<wpwrak>
hmm, my pc is getting fat and slow. almost 200 seconds for a cache-cold kernel build :-(
<wolfspraul>
I read that Microsoft is now asking for, and getting, 5-12 USD per Android device
<wpwrak>
we should do that too :)
<wolfspraul>
I find it interesting because it creates a communication challenge for copyleft hardware
<wpwrak>
maybe a little more, because microsoft are already rich, so we need it more :)
<wolfspraul>
first it devalues copyright, the whole concept
<wolfspraul>
then it goes about the 'microsoft tax' that some people still feel passionate about in totally new ways
<wpwrak>
how does it devalue copyright ?
<wolfspraul>
then it creates this impression that "unless you pay for patents _somewhere_, it's 'illegal'"
<wpwrak>
well, the patent system is pretty much like that ...
<wolfspraul>
because it doesn't matter who owns the copyright (free software / open source), what matters is who owns strong patents :-)
<wolfspraul>
so it definitely devalues the entire concept of copyright, _especially_ for free software
<wpwrak>
someone will certainly have some patent that you can be interpreted as interfering with, whatever you do
<wolfspraul>
sure, of course
<wolfspraul>
we are protected by our size, etc.
<wolfspraul>
but that is hard to communicate, and hard to get passion behind
<wolfspraul>
if this Android trend really continues, Microsoft may make more money from it than Google!
<wpwrak>
i'm not sure it devalues the copyright per se. both systems are restrictive - by default, you're not allowed to do anything. you can only be granted rights.
<wolfspraul>
sure it devalues copyright, to the outside
<wolfspraul>
outsider
<wpwrak>
of course, it devalues the assertion that Free whatever is truly free
<wolfspraul>
yes
<wolfspraul>
and that 'free software' has anything to do with 'microsoft tax'
<wolfspraul>
Android proves that it does not. wrong subject guys, the rules have changed. :-)
<wolfspraul>
while you are still perfecting your signed-by chains, the money is already following completely different rules...
<wolfspraul>
while it was only some trolls, we could dismiss it as something like a natural disaster
<wolfspraul>
but now it's Microsoft taxing Android, and it seems to be growing
<wolfspraul>
I already expect the press articles that the 'free software' concepts are a thing of the past, etc.
<wpwrak>
perhaps the best approach is to just set up a proxy company for each product. bill it all the cost, including R&D. if it's successful and the then inevitable patent strike happens, let it go bankrupt
<wolfspraul>
oh sure, in the real world I'm not worried one bit
<wolfspraul>
it creates a perception/motivation challenge
<wolfspraul>
not on my end :-)
<wolfspraul>
but with others, I think (that's my question)
<wpwrak>
is M1.org set up accordingly ?
<wolfspraul>
what do you mean?
<wpwrak>
or, rather, who will be the business end behind M1 ?
<wolfspraul>
my point is that it may demotivate people to engage in free software
<wolfspraul>
nah, there is no patent risk in Milkymist/NanoNote/OpenWrt/any of this
<wpwrak>
it may
<wolfspraul>
I'm not worried about that at all.
<wpwrak>
always the optimist ;-)
<wolfspraul>
no it's not
<wolfspraul>
I take that risk, done.
<wpwrak>
don't you think there are *some* patents that cover *something* in M1 ?
<wpwrak>
c'mon :)
<wolfspraul>
that's the wrong question
<wolfspraul>
because that is already true for every line of C code for at least 10 years
<wpwrak>
the real question is if it has a chance to appear on the patenteer's radar
<wpwrak>
so if you succeed, you die
<wolfspraul>
of course not, way too small
<wpwrak>
if you fail, you can do so in peace
<wolfspraul>
and too risky (for them)
<wolfspraul>
but if you grow (and make money), many people will be attracted by the money that they know you have
<wolfspraul>
the patent guys will be just one of 10-20 groups to watch out for
<wolfspraul>
you misunderstood my point
<wpwrak>
conclusion: you should make sure no money accumulates
<wolfspraul>
it's not me who is worried or demotivated
<wolfspraul>
I'm wondering what this development means to the bigger copyleft hardware/software picture, and how to communicate it to newcomers
<wolfspraul>
no no
<wolfspraul>
you can accumulate money, but then you need to deal with who comes and wants a share :-)
<wolfspraul>
that's natural
<wolfspraul>
my point is - don't you think that the development that Microsoft, of all people, is taxing Android, makes a difference in the free software world?
<wpwrak>
this mess is a completely normal consequence of the way the patent system works. absolutely nothing new there. maybe it will help more people to understand why patents are evil.
<wolfspraul>
the 'avoid microsoft tax' argument is definitely mood now
<wpwrak>
maybe some will have a rude awakening
<wolfspraul>
at least that needs to be removed from all sites, I'm sure it's still somewhere on gnu.org or fsf campaign pages...
<wolfspraul>
you can't continue to make that claim while Microsoft is counting the money it makes from free software
<wpwrak>
android could still avoid it. but then they'd have to give up control. sell bare linux machines.
<wolfspraul>
that won't happen. all the big players will pay.
<wolfspraul>
I think after HTC and Acer did, it's a done deal.
<wpwrak>
yeah, of course. but it's for their control that they're really paying.
<wolfspraul>
it creates a perception/communication problem
<wolfspraul>
ok, let me ask this way
<wolfspraul>
imagine you run a site that explains the advantages of free software to newbies
<wolfspraul>
ok?
<wolfspraul>
and somewhere you have the 'avoid microsoft tax' argument
<wolfspraul>
would you remove it? or keep it? add some notes to it?
