does solve the stated problem (sort of) May 09 13:55:38 I think that the PRU emulating SPI flash is possible but SCLK may not be fast enough for productive use. May 09 13:56:08 hmm. would have thought a few 10s of MHZ should be doable with a lot of care. May 09 13:56:19 the idea is to speed up development not slow it down :D May 09 13:57:05 bradfa: care to chime in here? May 09 13:58:10 lots of interesting stuff you could do with an spi slave emulator - such as writing device drivers whilst hardware is still being built... May 09 13:58:24 (i.e. the chips are at the fab rather than the boards) May 09 13:58:39 yeah it would be pretty nice May 09 13:59:12 obviously can do this in qemu but then you aren't getting all the odd timing off actual spi controllers etc. May 09 14:00:43 m_w: emulation of an actual SPI flash does not need to be the result of gsoc, imho, a somewhat generic SPI slave implementation (and what ever faults it has due to the issues you've brought up) is perfectly OK. The spi flash emulation was just an idea, it doesn't have to be the singular goal. May 09 14:01:35 m_w: for example, I have a need for a SPI slave where ever time CS asserts, the master is going to send me 4 bytes at 16 MHz. I don't need anything fancy to meet this kind of need and the slave never transmits, but a generic and easy to use Linux SPI slave interface would help me greatly if I wanted this slave to be a Linux device May 09 14:01:44 s/ever/every/ May 09 14:02:24 m_w: I haven't looked into your concern about McSPI and loading the transmit buffer yet, sorry, I'll try to take a look at that today and reply to your email tonight if I can May 09 14:02:40 okay May 09 14:02:45 if this project was easy and straightforward, it wouldn't be enough work to be a gsoc project :) May 09 14:04:01 :) May 09 14:04:24 well mainline SPI slave framework does not appear to be able to fit into a google decade of code May 09 14:04:58 looking at the past attempts on the mailing list May 09 14:05:32 does it have to be a generic driver? May 09 14:06:25 pmezydlo: if you want to get it upstream, I think making it as generic as possible is a good choice. But starting in a non-generic way to find the edges of what's possible is OK, I'd think. May 09 14:07:05 * bradfa had prompted pmezydlo to search for previous attempts at getting SPI slave mainlined prior to student selection... May 09 14:07:24 http://lists.infradead.org/pipermail/linux-arm-kernel/2012-August/116116.html May 09 14:07:45 here is a mention of a use of a bargepole May 09 14:09:43 Hi Jason May 09 14:09:45 Marek also brought forward some valid points in the email response May 09 14:11:12 http://idioms.thefreedictionary.com/I+wouldn't+touch+with+a+barge+pole May 09 14:11:42 for the idiom challenged May 09 14:12:32 thanks, m_w, I love idioms May 09 14:14:43 pmezydlo: the less generic the driver is the less work it will be May 09 14:15:22 for instance sending data when chip selected could be a google weekend of code May 09 14:16:18 yes, I understand this May 09 14:19:01 I will try to formulate a full description of what I mean and i will send this in emial. now i have to go to universities. May 09 14:19:45 pmezydlo: okay, hopefully we will have more feedback for you tomorrow May 09 15:45:12 hi Parthi May 09 15:46:15 alexhiam: seen amr? May 09 15:47:10 jkridner: nope May 09 15:48:31 I was listening in on your convo the other day through the gitter channel, I think I'm in agreement with the python strategy, though I'm not sure it should be running out of the box if it is just executing whatever python code gets sent to it as root May 09 16:10:45 alexhiam: are the shipments dispatched? May 09 16:13:51 kiran4399: I don't see tracking numbers yet, but jkridner would know better than I May 09 16:14:47 they have not. May 09 16:14:59 trying to get them out soon... some items requested I don't have. May 09 16:15:31 I will be moving sooner than I expected May 09 16:16:48 lemme put the new address on the spreadsheet May 09 16:16:52 m_w: thanks. May 09 16:19:29 alexhiam: I got the uImage after compiling the kernel and building the required kernel modules for bb blue.. May 09 16:19:41 alexhiam: I don't know how