<kyak>
i wonder what would happen if someone commits now
<kyak>
would my commit go on top due to the time?
<xiangfu>
kyak: no
<kyak>
xiangfu: ok, so there seems to be no problem with obfuscated time in git?
<xiangfu>
git commit is using something like parent child. you commit parent is "c631197a379aede07e8977a04274e99749bd17b4"
<wolfspraul>
yeah I think git is completely driven through the guids
<xiangfu>
it will never go to front of "c631197a379aede07e8977a04274e99749bd17b4", unless your commit goto upstream(or other branch) and we rebase current branch on top of it.
<kyak>
hmm ok.. but then it is possiblt to find the cheating with the next commit
<kyak>
if the next commit is correct in time, and the parent is my commit with obfuscated time, then it is clear that my time is wrong
<kyak>
don't worry, my employer is not that bitch.. i just think it's fun :)
<kyak>
always commiting at midnight
<xiangfu>
kyak: but I think 'git merge' is using data time. not sure, need find some document about 'git merge'
<kyak>
xiangfu: btw, could you update openwrt-xburst to the latest backfire?
<kyak>
there is cmake.mk backport, and David wanted to have it badly..
<xiangfu>
kyak: sure.
<kyak>
it will also hel pget rid of ERROR: please fix package/feeds/desktop/tint2/Makefile
<kyak>
tint2 is using cmake.mk
<kyak>
xiangfu: could you document this procedure? Or, could it be automated?
<kyak>
so we would catch up on backfire every once in a while
<xiangfu>
I remember I write one email , let me find out that.
<xiangfu>
it's not hard, only sometime need manually resolve the conflict.
<kyak>
ok, so it requires some manual job
<kyak>
xiangfu: another thing, is that every time after such merge something bad happens to my repo :)
<kyak>
for example, last time i got 126 commits ahead
<kyak>
and then after i commited a change, qi-bot gone crazy
<kyak>
though through the web interface i only saw my commit
<xiangfu>
after rebase, you better 1. backup non-commit work,  2. git fetch -a  3. git reset --hard origin/master  4. restore your work.
<kyak>
ok, let's assume i missed the rebase annoucement in IRC
<kyak>
how do i know if "git pull" is enough or i need to do the above?
<kyak>
(backup non-commit work - i assume it's git stash)
<xiangfu>
kyak: hmm...
<wolfspraul>
xiangfu: I think the 'updating openwrt-xburst to latest upstream backfire' procedure should be documented in the wiki
<wolfspraul>
just take your mail as a starting point (copy/paste). a wiki page is probably better suited to updating and improving over time.
<wolfspraul>
with a post to the list someone has to find the old post, edit it, resend a new mail with a complete new procedure. quite hard. updating a few bits and pieces on a wiki page is easier and more likely to happen.
<wolfspraul>
maybe it is already in the wiki somewhere...
<xiangfu>
kyak:Â Â 'git pull -r' should resolve that problem too,
<kyak>
xiangfu: the last time, i did git stash and git pull -r.. Then it still complained about conflicts, so i had to abort rebase. And then i did git pull
<kyak>
(and got 126 commits ahead)
<xiangfu>
kyak: the problem is there is conflicts when I do "git rebase" on top of 'backfire', so when you do "git pull -r", those conflicts come up again
<kyak>
i wonder where these conflicts come from? i don't remember that we changed a lot in base system
<xiangfu>
kyak: the rootfs size, and some kernel config.
<kyak>
xiangfu: ok, i'll just follow the "1. backup non-commit work,  2. git fetch -a  3. git reset --hard origin/master  4. restore your work" procedure for openwrt-xburst
<xiangfu>
wolfspraul: yes. I will create that page,
<xiangfu>
kyak: something wrong with my another pc, I will do rebase in 2~3 hours.
<wpwrak>
Jay7: hmm, i have a few GSM modules laying around ... ;-) they're kinda biggish, though
<wolfspraul>
somehow I think Nokia is dead with this, over the next 5-10 years. Also the new CEO did not have enough power to let go enough people, the infighting will start right away.
<wolfspraul>
too bad. at least Meego gets a little more r&d money for a while, which means some more good free sources being developed, and who knows there may be another strategy shift one day back towards Linux :-)
<Jay7>
I know that most part of nokia customers will be changed with time
<wolfspraul>
I think MS does not have enough momentum. they spent 500 million USD in marketing money to launch wp7, and reportedly only sold 2 million phones.
