<whitequark>
DocScrutinizer: I was playing with that chinese card and found an interesting glitch: if I set up netcat to send zeros over udp (feeding from 100M ethernet), that card easily sends/receives more than 92 Mbits of traffic in each direction
<whitequark>
contains Ralink RT3070L and a custom 1W ASIC amplifier
<wpwrak>
DocScrutinizer: his 1 W WLAN ghettoblaster ;-) (designed by the chinese from the dirty side of town)
<DocScrutinizer>
hah
<DocScrutinizer>
of course the card and usb cable are part of the antenna
<whitequark>
huh?!
<whitequark>
the _USB_cable_ is a part of antenna?
<wpwrak>
rejon: (wiki sections) pheew. manana. already too much word-wrestling for today :)
<DocScrutinizer>
there's no such thing like single-ended antennae - it's always a dipole of one kind or another
<DocScrutinizer>
that's what nobody really got when I explained where buzz of GTA02 came from
<whitequark>
so then, the antenna, card itself, my netbook, me, on whom the netbook lies, form one great kind of antenna?
<whitequark>
weird
<DocScrutinizer>
as there's no battery wit just one pole, there's no transmitter with just one antenna pin. It's always a balanced output, and if you don't use a symmetrical dipole for antenna, you connect one of both ends to "GND", whatever that means
<DocScrutinizer>
whitequark: yes
<whitequark>
... and the wireless performance will then depend on USB cable quality?
<whitequark>
and length
<whitequark>
and thickness
<DocScrutinizer>
partially, yes
<wpwrak>
whitequark: and exact distance to the moon ;-)
<DocScrutinizer>
you can see what things really are like when you look at large groundplane marconi radiators
<whitequark>
then, the whole idea of wifi dongle is flawed fundamentally
<DocScrutinizer>
nah, why?
<wpwrak>
whitequark: your laptop is your ground plane
<DocScrutinizer>
usually your real gndplane of the circuit forms the gndplane for the antenna
<whitequark>
well, you'll then never get good performance without proper grounding, and you'll never get one when plugging into some unknown laptop
<whitequark>
or am I wrong here?
<wpwrak>
whitequark: your laptop will have quite a lot of ground :)
<DocScrutinizer>
the size of active gndplane is hardly larger than the antenna itself
<wpwrak>
DocScrutinizer: hmm, TI claim that the size of the box you connect to matters (in their description of the PCB antenna i used)
<DocScrutinizer>
the virtual antenna size though, or simply the wavelength
<DocScrutinizer>
everything >2 or >3 lambda is irrelevant
<whitequark>
the dongle itself is about 80mm in length, while the wavelength is 125mm
<DocScrutinizer>
:nod:
<DocScrutinizer>
so the ground via USB will matter
<DocScrutinizer>
as you see with your ghettoblaster ;-)
<DocScrutinizer>
o/
<DocScrutinizer>
needs some food
<DocScrutinizer>
whitequark: placing the dongle on a steel table or tin foil may help a lot for performance
<DocScrutinizer>
should extend from footpoint of antenna all directions at last one wavelength
<DocScrutinizer>
of course horizontally, while antenna vertical
<whitequark>
I should make a tin foil hat probably
<DocScrutinizer>
I actually knew a guy who mounted his CB car lambda-5/8 on a copper dish of lambda/2 diameter and placed that contraption at his attic
<DocScrutinizer>
he claimed it to be the best antenna for 27MHz he ever had
<whitequark>
is testing the tin foil thing
<SpeedEvil>
That's a lot of copper
<whitequark>
hmm, looks like my testing method (nc -u server 1000 </dev/zero) isn't very good
<whitequark>
ping: sendmsg: No buffer space available
<whitequark>
never seen anything like that
<whitequark>
DocScrutinizer: placing it on a quite large sheet of tin foil actually halves the bandwidth
<DocScrutinizer>
whitequark: then the antenna of that device might be tuned to have no such classic gnd
<DocScrutinizer>
the problem probably is you make the antenna better with that gndplane, but same time you detune it, so it's a better antenna for the wrong frequency
<whitequark>
there's also some network stack (?) issue
<DocScrutinizer>
SpeedEvil: (lot of copper) indeed ;-P
<whitequark>
when I do the udp-flooding from Ethernet side of router, the card receives 93Mbps without any problems
<whitequark>
but when I do the same from the wifi side, it chokes just on the laptop
<whitequark>
tcp works fine, but the bandwidth is around 40Mbps
<aw>
wpwrak, and is "(set-render-type! 2)" a 'Normal' or 'High quality' type?
