<wpwrak>
live from the annual championship of efficient screen space usage. the winner is ... the yacas tutorial ! go to http://yacas.sourceforge.net/homepage.html then click on the "Tutorial" tab. (they've hidden the real URL somewhere in a style sheet or such. pretty good camouflage, too.)
<wolfspraul>
lekernel: we all don't know what this is :-)
<wolfspraul>
my qualification wouldn't be very, let's say, 'motivational', so I leave it up to the academic pros to do that :-)
<wpwrak>
some day we'll have a fine mess with all those probably short-lived projects in the qi-hw wiki. well, it just goes to confirm my theory that wiki as a distributed documentation tool is inherently flawed ;-)
<wolfspraul>
wpwrak: no no, don't worry.
<wolfspraul>
Carlos and I already have a perfect strategy for that.
<wolfspraul>
mediawiki has nice mass and batch deletion extensions...
<wpwrak>
;-))
<wolfspraul>
and they are already installed
<wolfspraul>
but let's give the students a fair chance
<wpwrak>
wouldn't it be better to set up a wiki at the university ? label the projects properly with their time slot, etc.
<wolfspraul>
I gladly donate some server storage and bandwidth for their projects :-)
<wpwrak>
that way, they can even stay after they're finished. that should be nicer for the students.
<wolfspraul>
well, now we get into the 'qualification' stage. Do they care? my view is this: they hobble up some fantastic plans, great on promises lousy on implementation. they need to somehow serve this past some superficial reviews to get their degrees, and the day they get their degree they care about these projects even less than I do :-)
<wpwrak>
the non-hierarchical nature of the wiki doesn't make mixing things very convenient. well, not without something like google to figure out what's actually relevant.
<wolfspraul>
Carlos will try to squeeze some long-term benefits out of it, but Carlos and I are pretty close in our judgment.
<wpwrak>
sure, the projects may die right after they're done. but you can still keep them where it doesn't cause trouble.
<wolfspraul>
of course if they are not in anybody's way
<wolfspraul>
mediawiki has strong category features
<wpwrak>
it's also nicer for students if they know their work isn't destined for destruction no matter what
<wolfspraul>
wpwrak: like I said. I'm willing to increase my respect for these works at least :-) to the level that the authors respect their own works...
<wolfspraul>
so no worries. nothing will be deleted just for fun. they are student projects. I hope they are in nobody's way.
<wolfspraul>
I think 99% of it is junk.
<wolfspraul>
but I guess in academia you need an unstoppable believe in 'talents' (ahem)
<wpwrak>
it's less about respecting the work or the authors but about putting things where they logically belong :)
<wolfspraul>
so I provide server resources, Carlos does the magic.
<wolfspraul>
well some of it may belong into the trash bin
<wolfspraul>
but not now, after they got their degrees
<wolfspraul>
we'll see
<wolfspraul>
who knows maybe something good comes out of it
<wolfspraul>
we should be optimistic, shouldn't we?
<kristianpaul>
wolfspraul: it is
<kristianpaul>
(99% talking)
<lekernel>
"some of it may belong into the trash bin, but not now, after they got their degrees"
<lekernel>
nice quote :)
<lekernel>
lols
<wolfspraul>
yeah well
<wolfspraul>
they themselves wouldn't even disagree :-)
<wolfspraul>
he he
<wpwrak>
students tend to know the limited lifespan of their efforts :)
<wpwrak>
some things have a real use, though. e.g., one of my students did some bug hunting in one of the little protocols in the ATM stack. found something in my implementation and also some things in the "competing" commercial implementation. that was useful.
<wpwrak>
then some others did things that didn't have a direct practical use but helped me to get a better understanding of the issue.
<wpwrak>
then of course, there was that one guy in finland who implemented a whole link-layer emulation stack. (LAN Emulation) wasn't my student but certainly did a heck of a work for his MSc thesis. that stuff is probably still round today.
<wpwrak>
so i wouldn't simply dismiss these projects but i'd also keep them separate from "official documentation".
<wpwrak>
wolfspraul: the glitch problem definitely is a DNS issue. seems that there's a cache somethere that forwards the query but immediately responds "not found". once the real response arrives, it's fine.
