<kyak>
wolfspraul: what i mean is that feeds/packages/Xorg/lib/qt4 is coming from "upstream"
<kyak>
and from time to time it is broken due to some commit there
<wolfspraul>
ok but which upstream?
<kyak>
openwrt
<wolfspraul>
which repository are you working against, which branch?
<kyak>
it comes from openwrt packages directly
<kyak>
Updating feed 'packages' from 'svn://svn.openwrt.org/openwrt/packages'
<wolfspraul>
do you compile from source or are you talking about problems with packages others have built?
<wolfspraul>
is that the OpenWrt development trunk?
<kyak>
yes.
<wolfspraul>
ah OK
<wolfspraul>
well you should probably expect some breakage there, no?
<kyak>
no
<kyak>
i expect them to test, then commit
<wolfspraul>
I don't know exactly how mirko works on qt4 stuff, either he works there, and then merges into the stable branches, or the other way round?
<wolfspraul>
if that's the equivalent of an unstable repo, maybe the testing is just 'in progress'
<wolfspraul>
I want to start doing some testing of my own on the NanoNote, and document it, but it will only be based on what we have in openwrt-xburst and the package feed on qi (which is months old right now, so doesn't really help...)
<kyak>
hm.. i'm surprised to hear this from you
<kyak>
actually, qi-packages are updated quite freqeuntly, just have a look at "latest updates"
<wolfspraul>
misunderstanding
<wolfspraul>
I mean a repository of pre-built binary .ipk files
<wolfspraul>
did we update it? if so I indeed missed it, let me check...
<kyak>
ah, it's updated together with the latest built
<kyak>
which was.. maybe you are right, a month ago
<wolfspraul>
we are talking about different things
<wolfspraul>
I think
<wolfspraul>
anyway I am trying to understand how you can feel better about qt4, because I think mirko is doing quite a bit of work there, and if you sync where you look with mirko, maybe things will get better? don't know
<kyak>
hmm, looks like packages weren't rebuilt with the latest image..
<kyak>
i'm not sure it is only mirko who commits to openwrt packages directly
<wolfspraul>
you mean to openwrt.org ? no sure not there are many people I think
<wolfspraul>
what are you trying to achieve?
<wolfspraul>
by installing prebuilt .ipks from openwrt.org you are leaving what anybody tests on a NanoNote, even tries to test :-)
<wolfspraul>
you are probably also not staying inside the upstream Backfire release, do you?
<kyak>
i try to build the latest opwnert-xburst image
<kyak>
from our git
<kyak>
but building of qt4 fails due to some previous commit
<unclouded>
kyak, we guessed yesterday that the pre-built .ipk files were automatically built along with the official release images but it seems that the repository of .ipks can be rebuilt independently of the official image releases
<kyak>
unclouded: sure, yes
<unclouded>
( not that's directly relevant to the current discussion)
<kyak>
i don't pay a lot of attention to pre-built .ipk files because i usually include all the software i need in my custom image
<kyak>
tslib.cpp:42:19: error: tslib.h: No such file or directory
<kyak>
this is the error building qt4
<kyak>
i don't knwo what was the update, but someone definitely missed some dependencies
<qi-bot>
[commit] Werner Almesberger: Introduce PLATFORM_ENTER, replacing the old PLATFORM_SETUP for DFU mode http://qi-hw.com/p/f32xbase/5ce904b
<qwebirc26516>
I seem to be having trouble with getting into usbboot. I have tried different cables (of which at least two positively worked with the unit before), shortening the usb-pins, not shortening them etc. No avail, I cannot get back into usb boot mode.
<qwebirc26516>
Now, this happened after I (apparently successful :(Â Â ) erased the NAND as per the wiki instructions, in preparation to re-flashing.
<kyak>
what do you see in dmesg?
<qwebirc26516>
Any ideas what else I can try? I have tried on two ubuntu machines so far, no avail
<kyak>
also, the suggested way is reflash_ben.sh.. why do you want to do it by hand?
<qwebirc26516>
reflash_ben.sh thre errors and did not proceed with flashing, thats why I cose to use usbboot.
<qwebirc26516>
thew
<qwebirc26516>
threw. damn ;)
<kyak>
what about dmesg?