to make an image from that.. May 09 16:19:50 alexhiam: You've got any clue?? May 09 16:20:09 kiran4399: are you using this? https://github.com/beagleboard/image-builder May 09 16:20:18 alexhiam: yeah.. May 09 16:20:34 alexhiam: but how to input an overlay and an uimage file? May 09 16:21:17 to the image_builder May 09 16:22:39 not sure, I haven't actually used that before, but I would start by reading through the shell scripts and configs May 09 16:23:04 and I bet jkridner knows his way around the image builder... May 09 16:23:24 jkridner: I put the new address in the comment field May 09 16:23:40 jkridner: ping :) May 09 16:25:55 Zeek about? May 09 16:33:04 ? May 09 16:33:19 kiran4399: overlay? May 09 16:33:31 as in, within /boot/uEnv.txt? May 09 16:36:15 Yeah.. May 09 16:36:21 jkridner: ^^ May 09 16:36:38 jkridner: I got the rootfs for debian.. May 09 16:36:41 why do you want to do that with image-builder? May 09 16:37:05 jkridner: Well.. I thought we could build image with image-builder.. May 09 16:39:27 https://github.com/beagleboard/image-builder/blob/master/target/boot/beagleboard.org.txt May 09 16:39:58 building with image-builder sounds great, but if you have an overlay, why not have your startup script do 'config-pin overlay CAPE'? May 09 16:40:20 kiran4399: fyi for dev and testing you're really not going to want to be building images for every revision! You can deploy your kernel image directly to the beaglebone May 09 16:40:52 alexhiam: yeah.. I know that.. but I thought it would be good if release an image for beaglebone blue.. May 09 16:41:27 I suspect we'll want the black, blue, green, etc. to all use the same image May 09 16:41:39 alexhiam: OK.. sounds cool. May 09 16:41:58 jkridner: am I right to think that? ^ May 09 16:42:02 kiran4399: the image should support both Black and Blue. May 09 16:42:31 kiran4399: for the first phase, you'd add the robotics cape overlay to the repo of overlays. http://github.com/beagleboard/bb.org-overlays May 09 16:42:41 then, it'll show up in the images. May 09 16:43:14 next phase, when the Blue is there, it'll have a different EEPROM and the dtb for the Blue will load. May 09 16:43:22 so, no need to change uEnv.txt. May 09 17:04:15 jkridner: by the second phase.. will you be done with the dtb's? May 09 17:04:39 jkridner: and by when are you planning to send the blue? May 09 17:15:25 Hi Jason May 09 17:16:32 Parthi: use irc nicks if you want to ping someone, otherwise their irc client won't notify them May 09 17:17:01 Oh ok , thank you May 09 17:21:46 Hi jkridner May 09 17:21:56 hi Parthi May 09 17:21:58 that worked. May 09 17:22:05 wow May 09 17:22:16 sorry I am new to irc May 09 17:22:32 I was hoping we could run into pdp7 and amr here, but I'm struggling to get them to idle here... May 09 17:22:38 and I'm not available very often either. :( May 09 17:22:57 amr is working as a GSoC intern updating bone101. May 09 17:23:15 oh ok May 09 17:23:23 pdp7 is a Foundation board member and has been working on Linux kernel drivers for various sensors. May 09 17:23:24 hi amr May 09 17:24:18 Hi pdp7 May 09 17:24:56 nice to meet you all trough irc :) May 09 17:25:06 I really hope amr can start idling in here a lot more May 09 17:28:17 why can't we have slack like chat app in beagleboard.org May 09 17:29:39 the bb.org community is full of kernel devs, and kernel devs like their irc :P May 09 17:29:53 ok :) May 09 17:30:46 there are actually slack and gitter channels, but they're just mirrors of the irc channels May 09 17:31:53 I found an open source project called rocket.chat ,I deployed in heroku and tested It worked fine May 09 17:33:38 yeah, I like rocket.chat, I think self-hosted is the way to go for something like that. But moving away from irc doesn't seem to be a very popular idea around here... May 09 17:34:44 Ok I will get comfortable with irc :) May 09 17:39:53 Gah irc obviously dropped out on me earlier... Wrote a nice essay.. Ah well.. May 09 17:40:02 Zeekhuge May 09 17:40:47 Email on that bug is good bit if you can spare time to do debugging then that's better! May 09 17:40:49 Hey jic23 May 09 17:41:25 Standard open source rule. Bug report good bug report with patches