<wolfspraul>
wpwrak: hmm
<wolfspraul>
maybe outside of reach for any improvement then
<wpwrak>
wolfspraul: dunno. since it seems to affect others too, it's probably close to the source.
<wpwrak>
s/doddle/doodle/
<wolfspraul>
wpwrak: he that is fun and crazy :-)
<wpwrak>
;-)
<wolfspraul>
should we put it on the qi homepage for a while?
<wpwrak>
hmm, maybe with better data ?
<wpwrak>
also, the compressed x scale makes things look a bit weird
<wolfspraul>
let's see, maybe an mtk project will join at some point
<wpwrak>
yeah, haven't quite figured out how to merge the telephony side
<wpwrak>
of course, the overall goal is to provide you with material to show to prospective investors/sponsors ;-)
<wpwrak>
"look, we even have a roadmap !" ;-)
<wpwrak>
it's in fact rjeffries' question about xue that made me think of this. because it's indeed sometimes hard to see how the various little projects interconnect.
<wpwrak>
(even if you know what each of them is doing ;-)
<wpwrak>
phew. after five layers of python, one finally gets to the C code of gnuradio. at least there's fftw at the bottom of this well, something known
<lekernel>
wpwrak: do you think python is slow?
<lekernel>
(i'm polling everyone's opinion about that :))
<wpwrak>
lekernel: slower than C for sure. whether that it "too slow" or "fast enough" depends ...
<wolfspraul>
wpwrak: I was thinking about a 'hardware block diagram', like those software block diagrams
<wolfspraul>
only that at the top, it starts/ends with the kernel, say Linux/RTEMS/Iris, then below that are blocks for IC design, KiCad/boom, RF stuff, mechanical process and tools, etc.
<wpwrak>
lekernel: i used it for some fair amount of data processing in my TMC package and i'd rather do all this in C. it's expect a speedup of about 10x, considering that most of the operations are quite simple. (such as walking a large array, doing floating-point comparisons on the values, and generating another array with the result)
<wolfspraul>
but your left-to-right diagram is also nice
<wpwrak>
ah, a stack. a stack implies order. do we have that ? ;)
<lekernel>
wpwrak: fyi milkymist spells with just one capital M at the beginning, and started in aug 2007 (without qi-hardware at that time)
<mth>
lekernel: python is slow, but often that doesn't matter if you do the complex control part in python and the heavy lifting in C modules
<wpwrak>
lekernel: ah, thanks ! i was indeed wondering about its origin date :)
<mth>
for processing numbers it really helps to use numpy
<lekernel>
failing to be popular since 2007, yay!
<lekernel>
big R&D is frustrating, really
<wpwrak>
lekernel: ;-)))
<wpwrak>
lekernel: well, that's about to change, isn't it ? how are things going with the CCC preparation ?
<lekernel>
heh, I hope so
<lekernel>
CCC preparation is a small task :)
<lekernel>
it's rather about getting the boards produced and flashed with good stuff
<lekernel>
btw we got a prototype case lately (thanks roh)
<wpwrak>
lekernel: (boards) yeah, that's what i meant ;-)
<wpwrak>
(case) wow ! that was really quick !
<lekernel>
one afternoon basically :)
<lekernel>
there are still a few rough edges but that's really good work for a start
<wpwrak>
(big r&d) i actually wondered about the timeline. normally, an MSc thesis doesn't take nearly that long. how did you manage that ?
<lekernel>
well, I just started before, and since this had never been published in academia and looked better than a crappy thesis, I managed to get it accepted
<wpwrak>
nice case. laser cutters rock.
<lekernel>
plus there are some improvements and studies that I did for the thesis itself
<wpwrak>
aah ! good move. yup, certainly beats the usual thesis stuff by a pretty wide margin.
<wpwrak>
(case) how did they do the groove in the top/bottom plates ?