<qwebirc26516>
dmesg is /var/log/messages - no messages, there is no new device present.
<kyak>
is your usb port functioning?
<qwebirc26516>
strange, isn't it...
<kyak>
try to connect other device to the same port
<qwebirc26516>
yes, I have checked that (and, used different ports, and, as mentioned, two different machines).
<qwebirc26516>
ie. a USB harddrive, using about the maximum power, works fine as do usb thumbdrives.
<kyak>
do you see the red LED when you plug in Ben?
<cts>
better.
<cts>
so, just to make suer I am not getting anything wrong here: I do have to take out the battery, wait, then connect the USB cable, then shorten the usb pins, then press power for two seconds, right? then I should see the USB device in lsusb (aas I did just this morning, before desaster struck...)?
<kyak>
no need to shorten. if your bootloader is not corrupted, just hold the "U" button while pressing the power on for two seconds
<cts>
someone just worked magic on my device. after trying at least 20 times, now I seem to be in usb boot mode (and, btw, the power-U mode didn't work for me, sorry).
<cts>
Ok, I'm good, the official version re-flashed fine. Thanks for holding your hands over my head ;)
<orly_owl>
What is needed to build a libre hardware webcam?
<wolfspraul>
orly_owl: how libre do you want to have it?
<orly_owl>
As much as possible.
<wolfspraul>
he :-)
<orly_owl>
I sense mocking.
<wolfspraul>
you need to tell us more about your background and what you try to achieve
<wolfspraul>
oh no
<wolfspraul>
totally not
<wolfspraul>
the best free camera I know right now is the Elphel 353
<wolfspraul>
but it's expensive and can hardly be called a 'webcam'
<orly_owl>
ok
<orly_owl>
yes i know it
<wolfspraul>
it's really a _long_ story with the 'libre'
<wolfspraul>
should the entire production process be openly documented? should all machines using during production be openly documented? should the entire production process be free of patents?
<wolfspraul>
those things are nearly impossible
<orly_owl>
I hear that the chipset/cpu it uses can output ogg video, which is nice.
<wolfspraul>
unless you want to go back to something like the flintstones
<wolfspraul>
yes and no
<wolfspraul>
the older Elphel 323 had an Ogg Theora encoder in the fpga, but it was never forward-ported to 353
<orly_owl>
looks up fpga
<orly_owl>
k
<wolfspraul>
just to finish the 'libre', some projects that are not satisfied with the state of things in the IC world use fpgas
<orly_owl>
why did they drop it
<wolfspraul>
but fpgas themselves are super proprietary ICs
<orly_owl>
:/
<wolfspraul>
yeah well
<wolfspraul>
if hardware is neglected for 20+ years that's what happens
<wolfspraul>
some people think you can design a 'freedom box' with Marvell technology :-)
<orly_owl>
it becomes rubbish like mr pc bios?
<wolfspraul>
Gandhi shopping for some AK47s...
<orly_owl>
:o
<orly_owl>
now that's an analogy i understand!
<wolfspraul>
when you say 'webcam', what specs and pricepoint do you have in mind?
<wolfspraul>
and why libre, which part should be libre?
<orly_owl>
cheap as hell, VGA
<wolfspraul>
the camera modules foudn in pretty much all webcams are proprietary ICs that an just give you jpegs
<wolfspraul>
it's a CMOS image sensor + IC logic to create JPEGs
<orly_owl>
the internal firmware should be libre abd the circuitry
<wolfspraul>
is that libre enough?
<wolfspraul>
well that's why I'm asking
<orly_owl>
well i guess i want a libre IC
<wolfspraul>
for the Elphel cameras this type of openess/freedom would by far not be enough
<orly_owl>
oh
<wolfspraul>
not flexible enough, too much hard coded in IC
<wolfspraul>
but maybe for you that's OK?
<orly_owl>
um
<wolfspraul>
I assume you don't want to make your own libre CMOS
<orly_owl>
i dont know
<wolfspraul>
unless you have a few billion USD for make a libre IC foundry
<orly_owl>
ah right
<wolfspraul>
s/for make/to make/
<orly_owl>
not handy, no
<orly_owl>
left my walet at home
<orly_owl>
wallet
<wolfspraul>
you want an external USB webcam?