excellent! May 09 17:41:40 okay May 09 17:41:59 actually that mail was basically to get Suman Anna's help May 09 17:42:08 as he is the author of the drivers May 09 17:42:40 IIRC - module unloading is broken May 09 17:42:46 Probably just a case of scatter of printf in driver probe to see where it dies. Good simplistic debugging. Can be fun :) May 09 17:42:49 soemthing to do with hwmod or relatedstuff. May 09 17:42:59 alexhiam: can I post weekly progress and stuff on my website? May 09 17:43:02 okay. May 09 17:43:31 kiran4399: you can, but just copy and paste the text to a mail on the ml so it lives there forever as well May 09 17:43:41 I will have to compile it then ... I'll do that, once the video is prepared. May 09 17:44:06 At present the first thing for me is video, before 13th. May 09 17:44:07 kiran4399: you can have a link, but also want the text of it on the ml May 09 17:44:07 Ah. Indeed video first then compile it :) May 09 17:44:25 alexhiam: right.. May 09 17:44:56 Funny really. Never occurred to me that there would not a binary available :) May 09 17:45:03 I will prepare the script tonight and will show it then .. May 09 17:45:34 Not > be May 09 17:45:50 Hate phone autocorrect sometimes May 09 17:45:57 Also, It would be nice to have your suggestions for the video. Maybe some points to show-off ! May 09 17:46:06 jic23: ^^ May 09 17:46:13 kiran4399: how's your video going? May 09 17:46:31 forget about modules May 09 17:46:32 alexhiam: It's done.. I just got to upload it.. May 09 17:46:39 easier and less trouble to link in May 09 17:46:40 awesome May 09 17:46:46 alexhiam: The audio quailty is not good. :-( May 09 17:46:51 Sure, though might not get a chance to look at script till morning. May 09 17:46:58 alexhiam: I'll see if can record it once more. May 09 17:47:05 okay, no problem :) May 09 17:47:11 kiran4399: just have a laptop mic or something? May 09 17:47:31 alexhiam: yeah.. last time I kept the mic too close.. May 09 17:47:34 :-P May 09 17:47:48 ah May 09 17:48:01 what are you recording the audio with? May 09 17:48:09 what software that is May 09 17:48:26 alexhiam: earphones mic.. senheiser :-P May 09 17:48:45 alexhiam: recordmydesktop May 09 17:48:49 well that should be good enough May 09 17:49:07 you could do the audio separate with audacity May 09 17:49:25 which gives you a vu meter so you know when you're over driving May 09 17:51:50 ds2: I need to get over the bug before the coding starts ? May 09 17:52:07 That is given I have to use this version. May 09 17:52:49 though it works fine as long as its not unloaded. May 09 17:54:10 Please suggest me an irc client for mac.. May 09 17:54:51 ZeekHuge: Why not just compile in and not use modules? May 09 17:55:06 the robot voice on henrix's video is difficult to understand May 09 17:56:18 it hurts my head even listening to it May 09 17:57:35 omg that's unbearable May 09 17:58:38 and unparsable May 09 17:59:45 I can read the slides myself :-/ just the boom boom music and a little dot to follow along like a little kid would have been better May 09 18:00:02 ds2: 'compile in' ? as in ? we have to build over the remoteproc module . isnt it ? May 09 18:01:51 when a driver is built into the kernel there is no module May 09 18:02:53 the kernel binary contains the driver May 09 18:03:40 this makes it such that the driver cannot be unloaded with rmmod May 09 18:04:07 ZeekHuge: why do we have to build it as a module? May 09 18:04:13 why do you need to repeated remove and load the driver anyways? May 09 18:04:26 CONFIG_FOO=y instead of CONFIG_FOO=m May 09 18:04:38 Parthi: I use hexchat May 09 18:05:02 looks like they have an os x build May 09 18:05:09 Oh thank you alex May 09 18:06:20 oh okay .. so not as a loadable kernel module, but as an in-kernel part. May 09 18:06:26 okay, got it. May 09 18:06:50 right. CONFIG_FOO=y instead of CONFIG_FOO=m. got it. May 09 18:07:02 Okay. May 09 18:08:07 Though that would probably consume more power. May 09 18:08:17 Even when the PRUs are not being used. May 09 18:08:23 right ? May 09 18:08:34 ds2, m_w ^^ May 09 18:08:57 makes no differnece if done right May 09 18:09:01 when the driver is loaded there should be no difference May 09 18:09:14 