<wpwrak>
perhaps we should then communicate the subtle difference between a blank and clean linux system and the bundle with all sorts of patent infringements. for all we know, the patents that affect android may even be on non-free code.
<wpwrak>
i think the argument is still valid. maybe explain why this case doesn't invalidate the argument but merely adds some constraints.
<wpwrak>
like you can't sell a freerunner with lovely free mplayer and open source mp3 codecs. at least not for long without trouble ;-)
<wolfspraul>
yes but that was contained, now it gets bigger
<wpwrak>
there is no containment with patents ;-)
<wolfspraul>
the amount ms charges is roughly the same as they would get for a winmobile license
<wolfspraul>
so you admit that:
<wolfspraul>
free software is only about copyrights, it's a specialist thing for lawyers
<wolfspraul>
copyrights are not that important in the business world, when it's about making money
<wolfspraul>
that's how you want to tell newcomers about free software?
<wpwrak>
copyrights are very important. you're mixing things incorrectly.
<wpwrak>
you create revenue by imposing restrictions
<wpwrak>
like a corrupt customs official who doesn't let your goods through
<wolfspraul>
yep
<wolfspraul>
:-)
<wolfspraul>
the oldest business model in the world, probably
<wpwrak>
then you get paid for relaxing the restrictions on a case by case basis
<wolfspraul>
setup a toll booth
<wpwrak>
and have some thugs nearby
<wpwrak>
today the thugs wear suits, but their ethics haven't changed much
<wolfspraul>
but if free software focuses on copyrights, its reach and power is limited by the growing power of patents?
<wolfspraul>
we will have a hard time countering such an argument
<wolfspraul>
it goes so far that someone will say "unless you pay one of the big guys, your stuff is illegal"
<wolfspraul>
which we've heard already 1000 times when we still did phones
<wpwrak>
of course it is limited by patents. where have you been the last one or two decades ? ;-)
<wolfspraul>
nah
<wolfspraul>
before there were trolls, or big corps in cross-licensing deals
<wolfspraul>
now we have Microsoft taxing Android
<wpwrak>
in europe, they've waged a little war against software patents
<wolfspraul>
the 'open source posterchild' by some, still
<wolfspraul>
that is a strong story, and it will spread
<wolfspraul>
and what we discuss here in our niche 50-people copyleft hardware irc channel is completely irrelevant compared to that
<wpwrak>
duh. truth be told, that story doesn't bother me much. what bothers me is the patent system that allows all such niceties.
<wolfspraul>
HTC and Acer pay now, I think everybody will over the next 12 months
<wolfspraul>
but it may create a motivation problem for newcomers
<wolfspraul>
with HTC and Acer paying, they got all of Taiwan. Every Taiwanese OEM will now enter negotiations and pay.
<wpwrak>
they can't limit the sharing of information this way. they can only limit production, shipping, and sales
<wolfspraul>
you know that
<wolfspraul>
next they need to tackle South Korea and Japan, they are probably on it already
<wpwrak>
also, you can try to find regional work-arounds
<wpwrak>
look at pharmaceuticals. there, they had the same things happening
<wolfspraul>
good thing that we don't use Android, phew
<wpwrak>
;-)
<wolfspraul>
soon if you have an Android device without paying for patents, the automatic assumption will be 'illegal'
<wpwrak>
you seem really rattled :)
<wolfspraul>
I think about the impact on newcomers.
<wolfspraul>
and I do think it makes a difference, but we see
<wolfspraul>
the 'microsoft tax' was a big thing to attract people to free software
<wolfspraul>
you will certainly agree
<wpwrak>
well, you can try to find out which patents exactly they pay for. see which apply to the free part of android, which to the non-free part.
<wolfspraul>
nah
<wolfspraul>
nobody will talk about that
<wpwrak>
then see which apply to sectors we already know are poisoned
<wolfspraul>
it will just be "Android patents - Microsoft"
<wolfspraul>
the smaller OEMs won't even look at the details
<wolfspraul>
HTC & Acer pay - we pay as well
<wolfspraul>
done
<wpwrak>
i thought you were worried about newcomers ?
<wolfspraul>
I mean individuals that discover free software.
<wolfspraul>
people, not companies
<wpwrak>
so explain to them why this doesn't mean the end of free software. see above.
<wpwrak>
android just happens to be set up such that the MS tax can be collected. that's part of their strategy. the problem is there (and in the patent system) not in free software or hardware per se.
<wpwrak>
with the provisio that free hardware is always more vulnerable than software, which is also something we know well
<wpwrak>
now what you can do is write some witty and insightful posting to the qi-hw list that ridicules stupid android for feeding the microtroll :)
<wpwrak>
a practical consequence of patent risks is that you should build firewalls against attacks, in case a project actually does take off. i don't think you've considered that yet.
<wpwrak>
when the attack comes, you can't do much to protect your future revenue. but you can at least protect your past revenue.
<wpwrak>
the best strategy is probably to just throw the towel and get a lot of press attention
<wpwrak>
but make sure your past revenue is secure. the patenteers don't usually come before you've made some substantial amount of money
<wpwrak>
thus it would be good to know what assets are protected. e.g., salaries most likely are. compensations for external providers not involved in the decision process probably too. not sure if this changes if they are part of the decision process.
<wpwrak>
(salaries) of employees. not sure about management.
<wolfspraul>
he
<wolfspraul>
understood
<wolfspraul>
you worry too much
<wolfspraul>
I will not write any mail about this, waste of time. And also I try to stay positive, not sure what the positive angle here is.