<orly_owl>
yeah
<wolfspraul>
there are no good camera/video standards in USB land, right?
<orly_owl>
actually there is
<wolfspraul>
oh good. didn't know.
<orly_owl>
UVC has brought some order
<orly_owl>
yep
<wolfspraul>
great, so there is a standard USB procotol now?
<orly_owl>
logitech are making UVC webcams
<wolfspraul>
then maybe you can treat everything behind USB as a black box, and that's free enough?
<wolfspraul>
kinda like a 3G dongle
<orly_owl>
yeah its a standard afaik
<orly_owl>
USB video device class
<wolfspraul>
not that I am satisfied with that but I just try to understand your motivations...
<wolfspraul>
I assume there are Linux drivers for that
<orly_owl>
yep
<wolfspraul>
so why don't you just buy a UVC webcam and done?
<orly_owl>
why would i be here if that was the case? :p
<orly_owl>
my motivations are: webcams have a massive markup on them, dont claim to support GNU+Linux on the box and are of (what i think is) poor video quality
<orly_owl>
i feel ripped off paying $50 to find that it cannot provide smooth video
<wolfspraul>
ok, #1 - #3
<wolfspraul>
#2 seems just a printing problem?
<orly_owl>
well
<wolfspraul>
yes but 50 USD retail for an entire product is a great pricepoint
<wolfspraul>
that thing costs maybe 12 USD or so in China
<wolfspraul>
if it's 50 in BestBuy, for example
<wolfspraul>
maybe 15 USD in China, but that's it
<orly_owl>
its more the case of being able to walk in to a shop and /know/ that this webcam will work on my GNU+Linux system. and i can say to someone who asks what webcam to buy 'look for <logo> on the box'
<wolfspraul>
ok but above you said you want a libre IC
<wolfspraul>
that's a massively bigger undertaking
<wolfspraul>
and it would need to be driven by desires beyond convenience or what's written on the box
<orly_owl>
that's the only way to stop it being a black box, isn't it?
<wolfspraul>
yes sure
<wolfspraul>
so there is Elphel, you know it already
<orly_owl>
yep
<wolfspraul>
wonderful people, I visited them in Salt Lake City a few months ago actually
<wolfspraul>
the cameras are expensive though, 1000-2000 USD until it's all said and done
<orly_owl>
yeah :/
<wolfspraul>
and that is not going to change, it may even go up
<wolfspraul>
but these cameras exist, for many years, Elphel is a true pioneer in copyleft hardware
<wolfspraul>
on the practical side, well, it's a long long way to what you are used to
<wolfspraul>
some winemakers are watching the ripeness of their grapes
<wolfspraul>
oceanographic institutes
<wolfspraul>
military (flew on global hawk), tanks
<zear>
isn't google using elphel for street view?
<orly_owl>
ah interesting
<wolfspraul>
yes
<wolfspraul>
many industrial applications
<wolfspraul>
it goes case by case
<wolfspraul>
but really I think what you are after is something totally different
<wolfspraul>
and maybe impossible
<wolfspraul>
what those tiny camera modules / black boxes in smartphones etc. can do nowadays is just unbelievable
<wolfspraul>
and whether they cost 10 or 20 USD, it's really all zero, because the real thing behind those innovations are many years and multi-billion USD investments by many companies
<zear>
wolfspraul, wasn't it global hawk that used no encryption for sending/receiving the data and talibs learned to receive the images from it? :D
<wolfspraul>
so when you come and really want to free something, suddenly you look at this huge pile of work, for years you would have only totally non-competitive stuff, etc.
<wolfspraul>
to compare, if I build a free phone for you now, I would be very happy if it could compete with the GSM phones in the early 90s
<wolfspraul>
another camera project we are just starting now is called Xue
<wolfspraul>
it's a different concept from Elphel, tries to be much cheaper
<wolfspraul>
I think it's great too, but will take at least several months before we even have first prototype boards, then there will be no software, etc.