<antoni`>
Hi all, I have read that the project atben is not wifi because there is not a real libre 802.11 hardware. Is this true?
<antoni`>
I use my wifi with free drivers...
<SpeedEvil>
It's not that simple.
<mth_>
antoni`: if I remember correctly from a discussion earlier tonight, the problem is that there is no hw with public specs that is suitable for mobile devices
<SpeedEvil>
The firmware is invariably closed
<SpeedEvil>
Well - almost invariably.
<SpeedEvil>
And the ones that are open or reverse engineered are not suitable.
<mth_>
if you're interested in the details, please check the logs, they're online at the Qi site
<roh>
well.. i think its many factors. one is documentation available, but also chip availability, power useage etc.
<DocScrutinizer>
invented another nice term (along with and same semantic cloud as "to tepco")Â Â >>man, are you poettering?!"
<DocScrutinizer>
'sth. is poettering' means 'it behaves invasive, destructive, and generally in an unpleasant pathological erratic way'
<DocScrutinizer>
this cancer is extremely poettering
<wpwrak>
aw: gerb2ppm input.pho output.pbm
<wpwrak>
aw: (warning) this one generates bitmaps, not those nice images. it's meant to be used as a building block in the future
<aw>
wpwrak, hi thanks, how about mixing them together when using 'Normal' & 'High quality' and gerb2ppm can also do it?
<aw>
wpwrak, strange. I found your time is 3:40 in Buenos Aires?
<wpwrak>
aw: (type 2) is should be "high quality", but i think gerbv more or less ignores that argument
<aw>
wpwrak, okay... so do i need to re-git clone your eda tool? or just download that 'gerb2ppm'?
<wpwrak>
aw: if you want to make the high quality images, you're need a different script. gerb2ppm is intended as a building block for future use. (i don't like how gerbv mixes the colors - you can get nice results, but it's very hard to get it right)
<wpwrak>
aw: (3:40) yup. had a quick nap before. woke up and decided to have a look for what's going on in the world :)
<aw>
wpwrak, yes, i saw that great one. :-) so that 'prettygerbv' also is done by 'gerb2ppm' or not the same one. or it specifically for altium m1? sorry i repeat it. :-)
<aw>
wpwrak, yeah...very late there. :-)
<wpwrak>
aw: the one in fab/ is for kicad. the one in fab/m1/ is for altium.
<aw>
wpwrak, okay. got it, thanks. u should be asleep now. :-)
<wpwrak>
yeah, will be soon :)
<DocScrutinizer>
(very late) nah, very early
<DocScrutinizer>
here it's even more late, or less early (but somewhat still too early in the morning)
<aw>
DocScrutinizer, hi, very later = very early; east = west. :-)
<DocScrutinizer>
hehe
<DocScrutinizer>
and at 180° lon one step can cost you a whole day
<DocScrutinizer>
funny thing, they seem to have managed it in a way this date-limit hits amazingly few solid ground
<DocScrutinizer>
from 65°-69°N there's a bit where you actually can walk across the line
<dvdk>
kyak: think i got it.  just filtering out all arrow messages at the lowest level
<dvdk>
now maybe we also need mouse wheel emulation?  haven't found out how to scroll via the keyboard yet
<kyak>
is it actually used?
<kyak>
it could be done via Volume keys :)
<dvdk>
btw there's no pgup/pgdown on the nanonote?  Red+volup/down doesnt seem to work
<kyak>
where did you try it?
<dvdk>
for example console scrolling
<dvdk>
(shift+pgup/pgdown)
<kyak>
this works for control scrolling
<kyak>
try shift+volume keys
<dvdk>
you mean without red?
<dvdk>
btw how do i exit from showkeys?  reboot?
<dvdk>
ah, eventually exuited
<dvdk>
(but console unusable now :/
<kyak>
sure, without red
<dvdk>
but i can scroll. nice
<dvdk>
didnt know
<kyak>
you don't want to press three keys to scroll the console, do you? :)
<kyak>
to exit the showkeys you just do nothing for 10 sec.. unfortunately, this console is ruined anyway :)
<kyak>
trying ase now
<kyak>
(always want to say "Arse") :)
<dvdk>
me too :)
<kyak>
it's great!