<lekernel>
mh, that was the messier part
<lekernel>
make two shallow cuts with the laser, then scrape the wood away with a screwdriver
<lekernel>
that was slow and messy
<wpwrak>
good ;-) i was already worried that laser cutters were getting way too powerful :)
<lekernel>
later we'll use a cnc
<wpwrak>
i was just about to mention that :)
<wpwrak>
with the mill it'll be easy. and they have a nice big one.
<wpwrak>
my mill only does pieces of 15 x 10 cm. so i'm no fan of "make the ben bigger" proposals :)
<wpwrak>
the sidewalls fit very nicely. you don't even seem to need a filler for the corners.
<wpwrak>
roh: congratulations to you, too ! very nice job !
<wolfspraul>
wpwrak: will you post your PDF in response to Ron's mail?
<wpwrak>
wolfspraul: not in its present state. maybe i'll make a program that generates such a diagram from raw data. that way, i'll be easier to rearrange things. i also need to verify a number of dates, etc.
<wpwrak>
wolfspraul: in any case, the answer for ron is probably more along the lines of a link to the projects list :)
<wolfspraul>
ah sure, I don't think you should get bogged down over such bureaucracy anyway
<wolfspraul>
Ron likes to lean back and ask wise guy questions...
<wolfspraul>
I appreciate the mental challenge to stay calm and friendly. That is indeed helpful. :-)
<wolfspraul>
reminds me of what I learn in China every day, so also goes along well with that...
<wolfspraul>
he's asking the right questions though, of course our stuff is not easy to understand etc.
<wpwrak>
wolfspraul: (bureaucracy) yeah, it's low priority :) first the antennas. then i might hack a faster version of rafa's jlime-pkg. then we'll see :)
<wpwrak>
yup, he asks good questions. i wish he'd also take on a redactorial/editorial role. i think that's what you need to make wiki-based documentation work. someone who keeps the big picture, steers thing in the right direction, and makes sure things connect.
<wolfspraul>
if you can activate him in that direction, we will hand you the 'ron award', the highest possible award in community contribution activation...
<wpwrak>
from the description, i take it that it is usually awarded posthumously ? :)
<wpwrak>
Ornotermes: kristianpaul found some SDR receiver for satellite and such
<Ornotermes>
oh?
<wpwrak>
Ornotermes: relatively inexpensive, estimated ~usd 150 i think. form factor: usb dongle.
<Ornotermes>
is there a link?
<wpwrak>
GRRR. if i let gnuplot do the math, it get signals of about -30 dB. if i do it in my C program, i get ~-5 dB. (the latter is actually roughly what i expect to get)
<wpwrak>
hates non-deterministic math
<wpwrak>
ah, where's kristianpaul when we need him ...
<wpwrak>
08:30 in bogota. he should be up by now.
<wpwrak>
(math) ah, my fault. forgot to scale it ...
<rafa>
wpwrak: maybe he just arrived from a night life
<wolfspraul>
he has a day job, maybe commuting to that right now?
<wpwrak>
hmm. results are still very different from what gnuradio's fft gives me. but at least they're no longer freakishly sensitive to parameter variations. that's something :)
<rafa>
wolfspraul: yes, surely. He often has the chance to be here still if he is working.. but sometimes there are really busy days as well
<rafa>
wolfspraul: I had a vague idea about dual booting to think between all.. there are a lot of pro and cons.. but well, it was an idea to put together other ideas before to define more technical details. Did you read the log about having just one kernel, just one rootfs partition and /data partition on nand for two or more distributions?
<wolfspraul>
I saw some discussions.
<wolfspraul>
not sure which one works best etc.
<wpwrak>
ah, t-shirt idea v2: front: "Eloi", with the o apple-shaped. below, "Morlock", with the k ben-symbol-shaped. back: "dinner is served".
<wolfspraul>
but it's good that we start thinking about it
<rafa>
wolfspraul: no.. I have not idea either
<wolfspraul>
and if 'many partitions' is totally outdated and old fashioned, well then, bring in the new stuff! chroot and what not, all fine by me. let's just think about the consequences first :-)
<wpwrak>
t-shirt #2 (for wolfgang :): front: "THE FLOW". back: "don't go with it"
<rafa>
wolfspraul: yes.. maybe we could put these ideas on some wiki.. so we can read from time to time to improve those
<rafa>
wolfspraul: chat forgets a bit
<wpwrak>
wolfspraul: many partitions = many more opportunities for noticing that something doesn't fit and needs resizing :)
<wpwrak>
rafa: (chat forgetting) while wiki preserves the obsolete concepts and explanations forever ;-)
<kyak>
rafa: what's the name of gui text editor in jlime?