<wolfspraul>
and we won't have something to compete with your 50 USD webcam for a long time
<orly_owl>
so basically
<orly_owl>
a few billion $ is needed to start anything
<wolfspraul>
no, I didn't say that
<orly_owl>
oh
<wolfspraul>
you can start anywhere
<wolfspraul>
but there will be big compromises you have to make
<orly_owl>
mhm
<wolfspraul>
so for example you listed 3 things you don't like about off-the-shelf webcams
<wolfspraul>
1. you feel a 50 USD price point 'rips you off'
<wolfspraul>
2. you don't like that the box has no label that says "works with GNU/Linux"
<wolfspraul>
what was #3? ... :-)
<wolfspraul>
3. poor video quality
<wolfspraul>
of course the #3 is because of the 50 USD, which is 12-15 at the source in China
<wolfspraul>
not exactly enough for anything good on the optical side
<wolfspraul>
and also not for a decent CMOS image sensor size
<wolfspraul>
at that pricepoint every penny matters, so it's designed to be able to be sold for 50 USD
<wolfspraul>
and then you even feel 'ripped off', so maybe you want to buy it for 40 USD, or 30?
<orly_owl>
a bit better, yeah
<wolfspraul>
so to make you happy, with your set of motivations, yeah, that's hard
<wolfspraul>
someone would indeed need a few billion just for the fun of seeing whether a free camera for 40 USD instead of a proprietary one for 50 USD (if that's even the case in the end) will sell better
<orly_owl>
then make it libre and $200, its a start
<wolfspraul>
well, nobody will make that kind of bet :-)
<wolfspraul>
he, OK
<wolfspraul>
that's a serious margin!
<wolfspraul>
yes, once you introduce this kind of margin, then even if the volume is only a few thousand units, let alone several ten thousand, you can do a lot of free innovations
<wolfspraul>
hardware is no mystery, no magic as a well known company tries to tell us
<orly_owl>
dell?
<wolfspraul>
but I can tell you, very few people are willing to pay 150 USD more for something only because on a very abstract level, somewhere inside, there is something 'libre'
<orly_owl>
:/
<wolfspraul>
most people prefer mentall to make a big black box around hardware, and outsource it to China or elsewhere
<wolfspraul>
mentally
<orly_owl>
but a few thousand nerds might, thus spurring libre webcams to a lower price point (hopefully)
<wolfspraul>
and they overlook that there's a lot of software that gets ignored that way, but after many years it's a bit late (and hard) to suddenly change that
<wolfspraul>
follow the Xue project
<wolfspraul>
but the main IC on that board alone will be around 50 USD :-)
<wolfspraul>
price to buy just the IC
<orly_owl>
per unit?
<wolfspraul>
sure
<wolfspraul>
trust me, the people that are selling webcams for 50 USD today have made several billion USD investments over many years
<wolfspraul>
methodically
<wolfspraul>
this is not falling from the sky suddenly
<orly_owl>
but its still a proprietry IC in Xue isnt it?
<wolfspraul>
I'm totally with you on changing the libre aspect, but if you neglect something for 20 years, well maybe it also takes 20 years to catch up?
<wolfspraul>
well, we are using (proprietary) FPGAs to start doing GPL licensed ICdesign
<wolfspraul>
plus there are no libre FPGAs, and at least to me evolutionary the libre ASIC must come before the libre FPGA
<wolfspraul>
and we try to not create dependencies to the specific proprietary technology we are using
<orly_owl>
thats a good plan
<wolfspraul>
so we try to keep our GPL licensed IC design portable from Xiling to Altera or other (proprietary) fpga makers
<wolfspraul>
that's as good as it gets right now
<orly_owl>
mhm
<wolfspraul>
at a later stage, we could aim for our own copyleft licensed ASIC production process, and then also to make our own copyleft fpgas
<wolfspraul>
but they still have to be somewhat competitive with the proprietary fpgas at the time, and then you are back to a problem similar like your webcam
<wolfspraul>
maybe we can make a free fpga, but it will look like one 10-20 years ago, and be 10* more expensive still
<wolfspraul>
like I said, that's what happens when you neglect a big area for many years, and others keep investing while you are not
<wolfspraul>
it will change, but it will take time, many years
<orly_owl>
:/
<wolfspraul>
it's not worse than the state of GNU software in the 80's, I think :-)
<orly_owl>
sounds like it
<wolfspraul>
the Windows 95 moment, the mega dominance of some proprietary ICs, stil has to come
<orly_owl>
back then you threw some computers and hackers into a room and out came GNU
<wolfspraul>
there's also a lot going on in hardware
<wolfspraul>
but in the bigger picture (and your perspective is very valid, I am not mocking it), it's really hard to have something competitive
<wolfspraul>
like you feeling ripped off with your 50 USD webcam
<wolfspraul>
:-)
<orly_owl>
yeah i know
<wolfspraul>
but someone has to invest hundreds of millions even more first
<wolfspraul>
and that has already happened, that's why this amazing piece of technology (your webcam) can be sold for _ONLY_ 50 USD
<wolfspraul>
and not, for example, 10,000 USD
<orly_owl>
consumer pc hardware is overpriced junk half the time
<orly_owl>
the jerky video from webcams just /really/ annoys me :/
<wolfspraul>
well
<wolfspraul>
let's do it better, it's hard but I am happy you are showing up here and interested
<wolfspraul>
over time your respect for what the current industry, proprietary as it may be, has achieved, will grow
<wolfspraul>
:-)
<orly_owl>
meh
<orly_owl>
maybe
<wolfspraul>
I am very interested in cameras, we get there.