<wolfspraul>
for you and me nothing changed, whether Acer pays for the MS patents or not
<wpwrak>
you're funny. first you complain about the sky falling, then you say i worry too much ;-)
<wolfspraul>
or HTC or others
<wolfspraul>
I wonder whether this development makes a difference to _NEWCOMERS_ who look at free software in 2011 for the first time
<wolfspraul>
someone born in 1998
<wolfspraul>
I expect many stories along those lines in coming months. And at least the old 'microsoft tax' arguments, which rely on copyright, won't work anymore. Since this is a tax beyond copyright.
<wolfspraul>
so it may impact someones willingness to learn more about copyright, overall
<wolfspraul>
if that's not what matters anyway...
<wolfspraul>
I am strictly thinking this through from the perspective of a newcomer.
<wpwrak>
patents have been the big enemy for a very very long time. copyright is our tool, not our problem.
<wpwrak>
if you take copyright away, all GPL becomes BSD, all CC-BY-SA becomes CC-BY. not the end of the world, but a difference.
<wpwrak>
if we don't actually steal code from somewhere, copyright is not against us. attempts to plant false accusations against us have been quite unsuccessful, even if spectacular (e.g., SCO)
<wpwrak>
patents are the enemy. have always been. patents are evil. if you think what is the foundation of what we consider humanity, they are crimes against it. patenteers should be judged and hanged ;-)
<wolfspraul>
and why should someone join a free software project if it's uneffective against this 'evil' thing anyway?
<wolfspraul>
you basically say the two are completely unrelated
<wpwrak>
because free software still escapes patents while it remains pure
<wolfspraul>
alright
<wolfspraul>
I got it
<wolfspraul>
the Microsoft-Android thing will still hang very big over this though, I'm sure
<wpwrak>
and you can create an incentive for parts of the world that have less to gain from patents to seek a different path. because doing so, they allow themselves to benefit from free software also when "impure"
<wpwrak>
(android) yeah. perhaps the ambiguous nature of android could use some rehashing
<wpwrak>
*grin* found something :)
<wpwrak>
likes the look of four screens around himself
<wpwrak>
should add a fifth for a ben with ubb-vga :)
<kristianpaul>
so now we can said "patent free, or almost patent free" ?
<kristianpaul>
when talking about M1 :-)
<kristianpaul>
hides from wpwrak
<kristianpaul>
he, just kidding
<wpwrak>
i wouldn't use "patent free". i don't like that this has been tagged on ben-wpan either. you can only declare something as "patent-free" in a retrospective perspective
<kristianpaul>
sure sure i know that
<kristianpaul>
dicuss that with rejon first :)
<wpwrak>
kristianpaul: naw, he has to do some money-begging missions to do. i'll be nice to him until that's done ;-)
<kristianpaul>
:-)
<wpwrak>
kristianpaul: how's your data gathering ? i lost a bit track of it, sorry. does it look better now ?
<kristianpaul>
wpwrak: i dint tried offline processing (even now that finally got SoftGNSS code for complex numbers), i finishing the port for one channel correlator from namuru
<kristianpaul>
i dint tried again*
<wpwrak>
so maybe next weekend will be the moment of the breakthrough ? :)
<kristianpaul>
at least i know what to do,and where look for
<kristianpaul>
you maybe remenber i pastebin some mail from namutu developer with some comments about the porting
<kristianpaul>
very usefulls btw, plus i know what can fail and with scope i can debug it
<kristianpaul>
wpwrak: breakthrough for C/A matching i think so, if all went okay
<kristianpaul>
s/matching/tracking
<wpwrak>
heh, cool :)
<wpwrak>
i admire your endurance. i don't know many people who wouldn't have given up months ago.
<kristianpaul>
well, 8 months ago i dint knew  nothing about GNSS/GPS, now i cant said that :-)
<rejon>
ha
<kyak>
wpwrak: do you know if lowpan-tools would work fine with libnl-2?
<lekernel>
I scavenged three 24-GHz ones yesterday from automatic door radars, and they have an empty diode slot marked "IF2"
<lekernel>
I'm wondering if I could throw a varactor in there to make tunable gunnplexers
<lekernel>
the IF2 slot is the same as the (populated) IF1 slot, and it'd even be better if I could use the reverse-biased detector diode from another gunnplexer as varactor (since those critters are usually extremely expensive and difficult to order...)
<wpwrak>
grmbl. shouldn't lit down to "briefly think about a problem"
<qi-bot>
[commit] Werner Almesberger: drivers/spi/Kconfig: clarified description of CONFIG_SPI_ATUSB (ben-wpan-stefan) http://qi-hw.com/p/qi-kernel/c211a33
<qi-bot>
[commit] Werner Almesberger: atusb: converted atusb_get_static_info to use usb_control_msg (ben-wpan-stefan) http://qi-hw.com/p/qi-kernel/429510e
<qi-bot>
[commit] Werner Almesberger: atusb: ignore interface unless class is USB_CLASS_VENDOR_SPEC (0xff) (ben-wpan-stefan) http://qi-hw.com/p/qi-kernel/23d5eb7
<qi-bot>
[commit] Werner Almesberger: atusb: merged USB control message submission into common function (ben-wpan-stefan) http://qi-hw.com/p/qi-kernel/e803b81
<qi-bot>
[commit] Werner Almesberger: atusb: clean up interface of atusb_read1 and atusb_write (ben-wpan-stefan) http://qi-hw.com/p/qi-kernel/191d55a
<qi-bot>
[commit] Werner Almesberger: atusb: corrected error and message length handing for transfers (ben-wpan-stefan) http://qi-hw.com/p/qi-kernel/b12e0e7
<wpwrak>
kyak: with -g (using SDL_gfx) for atrf-rssi and atrf-path ?