<kyak>
you disabled the arrows completely, right?
<kyak>
causes some inconvenience on dialogs
<kyak>
but it's ok
<dvdk>
no, you arrow events are let through if combined with 'shift'
<kyak>
ah ok
<kyak>
so anytime i want a keyboard arrow, i'll press the shift+arrow
<dvdk>
yes, but the shift will get through, too.  don't see a problem though
<kyak>
you made pointer acceleration good, it feels right :)
<dvdk>
kyak: it's a physics engine :)
<kyak>
i can draw shit on Ben now
<dvdk>
(i.e. accelerated motion with kind of drag)
<kyak>
do you mean you used the liballegro physical engine for that?
<dvdk>
:) nope, liballegro is too oldschool to have physics.  just look at the patch to see what i mean 'ak' is acceleration on keypress 'ad' acceleration due to drag, when not pressed
<dvdk>
so if you're quick you can even draw rounded shapes
<kyak>
that's pretty hard :) i fear i'll break the keypad
<dvdk>
maybe reduce the acceleration? :)
<kyak>
good it's fast liek hell...
<kyak>
i'm not sure, perhaps the cursor should have some inertia?
<kyak>
this way one could draw round shapes
<dvdk>
but it has inertia.  this is why it accelerates with a != infinity :)
<dvdk>
(but a is too close to infinity currently?
<dvdk>
what you mean is maybe drag is too high (i.e. deceleration when you released the key)
<kyak>
perhaps the deceleration is too fast
<kyak>
i.e. it stops momentarily
<dvdk>
just reduce 'ad', try 1 or 2
<dvdk>
(just patch the patch :)
<kyak>
i'll try that
<kyak>
dvdk: are you aware about other liballegro apps?
<dvdk>
there are not too many good games, i'm afraid :(Â Â
<dvdk>
(just updated the link in the applications wiki page)
<kyak>
all right, i'll have a look
<dvdk>
oops, pressing Fn+2 crashes
<dvdk>
looks like the NN's number keys cannot be used for zooming?  hmm, do they generate keypad codes?
<kyak>
hah, ad=1 turns pointer uncontrollable, feels like mario :)
<kyak>
might set to something higher
<dvdk>
(like mario): ROFL
<dvdk>
looks like zooming is impossible without mouse wheel nor number keys
<kyak>
any physical key generates some code, and then a modifier can be applied (like "fn" in case of Ben's numbers)
<dvdk>
Pressing Fn+1 is interpreted as Alt+N.  maybe allegro's keyboard driver is too low-level?
<kyak>
it probably uses it's own keyboard driver
<dvdk>
yes, it certainly is,  it uses sort of scan codes (maybe input events), that are then put into messages by the 'jinete' gui system that comes with ASE
<dvdk>
OUCH
<dvdk>
that means: more hacking
<dvdk>
yes, i cannot write numbers in ase's text input fields
<dvdk>
will have to look at allegro'skey driver.  different layouts can be configuerd
<kyak>
yeah, i noticed it has it's layouts
<kyak>
i tried selecting Russian, but nothing happened :)
<dvdk>
but thought those aren't used with unix, only with the dos port
<dvdk>
maybe something else going on
<dvdk>
ah, allegro source lkeybd.c
<dvdk>
it maps
<dvdk>
0x02 to key_1, isn't that supposed to be right?
<dvdk>
maybe it fails because 'alt' is pressed.  why does 'fn' map to alt??
<dvdk>
shoudln't Fn be hidden completely?
<dvdk>
ok, it even uses kernel's keyboard map:Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â Â ioctl (std_keyboard.fd, KDGKBENT, &kbe);
<dvdk>
but maybe wrongl?
<dvdk>
looks like it doesn't pass Fn as the right keboard map selector to the ioctl?
<kyak>
Fn is mapped to CtrlL
<dvdk>
ok, where can i find a readable description of NNs keymap?
<bartbes>
dvdk: btw, regarding wayland on fb, don't think so, one of the primary features of wayland is more direct access to opengl etc, also, doesn't wayland depend on stuff like KMS?
<DocScrutinizer>
dvdk: uh? 4 + (1<<6) ?
<DocScrutinizer>
probably you mean (4+1)<<6 ?