<wolfspraul>
you can follow what's happening at Elphel, or our very own Xue project
<orly_owl>
Xue sounds more reachable at least
<orly_owl>
how did you decide to make a network camera
<orly_owl>
and not a usb webcam
<wolfspraul>
I'm not even sure that has been decided like that.
<wolfspraul>
first Xue is a derivative from Milkymist, which currently has only USB host support, not USB client
<wolfspraul>
and even the USB host support is far from a complete USB implementation (on Milkymist)
<wolfspraul>
this is all such a massive construction site I don't even know where to start
<wolfspraul>
and then Xue is only in alpha stage right now, not 1 working board exists
<wolfspraul>
conceptual stage
<wolfspraul>
compared to that Elphel, for example, is shipping working cameras for 8 years or so
<wolfspraul>
Xue is really just an idea right now, and we are trying to implement it in reality (=hardware)
<wolfspraul>
once we have the first working boards a lot of SW development needs to happen
<wolfspraul>
then we can think about what products to go after
<wolfspraul>
and then a decision such as 'network camera' vs. 'usb webcam' will be discussed
<wolfspraul>
does network camera mean it's powered over Ethernet?
<wolfspraul>
does 'usb webcam' mean it must be < 100 USD retail, even less?
<wolfspraul>
no idea, really, we are far from even having to discuss such questions
<orly_owl>
oh right
<wolfspraul>
we are trying to do the entire Xue electrical design in KiCad
<wolfspraul>
a GPL licensed EDA tool
<orly_owl>
EDA
<orly_owl>
lets look that up
<wolfspraul>
that's a very daring move, and we will see whether we succeed or have to jump to a proprietary tool again, if we have to make a choice whether we want a functioning board soon, or instead work on freeing tools first
<wolfspraul>
hey, cool. the screenshot actually is KiCad!
<wolfspraul>
the advantage of the picture policied of the WMF
<wolfspraul>
policies
<wolfspraul>
in the industry, KiCad is very very rare
<wolfspraul>
to say the least
<wolfspraul>
:-)
<wolfspraul>
I hope we can succeed with Xue in KiCad, we'll see
<orly_owl>
sounds good!
<wolfspraul>
to me that feels like 100 questions before "is it a network camera or a usb webcam?"
<wolfspraul>
I have "will we be able to pull off the design in KiCad?" as my next big thing...
<wolfspraul>
btw, Elphel uses a proprietary EDA software as well
<orly_owl>
:/
<wolfspraul>
they want to make functioning cameras today :-)
<wolfspraul>
well it's a big construction site this libre hardware stuff
<wolfspraul>
we have to split the workload a little
<orly_owl>
yes its a good start
<wolfspraul>
so as long as we keep our goals in mind, it doesn't matter (to me) to use proprietary tools here and there to get started
<wolfspraul>
in fact I think that's great focus
<orly_owl>
as long as you work towards total freedom
<orly_owl>
so what motivates you to help with this project?
<wolfspraul>
correct
<wolfspraul>
it's a lot of fun, learn about new technologies all the time, make new friends. and I think it can become a real (money making) business too.