<rafa>
wpwrak: that is true.. but at least wiki could remembers us a bit those ideas during the process to think those
<wolfspraul>
rafa: if you get up one day, and you meet Werner, and he tells you all these things about how he loves u-boot, how he loves wikis, NAND, partitions, and many other beautiful things, you know something is wrong. Maybe you are in a dream? Maybe a new world? all is possible...
<rafa>
kyak: it gtkedit or gtkeditor.. no remember exactly. Let me find the original link.
<rafa>
kyak: btw.. all the gtk applications on jlime are gtk1 applications
<kyak>
rafa: okay..
<rafa>
kyak: because those were 3 or 4 times faster than gtk2.. of course.. gtk2 is modern and full of features
<rafa>
kyak: which we do not use on a 320x240 screen
<kyak>
oh ok!
<kyak>
i'm lookng for a simalicitic QT editor
<kyak>
*simplistic
<wpwrak>
wolfspraul: when this happens, i'd be very careful to avoid any physical contact with that "werner", for he'll certainly consist of antimatter :)
<kyak>
for some reason i believe qt is faster than gtk.. and more modern too
<kyak>
wpwrak: let me know when you have xterm running in openwrt :)
<rafa>
wolfspraul: yeah.. no sure if I know him or he is part of this Alice in Wonderland  :D
<rafa>
kyak: we had a long list of nice gtk1 applications a couple of years ago.. Unfortunately.. with the lot of changes on web server and stuff... I have not found that :(
<rafa>
kyak: but those were fast as the hell
<rafa>
kyak: I remember.. at my old jornada 660x (166mhz, 16mb of ram).. running them and immediately those appeared on screen
<rafa>
sorry, 66mhz
<kyak>
rafa: that's great.. unfortunately, those are usually unmaintaned
<kyak>
only few people need such apps
<rafa>
kyak: well.. they did just a simple and very know and used task.. like editor, calculator, browser
<rafa>
kyak: so you do not need maintainers
<rafa>
kyak: also.. they were so small to do easy modifications if you want
<rafa>
kyak: like I did with the dictionary (which is gtk1 as well)
<kyak>
well yes, you are right
<wpwrak>
(kyak) heh, i thought the "thanks" was ironic :)
<wpwrak>
grr. s/(kyak)/kyak: (tricked)/
<wpwrak>
should not multitask
<kyak>
wpwrak: it could have been ironic if i wasn't multitask :)
<kristianpaul>
wpwrak: hello
<kristianpaul>
sorry i travel to work every day
<kristianpaul>
btw i live in Buga, not bogota but yes is same timezone :)
<kristianpaul>
Ornotermes: hey
<lekernel>
does Ron do anything except not using Google and asking questions that slow down everyone else?
<kristianpaul>
hangon reading baclog just now
<wpwrak>
kristianpaul: b<vowel>g*a - matches :)
<wpwrak>
kristianpaul: and i was indeed thinking of the same time zone :)
<wpwrak>
lekernel: do we really want to know ? ;-)
<jekhor>
Hi. Can someone point me to actual version of NanoNote schematic & PCB? I found at least three versions of different boards on wiki, but without comments about actuality.
<wpwrak>
lekernel: btw, i don't see your presentation on the schedule ? or will it be "the concert" ? ;-)
<lekernel>
wpwrak: it's a workshop... they didn't put them on the schedule :(
<lekernel>
actually the workshop organization is a lot messier than I thought
<wpwrak>
lekernel: ah, what are the messy bits ?
<lekernel>
well, first they let you put an entry in the pentabarf, then you get no answer from them, and finally you discover that the workshops are on a "first come first serve edit the wiki schedule yourself" mode
<lekernel>
and, as you noticed, they aren't on the schedule
<wpwrak>
lekernel: hmm, that doesn't sound very nice. particularly not for firsttimers. did you still get a good slot ?