IIRC, the unloading doesn't power of off properly anyways May 09 18:09:24 okay. May 09 18:22:01 Okay ! So we have the birthday boy here ! May 09 18:22:14 Many Many happy returns of the day chanakya_vc ! May 09 18:22:47 Thanks ZeekHuge ;) Thank you so much!! May 09 18:23:23 I hope that I am soon at your place and then we'll have a party there ! May 09 18:23:52 (that would probably involve lots of coding for the project :P ) May 09 18:24:00 ZeekHuge, Yup come north. We will celebrate :) May 09 18:25:20 ofc we will! May 09 18:25:36 Chalo, I need to go now and study for the exams ! May 09 18:25:59 chanakya_vc: Once again, Happy birthday ! :) May 09 18:26:00 ZeekHuge, Best of luck and thanks for your wishes! May 09 18:26:09 :) May 09 18:28:10 chanakya_vc: Happy Birthday! May all your wishes come true :) May 09 18:29:10 Thanks pmezydlo , Thanks a lot for your wishes : ) May 09 18:29:48 Hi alexhiam May 09 18:30:25 i connected through hexchat May 09 18:42:43 Hi jkridner May 09 18:43:33 when can i get the grove modules? May 09 19:59:49 hi jkridner May 09 19:59:56 hi amragaey! May 09 20:00:09 alexhiam: you around? May 09 20:00:27 yo May 09 20:00:36 * jkridner hasn't looked at the new proposal doc yet May 09 20:00:45 you checked my email ? May 09 20:00:48 * alexhiam has May 09 20:01:00 I think the timeline is looking better May 09 20:01:09 cool. May 09 20:01:28 we all good with fixing BBUI first? May 09 20:01:35 though I'd prefer to see some more details as to the actual implementation of stuff May 09 20:01:52 yeah May 09 20:01:56 * jkridner wonders how we'll present the combo of BoneScript and PyBBIO demos within bone101. May 09 20:02:10 (that gives me a bit more time to get PyBBIO up and running on 4.x :P) May 09 20:02:20 * jkridner agrees, the BBUI details are still vague. May 09 20:02:43 python and node.js sections at the top level? May 09 20:03:45 that's valid. I was thinking tabs such that the code was the only delta. May 09 20:04:04 alexhiam: we skipping use of bonecards? May 09 20:04:13 that would be cool if they're parallel enough May 09 20:04:34 alexhiam: PyBBIO might have some additional functionality, but I think mostly they do the same sort of thing. May 09 20:04:37 hmm, how far you think we are from having the bonecards finished? May 09 20:04:51 If there's only code for 1, then no tabs is what I was thinking. May 09 20:05:07 alexhiam: I dunno... they are there, just buggy. May 09 20:05:16 don't know how long to flesh the bugs out. May 09 20:05:16 jkridner: yeah, I like that idea. The sections could be more like 'Digital Input', etc., instead of being organized by particular functions May 09 20:05:25 yeah. May 09 20:06:01 that way of organizing it would fit nicely with the bonecards... May 09 20:07:28 so, do you think of updating bone101 UI ? May 09 20:08:38 amragaey: always, but we have to take baby steps where we can. I'm thinking start with what is live (not bonecards) and move to bonecards later. May 09 20:09:12 http://jadonk.github.io/bone101/Support/bonecard/ documents a bit of the original bonecard layout. May 09 20:09:27 http://jadonk.github.io/bone101/Support/bonecard/create/ has the code for creating bonecards. May 09 20:09:56 wait, is the BBUI stuff really going to take until midterm? May 09 20:10:01 the concept is that each of the tutorials should be "forkable" and individuals should be able to collect decks of bonecards as helpful little tutorials. May 09 20:10:27 midterm is only 5 weeks. :( May 09 20:10:46 right May 09 20:10:50 I'm not fond of the weeks 1-2 deliverables though. May 09 20:11:00 playing and getting familiar is what you do no. May 09 20:11:04 s/no/now/ May 09 20:11:05 now. May 09 20:11:09 agreed May 09 20:11:32 there should be some kind of simple deliverable in week 1 and code progress 4-5 days a week. May 09 20:11:51 i'm starting my final exams this week until 4/6. May 09 20:12:26 I tried to make the timeline flexible, as I will increase my working hours to be full time after exams May 09 20:12:27 June 4? May 09 20:12:32 yes May 09 20:12:44 and if we're doing the python as just a server that runs whatever code you throw at it over the network as root (cringe), then that really might be closer to 1 week May 09 20:13:15 alexhiam: isn't that how all PyBBIO code is executed? May 09 20:13:23 yikes, so how much time will you have for gsoc the first 2 weeks? May 09 20:13:36 yes, but I have to go first on bonescript functions May 09 20:14:01 amragaey: then simplify the task, but "Getting familiar with the platform and having environment installed and get ready." is a NOP. (not an operation) May 09 20:14:15 jkridner: yeah, but you have to be the root user to write and run the code, where as this will run literally any code as root, e.g. 