<orly_owl>
ok
<orly_owl>
so
<orly_owl>
its a long way to the top if you wanna rock and roll? :)
<wolfspraul>
it's fun, you should join, but given your starting motivations maybe it's too much to ask from you? don't know
<orly_owl>
oh i feel too involved already
<wolfspraul>
if the printing on the box is not right for you, I would just take a big black pen and write "designed for GNU/Linux" on it, to make you happy
<wolfspraul>
problem solved :-)
<orly_owl>
im organising stuff will my local user group
<orly_owl>
stuff for software freedom day
<orly_owl>
getting CDs made
<orly_owl>
s/will/with/
<wolfspraul>
cool, maybe someone should buy our lovely little Ben NanoNote
<orly_owl>
i just feel very busy
<orly_owl>
yeah i should organise a group buy after SFD maybe
<wolfspraul>
jump into free software on embedded systems, OpenWrt, Jlime (OpenEmbedded), Debian with 32 MB RAM :-)
<wolfspraul>
that would be great
<orly_owl>
yep, openwrt on WRT54G here. their documentation is just horrible though. so confusing
<orly_owl>
i put familiar linux on a compaq ipaq. didnt do much with it though
<wpwrak>
murphy's law on USB: one you've built a sufficient large collection of USB devices, all of a sudden you run out of "good" cables and only find those that bring to light all the weird bugs
<bartbes>
oh that happens to all collectios
<bartbes>
*+n
<wpwrak>
aha ! bad contact in the connector. plus a bug in my USB library. and just how nicely they work together.
<qi-bot>
[commit] Werner Almesberger: lib/usb.c (open_usb): Stupid beginner's mistake: wrong operator precedence due http://qi-hw.com/p/f32xbase/79396b1
<qi-bot>
[commit] Werner Almesberger: Made communication with CNTR board more robust. Added documentation. http://qi-hw.com/p/ben-wpan/c7303e4
<johnny2>
hi can someone here help me with compatibility issue?
<kristianpaul>
johnny2: hey
<kristianpaul>
just people like questions here ;)
<kristianpaul>
just ask*
<johnny2>
hi
<johnny2>
i need to ask another question not related to compatibility
<johnny2>
i want to know which inexpensive graphic card can run crysis game on very high setting?
<mth>
johnny2: this channel is not about PC hardware: Qi Hardware is the name of a company that made a handheld Linux-based computer
<mth>
as for your question: inexpensive and high settings Crysis are not really compatible - but most reviews on sites like Anandtech or Tom's Hardware contain Crysis benchmarks so you can see how well the different cards score
<wpwrak>
mth: naw, the company is called something like "sharism at work ltd." :) and i think they're making more bens just these days.
<mth>
well, it's not easy to be both accurate and concise :)
<wpwrak>
mth: wolfgang did a good job at obfuscating his traces ;-)
<wolfspraul>
always
<wolfspraul>
we started as a Qi Hardware Inc. registered in California
<kristianpaul>
oh
<wolfspraul>
then later we thought it would be better to make Qi Hardware the community, and have a separate manufacturing company manufacturing stuff.
<wolfspraul>
of course the idea is that the manufacturing company is just one of many, since all knowledge is copylefted
<kristianpaul>
founded slides about shanzhai vs qi
<wolfspraul>
but there is only one manufacturer now, and that is indeed a company registered in Hong Kong, called Sharism at Work Ltd.
<wolfspraul>
somewhere in the middle we also had this idea to rename Qi Hardware to "Qi Inside", but only only we got pretty clear language letters from Intel lawyers
<wolfspraul>
and probably there were some other ideas as well
<wpwrak>
wolfspraul: (intel) oh, did you ask them or did they spot you the moment you tried ?
<wolfspraul>
they spotted the trademark registration
<kristianpaul>
ohhh
<wpwrak>
not bad :)
<wolfspraul>
and instead of using the official trademark 'opposition' path, threatened in mafia (aha, real world) style with crazy fines etc.
<kristianpaul>
well is 1nt3l... soo
<wolfspraul>
if I ever have some time, I go back to this
<wolfspraul>
I will register a "Wolfgang Inside" trademark and sell t-shirts
<kristianpaul>
lol
<wolfspraul>
tell them in advance I will do this, and that I will not give in to their threats, instead I see them at the official trademark dispute courthouse
<wolfspraul>
something they want to avoid at all costs, backroom deals are so much nicer...