<wpwrak>
;-)))
<lekernel>
seems I didn't get any slot so far...
<wpwrak>
the wednesday slot would be good
<wpwrak>
if people don't have to prepare anything for it, then tue, too
<lekernel>
combined with the pre-sale ticket mess, that pretty much sucks
<wpwrak>
well, there is "chaos" in the name :)
<lekernel>
yeah, grr
<lekernel>
next year => lecture :)
<lekernel>
I'm also losing a lot of visibility by not being on the schedule
<wpwrak>
grmbl. now .. my antenna sensitivity tests ... i do 100 test runs. each time exactly the same setup. what i'd expect are 100 results, bounding around roughly randomly within a certain range. what i get instead looks more like a -log curve with a bit of noise on top :-(
<wpwrak>
lekernel: yeah :-( well, maybe they'll put the workshops in a similarly visible place later.
<wpwrak>
workshops tend to be more for the in-crowd while lectures tend to draw more people who are just curious, though. so the workshop will be more of a challenge.
<wpwrak>
hmm. and then there are those freak runs where the power suddenly jumps by 8 dB. i wonder what got into the system with these ...
<wpwrak>
it looks a bit too consistent to be just an RF side problem. maybe there's a problem with the communication. well, i can debug this separately.
<wpwrak>
interesting. now i did the test with the 802.15.4 transmitter constantly running. (before i was turning it on and off between tests.) and i still get those surprise peaks. this suggests the usrp is to blame for them. grrr ...
<kyak>
kewl.. typing russian in Qt!
<kyak>
(have to hold the "Qi" button for now, Qt for some reason doesn't know how to proceed *_Lock from keymap)
<kyak>
god i'm so glad this is even feasible
<kyak>
Qt seems to be able to Lock only Caps, Num and Scroll
<kyak>
to be able to Lock CtrlR, for example, would require to modify Qt LinuxInput driver
<kyak>
hm-hm
<viric>
and what is ctrlr?
<kyak>
it's one of the keyboard modifiers, usually right control... could be anything..
<viric>
ah
<kyak>
for  Ben i've assigned it to "Qi" key
<kyak>
so it is possible to switch to alternative keymap
<viric>
in xorg?
<viric>
or the fb?
<kyak>
it works in console; and Qt has a kmap2qmap to convert it to Qt keymap (to use in linuxfb)
<kyak>
if there was X11 in openwrt, wouldn't be a problem at all
<viric>
there isn't even the tiny server?
<kyak>
there is something.. but i'm not sure
<mirko>
hey
<mirko>
kyak: ping
<mirko>
wolfgang told me anybody wrote a hello-world qt4 GUI program and it hangs/crashes as well when it is supposed to quit - however i forgot _who_ did so
<kristianpaul>
mirko: had you tried with nanomap?
<mirko>
yes
<kristianpaul>
what happended?
<mirko>
however it's far too complex for debugging qt4-issues on the device itself
<mirko>
the issue is the same though...
<mirko>
hangs and segfaults while syscalling rt_sissuspend
<kyak>
mirko: do you know why it was decided to go without X11? i understand it adds an overhead, but is it that noticable?
<kyak>
as i can see, openwrt supports X11 pretty well
<mirko>
kyak: it's overhead for features which aren't used on the NN
<calamarz>
sb has handy a url for the mxu_as awk script?
<lekernel>
kyak: name one advantage of X11 on embedded devices
<mirko>
kyak: well - the hello-world app quits fine now with linuxfb?
<mirko>
NanoMap still does not
<mirko>
kyak: show me the hellow-world app code please
<kristianpaul>
quak
<kristianpaul>
lekernel: remote X graphical terminal ()
<wpwrak>
rafa: i think jlime can start preparing the welcoming ceremony for kyak ;-)
<lekernel>
over what? SLIP/PPP? hahahahaha
<kristianpaul>
well sure not on those
<kristianpaul>
but on ethernet, well :)
<lekernel>
sure, you could use VNC or better RDP on those
<wpwrak>
kristianpaul: not so fast. they run PPP over high-speed links ;-)
<lekernel>
even the much-touted network supposed-transparency doesn't work with X
<mirko>
don't get it.. had the discussion sooo many times..
