**** BEGIN LOGGING AT Tue Jun 30 02:59:58 2015
Jun 30 05:22:31 <vvu> arianepaola: ping
Jun 30 05:22:40 <vvu> any good tutorials you followed for nw.js ?
Jun 30 05:56:43 <shubhangi> gm
Jun 30 11:12:44 <arianepaola> vvu: there is a list here https://github.com/nwjs/nw.js/wiki/Tutorials-on-Node-WebKit
Jun 30 12:18:46 <av500> vmayoral|pc: hi
Jun 30 12:18:54 <av500> can you join #beagle-mentors
Jun 30 12:19:06 <vmayoral|pc> sure
Jun 30 12:20:12 <vmayoral|pc> av500: need access
Jun 30 12:20:21 <av500> yup
Jun 30 12:20:23 <av500> sec
Jun 30 13:27:43 <vvu> arianepaola: thx
Jun 30 18:09:43 <alexanderhiam> neemo: still not seeing any framework code on github, how's that going?
Jun 30 18:14:15 <nerdboy> hmm, he was supposed to push what he had so far...
Jun 30 18:16:36 <Abhishek_> neemo: you could also push whatever you've done in a different branch if you want to clean up the master later.
Jun 30 18:16:52 <alexanderhiam> neemo: yes, please do push something ASAP
Jun 30 18:25:01 <alexanderhiam> neemo: even if it's not working, this is when we need to see and evaluate everything you've done and make sure you're on track, so it really all needs to be pushed
Jun 30 18:29:23 * vvu goes out of this danger zone :)
Jun 30 18:52:40 <jkridner> nerdboy: the merges are getting better each attempt.
Jun 30 18:56:13 <jkridner> pulled last request without changes.
Jun 30 18:56:48 <nerdboy> <ascii thumbs up>
Jun 30 18:57:47 <nerdboy> after all that, everyone in here should be git-ninja by now...
Jun 30 19:00:21 <Abhishek_> (y)
Jun 30 19:00:42 <jkridner> :-)
Jun 30 19:01:09 <Abhishek_> shubhangi: Pushed that overlay into the tree yet?
Jun 30 19:01:19 <Abhishek_> or it is still a gist?
Jun 30 19:04:33 <Abhishek_> nm, got it from the scrollback
Jun 30 19:58:49 <shubhangi> Abhishek_: does it work ?
Jun 30 20:22:06 * nerdboy enjoys the salsa roja for flavor and the habanero verde for some zip
Jun 30 20:26:53 * nerdboy forgot to ring the Taco Tuesday lunch bell
Jun 30 21:36:45 <moto-timo> arianepaola: sorry I missed our time this morning... something came up that I had to handle
Jun 30 21:48:42 <nerdboy> neemo: are you awake yet?
Jun 30 21:58:38 <DiegoTc> ebadawy: I wil be testing tonight
Jun 30 21:58:46 <DiegoTc> and open issues on github
Jun 30 21:59:24 <ebadawy> DiegoTc: great, thanks for that.
Jun 30 22:15:52 <vvu> azizulhakim: any updates ?
Jun 30 22:17:12 <azizulhakim> not much. tried to run the rpi image but unable to run it.
Jun 30 22:17:32 <azizulhakim> I think cache size might be responsible for high page fault
Jun 30 22:19:10 <vvu> can you change that ?
Jun 30 22:19:14 <vvu> in the vm of course
Jun 30 22:20:02 <azizulhakim> vmware or virtualbox doesn't give that facility. qemu might, but still didn't find something similar
Jun 30 22:20:41 <vvu> but are you sure vmware or virtualbox emulate same spec cpu ?
Jun 30 22:21:24 <vvu> don't you have a rpi available ?
Jun 30 22:21:35 <vvu> a.k.a. to borrow from somebody ?