<kyak>
lemme check it
<wpwrak>
zrafa: now it has happened: jlime has fallen behind openwrt on the features. you don't have these packages yet (at least ben-wpan tools didn't exist in this form back then :)
<kyak>
wpwrak: how do i select the SDL_gfx driver?
<wpwrak>
kyak: by compiling for jlime ;-) back then, SDL_gfx wasn't in the openwrt SDK and i haven't rebuild that one for a while. is it now there ? i also need libgsl (GNU Scientific Lib) for atrf-path
<wpwrak>
err, no, atrf-xtak
<wpwrak>
xtaL
<kyak>
wpwrak: SDL_gfx is in openwrt for.. a while now
<wpwrak>
what's the easiest/recommended way to get a host cross-development environment now ?
<kyak>
the easiest would be to follow the steps from "Compiling your own image"
<kyak>
perhaps it is possible with SDK, but i didn't try it for real
<wpwrak>
ah, no gsl requires for now. that's only used in the atusb variant. may come later, though. it's such a useful library :)
<wpwrak>
(i've skipped the manual patches for now. are they required ?)
<kyak>
(reading the instructions) yeah, you need to cd ../..
<kyak>
patches are not needed for now
<wpwrak>
kewl. make -j4 is running
<kyak>
i'm afraid it is going to build a ton of unnecessary stuff
<wpwrak>
yeah, i know :)
<wpwrak>
there's no pre-compiled .deb for the SDK, is there ?
<kyak>
you should've created your .config from scratch
<kyak>
or taken the minimal config and only selected SDL_gfx
<wpwrak>
naw, i want the "standard" SDK. if the standard is loaded with junk, so be it :)
<kyak>
there is a tar.gz for SDK, but as i said i didn't really try it
<wpwrak>
okay, let's keep this for later :)
<wpwrak>
wonders if that V=99 is such a good idea
<kyak>
not a good probably :)
<stefan_schmidt>
wpwrak: oops, I freed the wrong urb. The clock disabling is something I was not even aware of. :)
<wpwrak>
subtle, eh ? ;-)
<stefan_schmidt>
wpwrak: sync for the static info is what we had before. I just changed it to get started with async.
<kyak>
you could restart it without V=99, it shouldn't hurt anything..
<stefan_schmidt>
wpwrak: fine with having it back sync
<wpwrak>
stefan_schmidt: now i'm wrestingling with "long" read and write. read has the little quirk that the first byte does matter. i had overlooked that.
<wpwrak>
kyak: naw, then my "time" would reset, too :)
<stefan_schmidt>
wpwrak: hmm, the ignoring of all but vendor spec is an option.
<kyak>
wpwrak: are you into challenge with buildhost? :)
<stefan_schmidt>
wpwrak: and merging the functions was also on my list. Just wanted to have somethign working before doing to much cleanup and make my life even more harder :)
<stefan_schmidt>
heh, and some cleanup in all the tx[0], tx[1], rx+1 naming mess :)
<wpwrak>
kyak: yeah, i want to see how long this should really take :) how good is my competition ?
<kyak>
wpwrak: you are probably about to embarass qi-bot badly :)
<wpwrak>
kyak: let's see. my pc is something like 4 years old. old q6600.
<stefan_schmidt>
heh, the good old openmoko machines. :D
<wpwrak>
stefan_schmidt: yeah, similar specs. bought it in buenos aires, though
<wpwrak>
stefan_schmidt: thought that something harald has spent half a year on spec'ing can't be too bad ;-)
<stefan_schmidt>
wpwrak: yeah, got that far. Would not have make sense to put that one into your luggage ;)
<wpwrak>
kinda invoncenient ;-)
<stefan_schmidt>
has to wait for a complete new kernel build after and gcc upgrade of his distro...
<wpwrak>
kyak: btw, regarding http://en.qi-hardware.com/wiki/Building_Software_Image, i did the  git checkout -b local-build a0cb48741e20e3f2dd7eb546e57e0ae085eb7a31  variant instead of  git tag; git checkout release_2011-02-23
<wpwrak>
kyak: the  cd feeds/qipackages/ && git checkout release_2011-02-23  later on didn't seem to mind. hope that's good.
<wpwrak>
kyak: (i haven't used git tags yet)
<wpwrak>
meanwhile, the build is somewhere in a libc ...
<wpwrak>
doing stuff in parallel, which is a good sign
<wpwrak>
building again. i now changed all the "latest" to 2011-05-28
<wpwrak>
i think it would be good to have a snapshot that conserves all the external dependencies (downloads) at the time of building. otherwise, the whole thing is a moving target
<kyak>
latest is a symlink to 2011-05-28 atm
<kyak>
such kind of snapshot is saved
<kyak>
the VERSIONS file
<wpwrak>
ah, good. so you'd just need the source tarballs
<kyak>
you always know what was the base system and feeds state at the moment of release
<wpwrak>
(latest == 2011-05-28) could be. the dates look the same
<kyak>
must be!
<wpwrak>
(mirror) yes, it seems to fall back to the mirror. but first, it tries the original places. that's risky.