<lekernel>
it's just slow as hell
<mirko>
x11 is dead
<kristianpaul>
wpwrak: ah true :)
<mirko>
[/provoking]
<mirko>
;)
<bartbes>
wayland is the way(land)
<mirko>
xcb is promising as well
<wpwrak>
mirko: it died for the first time some 20+ years ago -and lived on as a zombie happily ever after ;-)
<kristianpaul>
rise hand who is not using X right now :)
<bartbes>
everyone on windows? :P
<mirko>
kristianpaul: X != x11 != graphical user interface
<lekernel>
I seriously wish I weren't - if only someone cared to kick that crap out instead of developing a patchwork of incompatible toolkits all depending on X
<lekernel>
unfortunately I'm stuck with an awful (and bloated) mess of Qt, GTK and even Motif apps
<bartbes>
Qt and GTK have ports to wayland, right?
<kristianpaul>
mirko: agreed
<kristianpaul>
mirko: as jlime uses xfdev relly well :)
<bartbes>
I believe the Qt port was quite advanced
<lekernel>
load three different widgets toolkit. absolutely brilliant!
<wpwrak>
mirko: let's see how Wayward looks once it has caught up on the basic feature set :) e.g., decent network support, runs well on common hardware - old and new, compatibility with the major widget sets, multi-head support, ...
<bartbes>
make it all run on cairo
<mirko>
lekernel: DirectFB support(ed) all of them...
<bartbes>
and profit?
<bartbes>
:P
<lekernel>
and add some extra layers on top of that (gtkmm, wxwidgets, ...) for extra-bloatedness
<bartbes>
and I want a ppa
<wpwrak>
lekernel: Motif ! pfui ! ;-)
<bartbes>
because my ubuntu install can be torn to pieces
<bartbes>
(preferably not, but it is already broken)
<wpwrak>
lekernel: and how is all that the fault of X ? you'll get the same quickly enough on any popular infrastructure
<lekernel>
not that much on Mac, except for ports of X applications
<lekernel>
which most Mac users hate btw
<mirko>
(which btw is not only because cocoa/aqua is so much better, but as well since x11 on macosx is so much PIA :))
<lekernel>
imo it's time to drop network pseudo-transparency and legacy crap like ICCCM and do like apple does - since apparently no one does it better
<wpwrak>
lekernel: well, mac is a bit of a special environment anyway. nice and tidy, but still much of a prison camp.
<mirko>
*popcorn*
<wpwrak>
what surprises me a bit is that people who want to replace X tend to want to replace X as their first step, before actually getting the alternative to work. e.g., the current wayward hype is all about rendering in the client and pushing bitmaps. you don't need to get rid of X to do this.
<lekernel>
why don't they put the widget set in the server? at least, this will force anyone to use the same widgets and get rid of the qt/gtk/motif/whatever mess
<wpwrak>
perhaps because people like to have a choice of widgets ?
<lekernel>
nice choice, that. it led to the complete fuckup the Linux GUI is
<wpwrak>
dunno. there is convergence on style. most of the basics look quite similar nowadays. similar enough that it's no big effort to use the common things in new applications. it used to be much worse in the old days.
<lekernel>
even ms-windows does that with some success
<wpwrak>
of course, many applications still need learning, but that's for stuff that's specific to them
<lekernel>
there are plenty of end user applications for all sorts of purposes, nearly all using the same widget set
<lekernel>
it's only on *nix that you have this mess around the GUI
<wpwrak>
you can probably just stay in one of those worlds and be happy. i'm sure gnome and kde each have the same set of basic utilities.
<lekernel>
and i'm sorry to say, but the x11-based GUIs are slower than the windows xp GUI
<lekernel>
(despite what the "recycle your old PC with Linux" people say)
<wpwrak>
i use my desktop to do work, not to run benchmarks, so i don't really know how slow it is ;-)
<wpwrak>
(and yes, i can watch movies in good quality on it too :)
<kristianpaul>
awesome !