'import os; os.system("rm -rf /")' May 09 20:14:34 from any device on the network May 09 20:15:05 or from any website that uses the socket.io code to detect a beaglebone on your lan... May 09 20:16:02 alexhiam: yeah, and I see people run unpatched BeagleBone's on the Internet. May 09 20:16:14 jkridner, this bonding period i'm reading bone101 code, I installed the ruby environment and ran jekyll and trying to get familar with BBUI May 09 20:16:19 I forgot the name of the search tool, but some return the server headers and you can just find them. May 09 20:17:02 We could create some kind of b.login()/b.logout()/b.setPassword() functions and encourage users to do a b.setPassword() early in owning the product. May 09 20:17:35 hmm, that could work May 09 20:17:42 a simple flag could maintain the login state and reject commands when the login is false. May 09 20:18:00 either that or implement a PyBBIO rpc server and sandbox the bone101 code, which is much more work May 09 20:18:07 we'd probably just tell folks to delete a file to reset the password to null. May 09 20:18:24 alexhiam: I'm looking for something amragaey can handle in this short timeframe. May 09 20:18:24 or to retrigger the password creation May 09 20:18:35 yeah, I'm onboard with that May 09 20:19:22 alexhiam, amragaey: but, remember that Cloud9 IDE on port 3000 is also wide-open as root. I guess people would need to know to shut down that service if they didn't want people running arbitrary code as root. May 09 20:20:32 right, but the difference here is that anyone could grab the socket.io code from the bb.org repo and run it on their website to detect and connect to a beaglebone on your lan, even if it's not internet facing itself, and run arbitrary code May 09 20:20:48 whereas you have to be on the lan or forward port 3000 to access c9 May 09 20:21:26 why wouldn't they have to be on the lan for bonescript too? May 09 20:21:38 oh, .. May 09 20:21:39 I get it... May 09 20:21:42 yeah May 09 20:21:46 on THEIR website. May 09 20:21:55 with bonescript its limited to the rpc api May 09 20:22:03 but with python it would be literally any code May 09 20:22:04 so, if you visit a malicious site, it could find any board on your land and screw it up. May 09 20:22:10 yeah May 09 20:22:22 alexhiam: you can execute arbitrary shell commands from BoneScript. May 09 20:22:23 in other words its a completely open door to a root user on your lan... May 09 20:22:32 ah, good point May 09 20:22:34 maybe we need to force a login access to this port May 09 20:22:40 so, I'm all for requiring a password May 09 20:22:48 amragaey: require or enable? May 09 20:23:06 my preference is to encourage users to run b.setPassword() early.... May 09 20:23:15 do you really think we should START with a password set? May 09 20:23:27 I mean, if it was a fixed password, it wouldn't provide any security anyway. May 09 20:23:31 it could start with a popup that tells you to set a password May 09 20:23:47 or that could be part of the initial setup somehow May 09 20:23:58 yes, so people will like to set a password at startup May 09 20:24:00 beagle-ui.js that does the pop-up today could detect if a password was set and request one be set. May 09 20:24:11 yeah, I like that May 09 20:24:29 k. May 09 20:25:32 do we care about that socket.io link being over http on the lan? May 09 20:25:43 we could use letsencrypt and do https there May 09 20:26:17 or just self signed certs May 09 20:26:32 since letsencrypt expires after 30 days May 09 20:27:12 amragaey: you understand the security issue we're talking about here? May 09 20:28:23 jkridner: though, since the logging in would