<kyak>
in any case, openwrt download system tries several download urls, and checks the md5sum of a package
<kyak>
so it's not risky
<wpwrak>
you can still fail because some random site at the far end of the intertubes is congested
<wpwrak>
well, maybe not fail
<wpwrak>
but spend a lot of time waiting
<kyak>
so? &
<kyak>
you have to download the tarball anyway
<kyak>
i'm angry when SF feeds me some mirror every time
<kyak>
*some crappy mirror
<wpwrak>
yeah ;-)
<kyak>
and so i wait
<wpwrak>
SF sucks
<kyak>
that i won't argue with :)
<wpwrak>
well, their download system. i have a few projects there but never touched that evil download system. just use them as free web space :)
<wpwrak>
qi-hw projects also has the problem of giving non-downloadable download links if using the download system
<wpwrak>
almost through toolchain :)
<wpwrak>
three bored cores are eagerly awaiting what comes next :)
<viric>
Anyone here using some working voip software?
<viric>
between computers would be fine, like skype.
<viric>
and I meant free software, ofc.
<wpwrak>
grmbl. need to do a full distribution upgrade of my test pc before even iz will run :-(((
<qi-bot>
[commit] Werner Almesberger: atusb: tentative addition of "long" read/write (untested) (ben-wpan-stefan) http://qi-hw.com/p/qi-kernel/09294c5
<kyak>
wpwrak: i've finished updating the "Building OpenWrt base on release files" section
<kyak>
wpwrak: could you check if your eye catches on something, cause you just went through it?
<stefan_schmidt>
wpwrak: ah, I just hit the long read write case when doing iz attach
<stefan_schmidt>
heh, an XXL buffer
<wpwrak>
stefan_schmidt: maybe you can try to see if it works with this patch. you should be able to send one frame with izchat before it hangs due to the missing interrupt
<stefan_schmidt>
wpwrak: ging to do this right now
<wpwrak>
stefan_schmidt: and you can see if anything actually got sent by watching with atrf-txrx from another pc
<wpwrak>
kewl :)
<stefan_schmidt>
wpwrak: no other pc around right now. And my sniffer hardware is in the lab and I'm at home
<wpwrak>
kyak: thanks ! is the  git branch -a  really useful ?
<kyak>
just to see what branches are existing
<kyak>
so that you could know which one to checkout
<kyak>
and which one is the latest
<wpwrak>
hmm. but then you should mention that there's a decision to take. most people will read this as a cookbook and just copy and paste
<xiangfu>
kyak: (updating the "Building OpenWrt...) thanks
<wpwrak>
so the  cd feeds/qipackages/ && git checkout release_2011-05-28  wasn't necessary after all ? :)  (i remember seeing it say that i'm already here)
<stefan_schmidt>
wpwrak: Async USB succeeded with length 19
<stefan_schmidt>
wpwrak: thats after a write2 classification
<wpwrak>
stefan_schmidt: sound promising :)
<kyak>
wpwrak: yup, not necessary
<stefan_schmidt>
wpwrak: will try to setup something to sniff on the air :)
<kyak>
if you've fetched the feeds.conf
<kyak>
xiangfu: np!
<kyak>
wpwrak: and i added a comment to git branch -a
<kyak>
wpwrak: "sed -i 's/=m$/=n/g' .config" is very important, if you don't want to build a whole bunch of staff (that is not even in release image, and just build as separate packages - "modules")
<wpwrak>
kyak: kewl. the rest looks good. i'm a bit uncertain about the "make" variants. seems that  V=99  would be used for debugging the build process while  IGNORE_ERRORS=m  would be used for just the opposite scenario. so shouldn't it be either  make V=99 (distro-developers)  or  make IGNORE_ERRORS=m (users)  instead of  make V=99 IGNORE_ERRORS=m (first recommendation)  vs.  make (offered as an afterthought)
<wpwrak>
(s/=m/=n/) ah, it's building now ...
<kyak>
perhaps IGNORE_ERRORS=m could be removed at all after there are no =m lines in config
<wpwrak>
(s/=m/=n) will i save more than the hour of build time i've already spent, most of it in the evil single-threaded bootstrap part ? :)
<wpwrak>
wishes it would parallelize packages not inside packages. most spend 90% of their time in configure or similar nonsense, so there's hardly any effect
<kyak>
you will save a lot of time without =m packages
<kyak>
and it also parallelizes not only inside pacakges
<wpwrak>
the 6 hours are with or without =m ?
<kyak>
though it may lead to strange results
<kyak>
for me it's around 3 hours without =m
<kyak>
i.e. building onyl those pacakges included in official image
<wpwrak>
(parallize) hmm. i rarely see more than one active process at a time
<stefan_schmidt>
hmm
<kyak>
do you run make with j= ?
<wpwrak>
ERROR: package/feeds/packages/dante failed to build.
<wpwrak>
luckily, this seems to have been one of those modules :)
<wpwrak>
with -j 4
<wpwrak>
err, -j4
<kyak>
hm... if it is module, it should've been removed
<wpwrak>
no, i this run didn't have the s/=m/=n/
<kyak>
damn
<wpwrak>
you did that later :)
<kyak>
how stupid am i?
<wpwrak>
mmh ?
<kyak>
=n won't work ):
<wpwrak>
hehe ;-)
<kyak>
it has to be # bla bla is not set
<wpwrak>
maybe /=m/d ? :)
<kyak>
yep. something like you said + make oldconfig..leemmm double check
<wpwrak>
maybe  make allnoconfig  else it'll ask about the things it forgot about
<kyak>
there is a yes "" | make oldconfig trick we already use :)
<wpwrak>
hehe :)
<wpwrak>
afk for a bit
<kyak>
ok, much better now
<wpwrak>
stefan_: you keep atusb->ctrl_urb for cleanup ?