<bartbes>
DocScrutinizer: I think he means 68, so that would be 4 + (1<<6)
<DocScrutinizer>
hmm, anyway I'd write that as 1<<6 + 4 then
<kyak>
dvdk: they should be F11/F12
<wpwrak>
oh, today is thursday. funny. thought it was already friday. my inner calendar is really messed up :)
<wpwrak>
tuxbrain_HxxHhzo: nice ... that's actually a valid sequence ;-))
<dvdk>
the ASE keycode fixing odyssey continues
<dvdk>
somebody knows which keycode Fn+Number generates?  Could it be that those generate keyPAD-numbers, not keyBOARD-numbers?
<wpwrak>
dvdk: btw, seems that /. dropped your submission. next try maybe tomorrow ?
<dvdk>
wpwrak: what about saturday?  maybe less standard news volume.  saturday worked nicely last time (ubb-vga)
<dvdk>
btw article needs lots of fixing.  don't like how it was changed for technical correctness.  "WIFI spec"?  don't think that normal people are going to know what is meant (?)
<wpwrak>
dvdk: (fixing) yeah, could use some streamlining. each word you can delete without changing the meaning is good ;-)
<wpwrak>
dvdk: (custom kbd drivers) hmm, what should i say with all my user-space bit-banging drivers ? :)
<dvdk>
thinks that the orthography of tuxbrain's WPAN pages needs to be fixed before posting any news.
<dvdk>
looks over tuxbrain's page and cannot find any errors. already fixed?
<dvdk>
wpwrak: but liballegro's drivers really implement an extremely productive API
<dvdk>
just test for stuff like key[KEY_LCONTROL] to see whether a key is currently pressed.  no limits for multiple keypresses (other than what the hardware enforces)
<wpwrak>
dvdk: (tuxbrain's page) the prices i the shop still don't seem to indicate the tax/export status
<whooo>
hi, I'm getting alot of illegal instructions and segfaults when trying to when trying to running debootstrap in a debian chroot on the nanonote, what could be the reason, problems with the FPU emulator?
<dvdk>
whoo: hmm, did the last kernel drop FPU emulation support?
<whooo>
dvdk: don't think so
<dvdk>
whooo: i once installed debian via debian-installer in chroot without problems.  very long time ago, though
<dvdk>
whooo: which debian suite are you using?
<dvdk>
whooo: setup the chroot correctly?  mounting /proc etc?
<whooo>
tried wheezy and sid
<whooo>
well, no, not really, let me try it
<whooo>
neither proc or sysfs seemed to help, then there is that if I run the command a couple of times, it works
<dvdk>
wpwrak: i find this wiki page much better than any /. article could ever be :)
<rejon>
should have a rant off between wpwrak and wolfspraul
<wpwrak>
dvdk: err ... that may actually be easier ;-)
<wpwrak>
rejon: (acawiki) sounds like a BIG undertaking :)
<rejon>
all worthy projects are :)
<rejon>
its contract for fabricatorz
<rejon>
but also good project
<rejon>
i see the path in a lot of these projects, just comes down to time and money :)
<wpwrak>
rejon: we should just get over with it and print our own money ;-)
<rejon>
ask wolfspraul about bitcoin ;)
<rejon>
we are definitely generating value in one domain ;)
<wpwrak>
rejon: naw, nothing as complicated as bitcoin. issue "rejon's legal tender, one unit, serial 1/1000" and digitally sign it. done. you've created a unique "coin".
<wpwrak>
rejon: and it's a well-defined quantity. then it all depends on why accepts it and who not. having some currencies declines is perfectly normal. try buying a train ticket in europe with yen :)
<wpwrak>
declineD
<dvdk>
good night
<wpwrak>
rejon: so it's all a question of convincing people that your currency actually has value and that its value will stay roughly the same or maybe increase. if it promises to drop, people will flee the currency, as they do with state-issued currencies. there's really no difference.
<mth>
do the NN distros use Dropbear? if so, are you happy with it?
<mth>
Dingux uses telnetd at the moment, but we might want to switch to an SSH server instead
<wpwrak>
mth: yes, it uses dropbear. i hate dropbear. learned to hate it at openmoko, when trying to write non-trivial scripts that would function with it predictably. openssh. there can only be one.
<mth>
is the memory overhead of openssh not too much for a 32MB machine?
<wpwrak>
dunno. didn't try to write on-trivial scripts on the ben yet, so i haven't ended my suffering yet :)
<wpwrak>
mth: but i can't really imagine that it could overwhelm the machine. maybe some space-starved 8 MB system.