be entirely through the js in the browser, a malicious site could just log your password... May 09 20:28:47 fair enough, but you should only give your password to a trusted site. May 09 20:29:14 I understand that we need to keep c9 ide running on 3000 secure, as not to keep anyone on the lan running a code as root May 09 20:29:20 * jkridner wishes OpenID was more pervasive. May 09 20:29:54 amragaey: I don't think we will address c9 security within the scope of your project.... we need to keep this something where you can succeed. :-) May 09 20:30:14 right, I guess the only concern would be a clone of bb.org, but if you can direct a person there I guess you're already on the same lan and could just use c9 May 09 20:30:56 amragaey: yeah, this is about that socket.io server to run bonescript and python code May 09 20:31:04 unrelated to cloud9 May 09 20:31:24 aha, I got it May 09 20:32:59 so we will enable users to set a password before running any arbitrary code on the bb May 09 20:34:00 here's an example https://auth0.com/blog/2014/01/15/auth-with-socket-io/ May 09 20:34:53 so we'd want the same server running the bonescript and pybbio code so they can use the same session May 09 20:36:50 if b.login() is used, it is within a socket.io session.... I believe the sessions are secure (ie., if you open up the socket again from somewhere else, you won't be able to send messages). May 09 20:37:33 that sounds right May 09 20:40:46 listeners are added in the onconnect() function here: https://github.com/jadonk/bonescript/blob/0.5.x/src/socket_handlers.js#L38 May 09 20:41:36 functions are called here: https://github.com/jadonk/bonescript/blob/0.5.x/src/socket_handlers.js#L90 May 09 20:42:06 if the local login state at that point is false, a substitute error-reporting function should be called instead. May 09 20:42:58 local meaning on the beaglebone, right? May 09 20:43:17 * jkridner really needs to enable one-shot functions to avoid run-away callback allocation at some point. May 09 20:43:31 alexhiam: local meaning in the context of a single connection/session. May 09 20:43:53 right now, that would be at every page load, I believe. May 09 20:44:10 sorry, internet disconnected May 09 20:44:23 amragaey: backlogs at /topic May 09 20:44:32 http://logs.nslu2-linux.org/livelogs/beagle-gsoc May 09 20:45:08 we could also have a token generated on the bone at login that gets passed to the client and stored as a cookie, then have that token expire after X minutes of not being used or something May 09 20:45:16 which would allow page refreshes May 09 20:47:16 Is the password is set for every session then ? May 09 20:48:11 it would need to be givenat the start of every session May 09 20:48:26 but only set when you first setup the bone, or when you want to reset May 09 20:49:07 aha, and it will be stored in bb server May 09 20:49:37 yeah, preferably as a hash May 09 20:51:39 Storing a hash on the BeagleBone makes sense to me. Not sure if it should be under /etc or /var. May 09 20:53:25 under /var I would think May 09 20:53:34 Having the browser store a cookie provided by the BoneScript server that goes invalid after a while is good. We can put some wrapper code in http://beagleboard.org/static/bonescript.js for b.login() to store the cookie. May 09 20:54:27 er, https://github.com/jadonk/bonescript/blob/0.5.x/src/bonescript.js May 09 20:54:51 the former was just to load the latter. both of those are client-side scripts. May 09 20:55:11 * alexhiam likes this plan May 09 20:55:26 should be pretty straightforward to implement I think May 09 20:56:56 amragaey: have you done any http session / cookie stuff before? May 09 21:02:34 I did basic session stuff in php, and in python frameworks It was built-in functions I used to just call it. May 09 21:04:50 pdp7: howdy... parthi was in here earlier. wanted to try to keep everyone aware of potential bone101 updates. there's also some contributions coming in from Seeed right now as they customize it for their boards. May 09 21:10:03 jkridner: bone101 contributions from seeed? May 09 21:10:16 yeah.... adding Green images, etc. May 09 21:10:26 oh, makes sense **** ENDING LOGGING AT Tue May 10 02:59:59 2016