<wpwrak>
hmm .. nobody seems to use spi->bus_lock_flag. so perhaps we shouldn't either. tricky ...
<wpwrak>
ERROR: package/feeds/packages/elfutils failed to build.
<wpwrak>
fairly peaceful so far
<stefan_>
wpwrak: I'm using one atusb in my kvm test system and the other one directly from my host for atrf-txrx but so far nothing showed up during iz attach
<stefan_>
wpwrak: (ctrl_urb) maybe its now longer needed after your reworks. need to check
<stefan_>
wpwrak: (bus_lock_flag) its only useful when we have more then one device on the same SPI bus?
<wpwrak>
(ctrl_urb) i could imagine a role during driver removal, in case there's still something pending. but i'm not entirely sure we need to worry about this case.
<wpwrak>
(bus_lock_flag) let's hope for the best :)
<wpwrak>
ERROR: package/feeds/packages/libmicrohttpd failed to build.
<wpwrak>
(still going. all smooth so far.)
<stefan_>
wpwrak: looking through the code it does not make much sense to keep it in the struct. IIRC it was there to get passed around
<stefan_>
wpwrak: I don't see much use of driver removal of it.
<stefan_>
wpwrak: I would say we move it out of the struct and have it only local in submit_control_msg
<kyak>
wpwrak: btw, if sdl-gfx has already built, you can try it
<wpwrak>
kyak: (list) ah, good. still a few more to come then :)
<wpwrak>
kyak: (already built) naw, i'll wait for the whole thing to settle. don't want to chase ghosts :)
<kyak>
hehe, ok :)
<stefan_>
wpwrak: should I kill ctrl_urb from the struct or do you already have it done?
<wpwrak>
no, i haven't touched it yet
<stefan_>
wpwrak: ok, I'll go ahead then
<wpwrak>
we also don't really need to keep the static info in struct atusb_local. i think you killed the reference to it by kicking out the sysfs files
<stefan_>
wpwrak: hmm, right
<wpwrak>
i'll remove it once you push
<wpwrak>
will make things quite a bit simpler :)
<stefan_>
wpwrak: we don't need the infos if we have different firmware versions with different features?
<kyak>
$(STAGING_DIR) expands to something like /home/bas/build/openwrt-xburst/staging_dir/target-mipsel_uClibc-0.9.32
<wpwrak>
(still waiting for my test pc to finish upgrading. then i'll be able to ru(i)n kernels again)
<kyak>
by openwrt build system
<zrafa>
wpwrak: there is no a lot of motivation to maintain a distribution for a machine with a fixed number of users. At least we have not gotten many requests from jlime users..
<zrafa>
wpwrak: sujan who is using the nn in nepal for schools is using jlime. He did some request which I will try to help
<wpwrak>
zrafa: maybe they're all quietly happy ? :)
<zrafa>
wpwrak: but no for all of them :) (one was: is there some way to watch ppt (M$office) on nn screen for teachers?
<wpwrak>
zrafa: e.g., i use jlime all the time and it just works :)
<zrafa>
wpwrak: haha.. maybe :)
<wpwrak>
kyak: (STAGING_DIR) yes, i understand. but isn't there a more generic name for such things ?
<qi-bot>
[commit] Werner Almesberger: atusb: remove more unused items (usb_irq, urb_completion) (ben-wpan-stefan) http://qi-hw.com/p/qi-kernel/2723787
<qi-bot>
[commit] Werner Almesberger: atusb: added section titles and moved code in a more logical order (ben-wpan-stefan) http://qi-hw.com/p/qi-kernel/0d9f58b
<stefan_>
wpwrak: we should also move the prink's to dev_info or dev_dbg
<wpwrak>
merge some of them away first :)
<stefan_>
heh
<wpwrak>
here we are
<qi-bot>
[commit] Werner Almesberger: atusb: merged error checking and request completion for control transfers (ben-wpan-stefan) http://qi-hw.com/p/qi-kernel/247b48a
<wpwrak>
*hmm* ...
<wpwrak>
./configure: line 11314: syntax error near unexpected token `NL,'
<wpwrak>
./configure: line 11314: `PKG_CHECK_MODULES(NL, libnl-1)'
<wpwrak>
but it's installed ... (libnl-dev)
<wpwrak>
hates autocrap
<wpwrak>
blargh make -r world: build failed
<wpwrak>
probably  make[3] -C feeds/qipackages/ne compile
<wpwrak>
trying the demodularized approach now
<stefan_>
wpwrak: All still working after all the cleanups. Good. :)
<stefan_>
wpwrak: now we need to find out what happens during iz attach and more
<wpwrak>
kyak: yes "" | make oldconfig  still accepts a number of things
<wpwrak>
yup. something gets sent
<stefan_>
wpwrak: seen it on the air?
<stefan_>
wpwrak: With atrf-txrx?
<wpwrak>
aye
<wpwrak>
maybe you receiver was on the wrong channel
<stefan_>
wpwrak: hmm
<wpwrak>
and the message looks the same as from a ben with atben. nice :)
<stefan_>
wpwrak: aye, 11 != 15
<stefan_>
wpwrak: Really impressed by your remote analyze skills ;)
<stefan_>
same channel gives a packet, yay
<wpwrak>
now, the interrupt ...
<roh>
*g*
<wpwrak>
kyak: make -r world: build failed. Please re-run make with V=99 to see what's going on
<wpwrak>
let'e V=99 then ...