Jun 30 22:23:39 <azizulhakim> I can buy one
Jun 30 22:24:08 <azizulhakim> and vmware or virtualbox might not emulate exactly the same spec
Jun 30 22:24:26 <vvu> try 1st the vmware virtualbox
Jun 30 22:24:27 <azizulhakim> I think they emulate the same cpu as the host machine
Jun 30 22:24:39 <vvu> 2 laptops if so ? :)
Jun 30 22:25:19 <azizulhakim> yes, you can reduce the ram size though
Jun 30 22:29:10 <vvu> mhmh we need to find a fix for this
Jun 30 22:29:16 <vvu> ortherwise that is bad
Jun 30 22:35:05 <azizulhakim> vvu: any idea in your mind?
Jun 30 22:35:36 <vvu> in the current state if you remove audio and input and leave just video, how does it behave ?
Jun 30 22:35:41 <vvu> do we have an okish framerate ?
Jun 30 22:37:20 <azizulhakim> not so good, it takes around 25 sec to transfer a 7 second video
Jun 30 22:37:45 <vvu> but in the logs you sent me you were playing the video file 23 seconds
Jun 30 22:37:47 <vvu> why so ?
Jun 30 22:37:53 <vvu> and in the normal pc and vm just 7
Jun 30 22:39:07 <azizulhakim> the data transfer takes 23 second time. so even though the video is 7 sec long, you'll see it for 23 second in bone
Jun 30 22:41:17 <vvu> ah...so the cache may be the issue
Jun 30 22:41:25 <vvu> ds2: ping
Jun 30 23:22:25 <vvu> azizulhakim: i think we should ask everybody from here about this :)
Jun 30 23:22:31 <vvu> all hands on deck!
Jun 30 23:26:28 <azizulhakim> :D
Jun 30 23:27:18 <azizulhakim> how do we overcome the cache issue if that is responsible for it
Jun 30 23:51:25 <vvu> nobody wants to help...
Jul 01 01:10:14 <neemo> nerdboy: alexanderhiam Abhishek_ Missed out last night, crashed to bed after I got home
Jul 01 01:10:25 <neemo> I've pushed the 3D fitting up
Jul 01 01:11:38 <neemo> Used Yury's work for getting the params out and tried fitting, still have to figure out how to get a rotation matrix from the eigenvector output
Jul 01 01:18:51 <neemo> I've started the framework code just recently and didn't write much since I've spent most of the last two days at the chinese exit/entry bureau sorting out visa stuff :/
Jul 01 01:20:06 <neemo> nerdboy: alexanderhiam Abhishek_ I'll be here most of the day (apart from a couple of hours in the evening), so ping when you're back online
Jul 01 01:38:30 <nerdboy> you should include license for your stuff too
Jul 01 01:39:58 <nerdboy> so what do you have for a framework so far?  more than a glimmer?  smaller than a breadbox?
Jul 01 01:41:37 <nerdboy> can we get a first cut at an interface spec/header file(s) with some doxygen comments?
Jul 01 01:44:16 <nerdboy> or maybe start with napkin-diagram and a description...
Jul 01 01:54:59 <neemo> nerdboy: pretty much
Jul 01 01:55:25 <neemo> I'm working on the napkin diagram and reading on how to not mess up the framework/API
Jul 01 01:55:56 <neemo> since my experience building such is, well, not big
Jul 01 01:56:02 <nerdboy> it's okay to have public and private includes
Jul 01 01:56:24 <nerdboy> that's why you start with the interface
Jul 01 01:56:37 <neemo> yeah, figured that
Jul 01 01:56:41 <nerdboy> sketch out some functions/parameters
Jul 01 01:57:00 <neemo> I'll be reading and transfering my ideas to code and/or diagrams or something I can share
Jul 01 01:57:28 <neemo> so we can have a discussion and so you guys can give me advice on where to go from there
Jul 01 01:57:32 <nerdboy> or oblects/methods but i would not ask for c++ unless you're actually really good/careful with it...
Jul 01 01:58:25 <nerdboy> c is more straight-forward and interfaces reasonably well with other stuff
Jul 01 01:58:36 <neemo> I'll be at it today and share with you guys before the meeting I guess
Jul 01 01:59:12 <neemo> I'd prefer C as well, have less experience doing real stuff with C++
Jul 01 01:59:12 <nerdboy> fortran would be appropriate for matrix decomposition
Jul 01 01:59:24 <neemo> but happy to expand later
Jul 01 01:59:35 <alexanderhiam> remember there's the userspace API in C, as well as the driver interface, most likely through sysfs
Jul 01 01:59:38 <nerdboy> depends on libs you need
Jul 01 01:59:51 <neemo> dont most good math C libraries run their stuff in fortran in the bg?