<kristianpaul>
hides
<rafa>
lekernel: one advantage of xfbdev on little devices is that it works out of the box. We have around 10 different WM to use and control applications running at the same time.. Then there are many applications written during 15 years to test and use. Check this : http://www.youtube.com/watch?v=uKAa_JbEokk .. WE did this env without to write a just one line of source code.
<rafa>
lekernel: there a lot of problems and cons.. i know.. but in comparison.. i see that qi-openwrt uses gmenu2x.. a crapy thing
<rafa>
which has improved a lot writting/modifying it
<rafa>
if they want to change gmenu2x there is not choice
<rafa>
and they spending months and months improving that
<rafa>
well. with xfbdev there we have around 10+, 20 perhaps differents goods GUI menus to use as launcher or similar..
<rafa>
lekernel: but again.. I know that there are a lot of cons with X.. but we used it because we do not need to write any line of source code... it just works..
<lekernel>
that's what I said: everyone focuses on the X11 crap
<rafa>
and we have a lot of applications there to use and we like it
<lekernel>
so in the end you have plenty of apps
<lekernel>
major bugs are fixed
<lekernel>
etc.
<lekernel>
but that doesn't make X good per se
<lekernel>
it just makes it a necessary evil
<rafa>
lekernel: yes.. of course. I understand you..
<rafa>
lekernel: and we use it at jlime because we are not a lot of people and we are not good developing new stuff.. we are godd  using and setting the current stuff
<kristianpaul>
okay i hope i can play videos on milkymist that will be awesome :D
<qi-bot>
[commit] Werner Almesberger: usrp/range: obtain avg/min/max from a series of values and eliminate outliers http://qi-hw.com/p/ben-wpan/c4d906b
<qi-bot>
[commit] Werner Almesberger: usrp/README: description of the preparation for antenna measurements. http://qi-hw.com/p/ben-wpan/cc5090b
<mirko>
it hangs while issueing the rt_sigsuspend system call
<mirko>
seems uclibc related
<mirko>
since we switched back to uclibc 0.9.30 we lost the improved threaing support
<mirko>
*threading
<wolfspraul>
is it possible that the problem would go away if we switched back from at 4.7.0 to 4.6.2 ?
<mirko>
no, most likely not
<mirko>
it's not qt's fault i think
<mirko>
and we can't switch back that easy
<wolfspraul>
sure, maybe we should 'switch' forward ;-)
<mirko>
wolfspraul: well, i did once (uclibc 0.9.32) and it did break SDL ;)
<mirko>
wolfspraul: which needs to be evaluated once again - much improvements and commits happened to uclibc since then
<wolfspraul>
there was a small msmtp patch Xiangfu sent to the openwrt mailing list a few days ago
<wolfspraul>
I havent' followed whether someone thought it's good and applied it
<wolfspraul>
it was about adding 3 scripts to the package (a patch to the Makefile)
<mirko>
hm, i don't think it's release critical
<wolfspraul>
he, no :-)
<wolfspraul>
it's just a patch...
<wolfspraul>
but those kinds of small things are either applied right away or forgotten
<mirko>
didn't see it yet - just got headaches how to handle the openwrt-packages-situation :)
<wolfspraul>
you set the openwrt-packages ncurses to broken because of a conflict. I don't know what the reason behind the openwrt-packages ncurses was, I think I saw kyak posting something about it here in irc...
<mirko>
well, it broke alsa do i marked it as broken
<wolfspraul>
I think lars did something on the alsa side.
<mirko>
no, alsa didn't get changed
<mirko>
at least not in the "testing"-branch (fortunately it didn't get changed :))
<wolfspraul>
mirko: how does the versioning of feeds work?
<mirko>
wolfspraul: what do you mean?
<wolfspraul>
for example you reverted a commit in openwrt-packages now, because it didn't match the openwrt-xburst/testing branch
<wolfspraul>
so how can the openwrt-xburst/testing branch (or any OpenWrt build) pin down a feed at a certain revision?