<wpwrak>
grmbl. fighting ubuntu takes about 80% of the time, the actual driver maybe 20% :-(
<wpwrak>
i've now replaced /sbin/init with a shell script. sanity restored.
<wpwrak>
it's amazing just how much doesn't break if you don't launch all their junk
<stefan_>
re
<stefan_>
wpwrak: heh
<stefan_>
wpwrak: to fancy distro choosen? ;)
<stefan_>
I'm wondering if I have any partial PC around to do some testing between two atusb's as well
<wpwrak>
naw, some 8.10 that lacked packages they had already deleted and that didn't handle an update to ludicrous well either
<stefan_>
hmm, maybe I just use the laptop of my gf with the same kvm setup I use here on mine
<stefan_>
yup, seems the easiest solution
<kyak>
wpwrak: $(STAGING_DIR) - not sure what you mean by "more generic". It also doesn't really matter how you name it, does it?
<kyak>
wpwrak: where does it fail this time? Must be some package belonging to image
<wpwrak>
kyak: seems that it was a parallelism glitch. i let it run for ~1 h with V=99 and without -j and it just kept on going
<wpwrak>
kyak: now back to -j4
<kyak>
wpwrak: i dared to do a full build with -j2 once, it failed only once during the process :)
<kyak>
-j4 might be too much for it
<stefan_>
re
<stefan_>
wpwrak: having my test setup with two different pc's now
<stefan_>
wpwrak: somewhere from at86rf230_channel
<stefan_>
Interesting. It does not happen if one tried iz assoc before with the same channel
<wpwrak>
(izcoord) ah, how ?
<wpwrak>
jus jad a little nap to refresh my brain :)
<wpwrak>
meanwhile, OpenWRT build failed around libsdl_mixer .. let's see if this was another parallelism glitch
<wpwrak>
ah .. that's why  my interrupts don't get handled ;-)
<wpwrak>
stefan_: shall i just revert 88bf9af396b0ad297328394c29d6ab5ea04e5aac ?
<wpwrak>
opwnwrt build now fails at  make[3] -C feeds/qipackages/pygame compile
<wpwrak>
next at  make[3] -C feeds/desktop/data/hicolor-icon-theme compile. the failures now become more frequent. may have reached some convergence point
<kyak>
))))
<kyak>
you sould definitely plot it
<wpwrak>
it's a pity that "make" doesn't give any hint at all what happened
<kyak>
maybe it some kind of a new law
<wpwrak>
it would be nice to at least know which package failed ...
<kyak>
you mean, not without V=99?
<wpwrak>
yup
<wpwrak>
with V=99, you get drowned. without it, you're blind. there's something roughly in the middle that's missing ;)
<kyak>
yeah, it's hard to understand what exactly has failed when using more than 1 jobs
<wpwrak>
stefan_: reverted your patch. let's see if the interrupts now get handled ...
<wpwrak>
meanwhile, in the openwrt camp: next stop,  make[3] -C feeds/qipackages/qstardict compile
<wpwrak>
i wonder if openwrt build just randomizes the package order for each run
<kyak>
that's unlikely :)
<wpwrak>
ah, and where do the colors come from ? somebody seems to use a black xterm background. this is very nice, except for those who don't, and who then see things like lime text on beige :-(
<wpwrak>
next stop:Â Â make[3] -C feeds/qipackages/surfraw compile
<kyak>
it's failing on something else
<kyak>
run in a single job if you need to see it
<wpwrak>
last time i went to single job, it didn't hit anything for more than an hour
<wpwrak>
i don't quite trust that build process :)
<kyak>
it just didn't get to the failure point, it must be in another branch of tree that make is building
<wpwrak>
now running single-threaded ...
<wpwrak>
hmm. i get one interrupt, but not two. everything else looks okay. weird.
<stefan_>
wpwrak: (revert) please do. It was just for debugging and failed in earlier stages due to expecting an int
<wpwrak>
kyak: the single-threaded build now runs well, as before when i tried single-threaded
<stefan_>
wpwrak: ah, you already reverted
<wpwrak>
fascinating. now i've managed to hang dmesg ;-)
<stefan_>
heh
<stefan_>
updates
<wpwrak>
ah not, it only becomes invisible :)
<stefan_>
quite useful such an invisible dmesg ;)
<wpwrak>
stefan_: shall i push my current heavily hacked-up version ? (tons of printks all over the place)
<stefan_>
wpwrak: no need
<stefan_>
I stay with what I have and see if I can figure out why it dislikes the channel setting
<stefan_>
Not that easy to debug as it hangs kvm completely
<wpwrak>
sounds suckish
<wpwrak>
kyak: single-threaded build seems fine. just horribly slow.
<stefan_>
hmm, need to route the serial console out of kvm it seems. Can't make a clue of the problem without it
<kyak>
wpwrak: as i mentioned, -j2 wasn't so bad for me.. perhaps you could try
<wpwrak>
interrupts seem to work more or less. let's see if i can actually have some meaningful conversation ...
<stefan_>
wpwrak: yay
<wpwrak>
kyak: ah, now i'm hitting something.  make[3] -C feeds/qipackages/pygame compile
<kyak>
pygame.. not sure if it's a part of the image
<wpwrak>
stefan_: eceived short address 8001, status 00 ;-)
<wpwrak>
cleaning out some of the excessive debugging stuff
<stefan_>
wpwrak: wtf?
<stefan_>
wpwrak: hero!!
<stefan_>
wpwrak: izchat also working?