<mth>
the Dingoo has 32MB (some have 64MB), but SSH would be used for debugging and the program being debugged might take up a lot of space
<roh>
dropbear works.
<mth>
we do have zram swap though, so effectively we have a bit more than 32MB to use
<roh>
if you need key auth and or x-forwards, key management, scp, sftp... replace it with openssh. not much of a choice
<mth>
key auth is a must: I don't want to have to type my password every time
<roh>
also openssh isnt that huge.
<roh>
leave away other senseless crap like games ;)
<mth>
eh, games and emulators are 80% of what the Dingoo is used for
<roh>
meh
<roh>
uh.. dropbear supposedly can handle keys.. thats new to me
<wpwrak>
dropbear has key auth. it's just different from openssh's (different file locations)
<wpwrak>
the problem are more subtle things. e.g., when it tries from the terminal or how it passes the exit code.
<mth>
the main use is a developer being able to log in and view a log file or run gdb or strace
<wpwrak>
roh: ssh ben 'cat >>/etc/dropbear/authorized_keys' <~/.ssh/id_rsa.pub
<mth>
I don't expect very complex scripting to be done, although you never know
<wpwrak>
mth: things always get more complex, never simpler ;-)
<mth>
yep :)
<mth>
although the main thing right now is whether we should switch away from telnetd
<mth>
I guess switching from Dropbear to OpenSSH later would have a much smaller impact
<wpwrak>
mth: that's an easy one: YES ;-) you should have done that some 1-2 decades ago :)
<roh>
wpwrak: whatabork
<wpwrak>
roh: mh ?
<mth>
most people don't put their Dingoo on a routable IP address, so the danger isn't too large, but it's still not good
<roh>
wpwrak: well .. /etc often isnt writeable on such systems but only has a few symlinks to a /var that is.
<wpwrak>
mth: yup. just a bad habit some old folks used to have. like chewing tobacco :)
<mth>
by the way, currently the Dingoo runs a DHCP daemon to give the PC an IP address
<wpwrak>
roh: well, that's how it works on the ben (openwrt and jlime). ymmv :)
<mth>
we could use link-local addresses instead (cleanly in IPv6 or hackish in IPv4 with 169.254.x.x)
<mth>
but then how would the user know which IP address to connect to?
<wpwrak>
mth: now you just need a persistent and pre-loaded web cache and you don't even need that internet access :)
<mth>
I guess we could make an offline wikipedia package
<mth>
that should keep people busy for a while, they don't even notice they don't have the whole internet :)
<wpwrak>
make a google and facebook "currently out of order - please come back later" page and it may take people a while to notice ;-)
<mth>
I guess zeroconf could be used to advertise services on a link-local address, but avahi does not seem to be very lean (or am I mistaken?) and there aren't a lot of clients with zeroconf support except from Apple
<roh>
mth: lean?
<roh>
mth: basically everything which uses mdns and not an apple product uses, links against or communicates wit avahi (e.g. via dbus) to make things work.
<mth>
roh: I mean that avahi + dependencies would have a rather high memory footprint
<mth>
for an embedded device
<mth>
but I didn't actually test it, so I could be mistaken
<roh>
mth: not really. you can compile it quite minimal.
<mth>
ok
<roh>
i have it on my opentwrt router.
<roh>
and all other implementations are much worse, more buggy and not smaller afaik.
<roh>
even apple people started noting that its much more useable than their own libs
<roh>
i use it as an mdns-repeater on my router
<roh>
wifi and wired lan are different ethernet segments
<mth>
I too, but by router is a PC with a Geode processor, so not in the same category as a mobile device
<mth>
but is there support for zeroconf in command line apps or only in GUI apps?
<mth>
could I do "ssh dingoo" for example if the Dingoo would advertise its sshd service?
<roh>
dingoo.local
<roh>
usually one puts the mdns stuff behind .local
<roh>
/etc/nsswitch.conf has
<roh>
hosts:Â Â Â Â Â Â Â Â Â Â files mdns4_minimal [NOTFOUND=return] dns mdns4
<roh>
in it on ubuntu
<mth>
ah, that's interesting
<mth>
on openSUSE it's like this:
<mth>
hosts:Â Â Â Â Â Â Â Â Â Â files mdns4_minimal [NOTFOUND=return] dns