<mirko>
wolfspraul: i created a "testing" branch in openwrt-packages.git as well
<mirko>
and based it on the commit i created the testing image with
<wolfspraul>
we can do that because we can create that branch, but what if the feed is maintained by someone else?
<mirko>
so i didn't need to revert anything
<wolfspraul>
is it possible to 'pin' a feed at a certain revision?
<mirko>
you mean inside feeds.conf(.default)?
<wolfspraul>
yes, for example there
<mirko>
no, there's no functionality like that (yet)
<mirko>
that's why i created the testing branch
<mirko>
think that's the most elegant way
<wolfspraul>
yes, for now it makes sense
<wolfspraul>
mirko: so a general revision capability will come to feeds.conf at some point?
<kristianpaul>
Xbusrt SIMD for FFTW vs FFT Aceleration on Fpga, lets see wich one got first ;-)
<wpwrak>
hmm, how big a stopwatch will i need ? :)
<kristianpaul>
hope no so big, first PRN code match is schedulled for 15dc acording to Free GPS Stakc wiki page
<kristianpaul>
15 dec
<wpwrak>
ah, so you can just lean back and in one month it'll happen :)
<kristianpaul>
he well is not so that way as you alrady know
<kristianpaul>
FFT on fpga is great an easy using IP cores, i'm NOT going do that, long way is DIY or look what is already done..
<kristianpaul>
SIMD well at least have some instructions documented..
<kristianpaul>
well about FFT on FPGA i need to do more serch..
<kristianpaul>
ah sure, i can run fft on my simd support amd procesor first and see waht i got
<wpwrak>
why do the fft on the fpga ? just grab a stream of data, send it to the pc and process it there
<kristianpaul>
is a "planned" date
<wpwrak>
once happy with the result, see if you can do the same on the ben
<kristianpaul>
sure
<kristianpaul>
thats first plan
<wpwrak>
(amd) yup
<kristianpaul>
the shedulle said before PRN thing, get raw data
<kristianpaul>
now i'm reading what format to use in order get life easier to gps-sdr
<wpwrak>
floats :) fftw likes floats ;-) even more if they're complex. and double, i guess, too :)
<kristianpaul>
but wait, i just have a pair of 2 bit signed data..
<kristianpaul>
yes is complex
<kristianpaul>
was real before some secret magic in the chip suposed to make my like easier??..
<kristianpaul>
s/like/life
<wpwrak>
it's probably just how the adc is implemented :)
<kristianpaul>
yeah sure
<kristianpaul>
the little drawing in the datsheet said that too
<wpwrak>
so you need to find algorithms that work with this sort of data. i'm sure they exist ... somewhere :)
<wpwrak>
nore sure if fft is really part of it, though
<kristianpaul>
sure lots
<kristianpaul>
fastgps
<kristianpaul>
gps-sdr
<kristianpaul>
and the non avaliable for me mathlab code of the books i'm reading
<kristianpaul>
wpwrak: it is !!
<kristianpaul>
at least i want do serial correlation on time domain fft is the way
<kristianpaul>
what i know i ahve to do is 1, precalculate some fft acording to one PRN code i'll try match, the multiply this by the fft from the data received, then i may get a peak (amplitude) wich show i got the "Gold Code" so i can jump data correlation steps (now real data from GPS !! :D)
<kristianpaul>
thi guy gives better idea or what to follow in some way of course
<wpwrak>
ok. you're already hovering around the edge of my signal processing knowledge :-)
<kristianpaul>
nah, just the basic concept, after skiming some books,  wait me for when i have to learn about satellite tracking and ephemeris
<kristianpaul>
that will be fun
<kristianpaul>
but now step by step i still stuck on data aqusition
<kristianpaul>
and YES i will make posible as fast as i can raw data for other people to look at
<kristianpaul>
well i'm off, read you later
<kristianpaul>
but look FFT is one of the most imporant task will may consume resources, next steps will require advancend logic not worhtly to implement in hardware or acelerate some how
<kristianpaul>
and so far our nanonote is not dual-cored :)
<wpwrak>
fft will be a bit of work :)
<wpwrak>
well, there are many optimized variants ...