<kyak>
wpwrak: yeah, it is.. You can just disable it from menuconfig for now
<wpwrak>
kyak: more like -j3 ... also wrestling with ubuntu :)
<kyak>
wpwrak: hm, that's interesting. You have smpeg installed on host?
<wpwrak>
apparently yes
<kyak>
it's a bug in pygame then
<kyak>
i'll write it down :) just disable it for now
<stefan_>
ha, -nographic and some serial foo for the kernel and getty gives me all the kvm output on the terminal I start it. great :)
<wpwrak>
oh, it gets even more interesting
<wpwrak>
ah wiat, thats my xterm that's confused ...
<wpwrak>
lemme fix that first
<wpwrak>
removed. running with j4 again
<stefan_>
holy shit this stack needs some more checks on input data
<wpwrak>
stefan_: error checking ? i didn't see any ;-)
<stefan_>
wpwrak: giving no address behind -s for the coordinator let the whole kernel crash badly :)
<stefan_>
wpwrak: giving him the usual -s 1 and all is fine
<wpwrak>
stefan_: also launching izcoordinator regularly produces some erhm, interesting output
<stefan_>
wpwrak: waiting for your int code now to test myself between two stations
<wpwrak>
just a moment. almost done trimming down the printks ...
<stefan_>
wpwrak: take your time. Need to get some sweets and something to drink
<stefan_>
feels great btw that its near to working now
<stefan_>
even if it took me over a week for the small steps and you only two days to get it into a working state
<stefan_>
I put this under experience ;)
<wpwrak>
get a baileys, drambuie, or jaegermeister, then you have both :)
<wpwrak>
(experience) yeah, it helps sometimes :)
<stefan_>
wpwrak: btw, HQ of jaegermeister is only 18km from here :)
<wpwrak>
wow. have you installed a pipeline yet ?
<stefan_>
wpwrak: but I had enough bear, whisky and rum the weekend. Partying until dawn two days in a row was pretty hard though
<stefan_>
wpwrak: quite funny that a lot people now about wolfenbuettel because of jaegermeister
<wpwrak>
stefan_: tell me about it .... FISL was taxing
<stefan_>
wpwrak: :)
<wpwrak>
*grin*
<wpwrak>
oopsie. now i broke something
<wpwrak>
hmm. seems that  can't defer interrupts the way i'm doing it now
<wpwrak>
guess i'll need a tasklet
<wpwrak>
let's see if the thing now always crashes ...
<wpwrak>
kyak:Â Â make[3] -C target/sdk install takes an interesting amount of time
<wpwrak>
ah, bzip2 :)
<wpwrak>
doens't crash all the time. good. time to commit
<kyak>
wpwrak: yeah, but good news is that you are almost done :)
<wpwrak>
done ! :)
<qi-bot>
[commit] Werner Almesberger: Revert "ieee802154/at86rf230: Be more verbose and disable irq handling for now" (ben-wpan-stefan) http://qi-hw.com/p/qi-kernel/7a7da98
<qi-bot>
[commit] Werner Almesberger: atusb: moved firmware version info before AT86RF230 driver bringup (ben-wpan-stefan) http://qi-hw.com/p/qi-kernel/4c977e4
<stefan_>
wpwrak: izchat works fine here
<wpwrak>
kewl :)
<stefan_>
wpwrak: thats really great. Finally I can go back to work on the layers above the kernel
<stefan_>
wpwrak: Need to start working on my basic lowpan perf tool again
<wpwrak>
there are still a few issues left. fixing ...
<stefan_>
wpwrak: already tested dirtpan?
<wpwrak>
not yet. still have too many known bugs :)
<stefan_>
heh
<qi-bot>
[commit] Werner Almesberger: atusb: initially disable the interrupt delivery tasklet (ben-wpan-stefan) http://qi-hw.com/p/qi-kernel/206feca
<stefan_>
wpwrak: the junk behind the date? Buffer to long?
<wpwrak>
not NUL-terminated
<wpwrak>
see commit log ;-)
<wpwrak>
new acronym: RTFL ;-)
<stefan_>
build[retval] = 0;
<stefan_>
Thats a neat way of express a NULL termination :)
<wpwrak>
pretty standard, no ?
<stefan_>
It should as the len is often given by the return value but I never seen it like this
<stefan_>
perhaps I just don't remember
<wpwrak>
kyak: interesting ... in my build from Oct 30Â Â 2010, my toolchain was toolchain-mipsel_gcc-4.3.3+cs_uClibc-0.9.32
<wpwrak>
kyak: now it seems to be toolchain-mipsel_gcc-4.3.3+cs_uClibc-0.9.30.1
<wpwrak>
kyak: does this make sense ?
<wpwrak>
hmm .. warning: implicit declaration of function 'strtof'
<wpwrak>
yet i #include <stdlib.h>, which is also what POSIX says i should do
<wpwrak>
it links okay, so the function is around
<wpwrak>
and no SDL, it seems :-(
<wpwrak>
yeah. no sign of it
<wpwrak>
sigh
<wpwrak>
reverted to older but more up to date toolchain. let's give this another try in a while then
<qi-bot>
[commit] Werner Almesberger: atusb: implement interrupt synchronization via a 10 ms delay timer (ben-wpan-stefan) http://qi-hw.com/p/qi-kernel/eb1a51e
<qi-bot>
[commit] Werner Almesberger: atusb: updated to do list and added missing \n to dev_info (ben-wpan-stefan) http://qi-hw.com/p/qi-kernel/1b9d72f