Jul 01 02:00:04 <nerdboy> could use libsysfs underneath
Jul 01 02:00:06 <alexanderhiam> it would be good to see outlines of both (which would be similar APIs)
Jul 01 02:01:49 <alexanderhiam> have you done any linux driver work before?
Jul 01 02:01:58 <nerdboy> i'm sure there are c interfaces to some of the fortran stuff but a lot of math/science code is still in pure fortran
Jul 01 02:02:29 <alexanderhiam> I would say implement it all in C, then you can optimize later
Jul 01 02:02:43 <neemo> yeah, lots of planning and redesigning going forward
Jul 01 02:02:45 <nerdboy> sysfs interface isn't exactly a kernel driver interface
Jul 01 02:03:02 <neemo> alexanderhiam: yeah I'd agree with C
Jul 01 02:03:21 <neemo> alexanderhiam: also, no, I haven't done any linux drivers
Jul 01 02:03:32 <alexanderhiam> a sysfs interface is nice for having APIs in lots of languages
Jul 01 02:03:45 <neemo> I've read the basics and researched it a bit to prepare for this
Jul 01 02:04:13 <nerdboy> see if you can make a c interface using libsysfs
Jul 01 02:04:24 <nerdboy> then wrap it as needed
Jul 01 02:04:29 <neemo> alexanderhiam: if you've got any link that can get me to practical stuff fast, I'd be happy to digest it
Jul 01 02:04:45 <alexanderhiam> it would probably be good to start by writing a hello world driver
Jul 01 02:05:15 <neemo> nerdboy: also for the libraries stuff, I've looked into what other people have been doing with fitting estimation math
Jul 01 02:06:05 <neemo> this wasa good discussion over at SO http://stackoverflow.com/questions/7272252/fitting-an-ellipsoid-to-3d-data-points
Jul 01 02:06:12 <neemo> /wasa/was a/
Jul 01 02:06:37 <neemo> they mentioned http://www.gnu.org/software/gsl/ is a good idea for least squares fitting
Jul 01 02:07:29 <neemo> so we could use that as a starting point, not sure if it makes sense to reinvent the well on matrix multiplications/inversions etc
Jul 01 02:08:15 <nerdboy> https://www.kernel.org/pub/linux/kernel/people/mochel/doc/  <=  sysfs docs
Jul 01 02:08:49 <nerdboy> just look for the specific matrix libs you need first
Jul 01 02:09:08 <nerdboy> at least a dozen in portage...
Jul 01 02:09:26 <nerdboy> install sysfsutils doc package
Jul 01 02:09:30 <neemo> Ok
Jul 01 02:09:35 <neemo> will do
Jul 01 02:09:49 <nerdboy> libsysfs.txt
Jul 01 02:11:17 <nerdboy> https://bpaste.net/show/1dde02d7e120
Jul 01 02:11:36 <nerdboy> ^^ esearch -cS matrix
Jul 01 02:11:53 <nerdboy> couple extraneous hits...
Jul 01 02:12:41 <neemo> on it
Jul 01 02:12:55 <neemo> (reading time) :)
Jul 01 02:13:03 <nerdboy> hmm, never looked at that one...
Jul 01 02:13:08 <nerdboy> Matrix Algebra on GPU and Multicore Architectures
Jul 01 02:13:38 <nerdboy> anything called magma has to be good
Jul 01 02:13:54 <neemo> lol
Jul 01 02:13:58 * nerdboy waits for the "hot" punchline
Jul 01 02:14:48 <nerdboy> mumps is a somewhat less attractive name...
Jul 01 02:15:47 <nerdboy> jkridner: when do we get beaglebonequad ?
Jul 01 02:16:34 <neemo> that would be nice
Jul 01 02:17:11 <neemo> although still not enough to solve that monstrosity
Jul 01 02:17:21 <nerdboy> for now you are limited to one core and the PRUs...