<wolfspraul>
but I have no time for this right now...
<wolfspraul>
kristianpaul: oh trust me I will do it. I like this kind of stuff. :-)
<wpwrak>
so many windmills to fight in IPland ;-)
<wolfspraul>
yes, that's why not now
<wolfspraul>
but I really didn't like how they threaten you behind the curtain
<wolfspraul>
what is the legal system for? they make a mockery of it
<wolfspraul>
one would think the trademark office exists to settle disputes, but they don't like the uncertainty of due process, of course...
<mth>
or trademark "Evil Inside" and let Intel break their heads over whether they should claim they own that or not ;)
<wolfspraul>
that was the main part that pissed me off
<wolfspraul>
"fear tax" all over
<wolfspraul>
I have no problem following the law, and if the law (trademark law!) says that {word} Inside is Intel's, fine, I will follow
<wolfspraul>
but I never get the chance to find out from an independent party (aka judge)
<wolfspraul>
that's not right
<wolfspraul>
anyway, that's just the Qi Inside side story
<wolfspraul>
no worries guys, I will not get distracted on that one...
<wolfspraul>
mth: they will come after you.
<wolfspraul>
these lawyers have nice cushioned deals with Intel and I'm sure they charge the hell out of Intel for every letter and sneaky action they can come up with against you.
<wolfspraul>
the law is theirs, to the last point.
<wolfspraul>
if you are paid 500 USD an hour it's also worth to be a little creative... :-)
<kristianpaul>
heh..
<mth>
yeah, it's like spending money on weapons, most are bought with defensive intentions but they tend to get used offensively eventually
<wolfspraul>
is the Qi Hardware vs. Sharism thing clear now?
<wolfspraul>
Sharism is just a small manufacturer, needs to be profitable. It's a primitive little business.
<wolfspraul>
Qi Hardware is a grandiose free the world movement.
<wolfspraul>
:-)
<mth>
it's clear now and it wasn't clear to me before, so all this was actually useful :)
<kristianpaul>
yes and the copyleft hardware words let me add :)
<wpwrak>
hmm .. with a 15 pF load and sloppy soldering on board #1 my crystal is 13 ppm slow. with a 10 pF load and neat soldering on board #2, it's about 20 ppm fast. before, without load caps, they were both within 5 ppm of each other. what conclusions do i draw from this ? *scratch,scratch*
<mth>
probably that you have an insufficient number of measurements?
<wpwrak>
that's always a good answer ;-)
<wpwrak>
"after ten years of research, we conclude that more research is needed" :)
<mth>
although once of my teachers said that 2 measurements is always good, since you can draw a straight line through them
<wpwrak>
i did actually two measurements on each board. they differ by 1 ppm each. (i should add that my theoretical error bounds are in the order of 100 ppm, but they seems to be grossly inflated. the main factor there is relative NTP accuracy, which seems to be pretty good.)
<mth>
and you have factors like temperature at similar values?
<mth>
I'm mostly a software person, so correct me if I say stupid things about hardware
<wpwrak>
mth: temperature may vary by a few K, yes. there's little i can do about this :-(
<wpwrak>
the good news is that fixing the clock that's a bit too fast should be well within the reach of the trim capacitors in the chip. let's see how this goes ...
<mth>
do you need clocks that are this accurate?
<mth>
you could always reduce the error margin by simply doing a longer measurement, right? since the NTP accuracy only comes into play at the start and end of the measurement and is distributed over the entire interval
<wpwrak>
mth: i need 40 ppm over the (ben's) operating temperature range. the crystal has a base accuracy of 15 ppm plus 15 ppm over its temperature range (-10 - 75 C)
<wpwrak>
mth: (long measurement) yes, that's what i'm doing. the ones i'm running now take about 1000 seconds and seem to be good to maybe 1-2 ppm.
<unclouded>
hey xiangfu, have you got time to test a nightsky .ipk with some debugging in it?
<xiangfu>
unclouded: but I don't know how to do that :(
<unclouded>
I'll supply a new .ipk that prints some debugging when it starts