Jul 01 02:17:38 <neemo> nerdboy: alexanderhiam any idea how the eig() function works in MATLAB, and/or how to extract the rotational matrix from the evecs output :?
Jul 01 02:17:51 <nerdboy> think about ways to pre-compute parts of the solution if necessary
Jul 01 02:17:58 <neemo> yeah
Jul 01 02:18:23 <neemo> that's what I was doing with that ugly (obscene even) non-linear equation
Jul 01 02:18:32 <nerdboy> maybe even precalculate solution space and pick one at runtime
Jul 01 02:18:52 <neemo> essetially, when it's solved, and it has to be solved only once for each "type" of fitting
Jul 01 02:19:32 <nerdboy> seen that done with precomputed wave solution and lookup tables
Jul 01 02:19:47 <neemo> type being 2D, 3D, or when we move to time-variant compensation, every change for the number of telemetry data inputs needs a recompute of the matrix
Jul 01 02:19:57 <neemo> so for example
Jul 01 02:20:07 <nerdboy> different problem than interface design
Jul 01 02:20:40 <nerdboy> draft something and we can "iterate" on it
Jul 01 02:21:02 <neemo> when you have a cubesat, you're not gonna change the number of solar panels, or the number of comm devices
Jul 01 02:21:03 <nerdboy> assume you have a backend solver for now
Jul 01 02:21:08 <neemo> ok
Jul 01 02:21:38 <neemo> ^ -> part missing after devices is, once you have a finished design for you cubesat
Jul 01 02:22:18 <neemo> it's gonna change during cubesat devel, presumably :)
Jul 01 02:22:30 <neemo> hardware devel that is
Jul 01 02:22:32 <nerdboy> you need a framework-ish thing that works for this problem but isn't specifically designed for this application
Jul 01 02:22:39 <nerdboy> if that makes sense...
Jul 01 02:22:43 <neemo> yeah
Jul 01 02:23:03 <nerdboy> data/control flow and interface(s) first
Jul 01 02:23:39 <nerdboy> dia exports nicely to svg...
Jul 01 02:24:09 <nerdboy> so does graphviz *if you already know dot notation*
Jul 01 02:24:27 <neemo> it's supposed to ease up cubesat development for any institution/DIY hacker and provide the usual stuff you would expect but be open to expansions for specific use cases
Jul 01 02:24:28 <nerdboy> should save that for later
Jul 01 02:24:48 <neemo> that's what I had in mind anyway
Jul 01 02:25:07 <neemo> yeah, dot is good for fast iterating
Jul 01 02:25:11 <nerdboy> so you could start with a system/architecture diagram
Jul 01 02:25:23 <nerdboy> might help clarify the mission
Jul 01 02:25:34 <neemo> that makes sense
Jul 01 02:25:42 <nerdboy> but only if you can do one fairly quickly
Jul 01 02:25:51 <neemo> so think of use cases and make a design that's flexible from there?
Jul 01 02:27:44 <nerdboy> you can (briefly) check out definition/examples if you want
Jul 01 02:28:26 <nerdboy> i'm talking operational concept for your satellite system
Jul 01 02:28:50 <nerdboy> the briefing you would give the General
Jul 01 02:29:30 <neemo> yeah, that's what I had in mind
Jul 01 02:29:35 <nerdboy> https://github.com/VCTLabs/elc2015-presentations/tree/master/prebuilt  <=  some examples of ops concept in security slides
Jul 01 02:29:43 * nerdboy likes NASA examples
Jul 01 02:29:45 <neemo> or rather, sir, yes sir
Jul 01 02:30:33 <nerdboy> https://en.wikipedia.org/wiki/Operational_View  <=  another good wikipedia page
Jul 01 02:30:56 * nerdboy farts in the general direction of wikipedia jokes
Jul 01 02:37:12 <neemo> Ok im on it
Jul 01 02:37:25 <neemo> operational diagram and a general outline of the framework
Jul 01 02:37:56 * neemo takes the comfy chair, reading time 
Jul 01 02:38:15 <neemo> also, brb, reboot
**** ENDING LOGGING AT Wed Jul 01 02:59:58 2015