<wolfspraul>
that's 250 USD / phone in marketing costs alone
<wpwrak>
wolfspraul: (nokia dead) yup
<kyak>
i;m pissed at nokia. but it definitely won't die
<wolfspraul>
A minor casualty seems to be the recently announce microsoft-openstreetmap collaboration
<wpwrak>
kyak: remember nortel. also dinosaurs die :)
<wolfspraul>
because now the new pair needs to use the navteq stuff nokia bought for 8 billion usd, so the free stuff is in the way
<wolfspraul>
wpwrak: looking at this today I need to say I would now agree that the new CEO was more or less 'sent' there
<wolfspraul>
he doesn't even have the power to shakeup the board
<wolfspraul>
so he only drives the new alliance contract home (to ms). doesn't look like rescuing nokia, looks more like business development for ms.
<wolfspraul>
all those old guys left will never reformat their brains around wm7, that will be a mega-mess internally
<wolfspraul>
where powerful people with big budgets pull in all sorts of different directions
<wpwrak>
wolfspraul: (osm) not sure. osm is such small fish, they can hide it somewhere else.
<wpwrak>
wolfspraul: (change of direction) yeah, it'll be a mess. at least they didn't kill me-ego outright. let's see if anything happens there.
<wolfspraul>
some will leave, many will stay and work even harder - against the big strategy of their own megacorp
<wpwrak>
wolfspraul: some of the "smooth" transition may also be simply because they may prefer to shut down things bit by bit. e.g., downgrade symbian and finally let it die.
<wolfspraul>
for symbian they already said that
<wpwrak>
wolfspraul: (work against) in the case of meego, this is good :)
<wolfspraul>
sell another 150 million units then 'end development'
<wpwrak>
(symbian) ah, i'm behind then :)
<wolfspraul>
meego sounds to me like he didn't have enough power to shut it down outright, but since he drove his main assignment home (ms alliance), that bit of cost drag can just be dragged along a little longer.
<wolfspraul>
that is the strangest thing I've seen in a while. totally not product or business focused.
<wpwrak>
wolfspraul: you also have to consider that nokia has strong ties to finland. this also complicates mass layoffs.
<wpwrak>
wolfspraul: (not focused) meego ?
<wolfspraul>
wpwrak: let me quote what they say about meego in the press release :-) then you will see...
<wolfspraul>
"Under the new strategy, MeeGo becomes an open-source, mobile operating system project. MeeGo will place increased emphasis on longer-term market exploration of next-generation devices, platforms and user experiences. Nokia still plans to ship a MeeGo-related product later this year."
<wolfspraul>
1. project
<wolfspraul>
2. longer-term exploration
<wolfspraul>
3. still plans
<wolfspraul>
I mean it cannot be any clearer than that :-)
<wolfspraul>
A DRAG
<wolfspraul>
that the new CEO was not able to kill right away because of internal resistance
<wolfspraul>
that doesn't make him look like a good CEO at all, imho. where is shareholder value here? or any value? it's just nonsense.
<wolfspraul>
the meego guys will try to squeeze out as many dollars and days as they can get...
<wpwrak>
wolfspraul: (meego) well, that sounds actually good. remove the pressure. if it can succeed at all, then only if it doesn't have the whole company look over its shoulders.
<wolfspraul>
it will become disconnected from all business processes. marketing, sales, product groups, etc.
<wolfspraul>
but yes, the people there will get some more 'funding', and will try to squeeze out the most of it.
<wpwrak>
wolfspraul: (disconnected) dunno. depends on how they're organized internally. ideally, meego would get its own (small) marketing team that works on a plan.
<wolfspraul>
I know a bit of American business lingo, I can tell you that paragraph is BAD :-) At least until the day the new CEO is ousted, which will surely come.
<wpwrak>
wolfspraul: meego still needs to make the first step: become visible. you sometimes can combine this with launching a successful product, but since it's entering a market with several similar competitors, that's hard.
<wolfspraul>
will you buy the 'still to be released' Meego device when it comes out?
<wpwrak>
wolfspraul: dunno. chances are that i wouldn't because i've gotten used to keeping the feature set of my phones at the bare minimum.
<wpwrak>
wolfspraul: but then, if it's open and "nice", who knows
<wpwrak>
wolfspraul: anyway, for the open source world, today's results seem to be about the best likely outcome. one proprietary system gets replaced with another one. the open system doens't get unrealistic pressure put on it (or at least so it seems), and only victim may be osm, but not even this is clear.
<wpwrak>
wolfspraul: (osm) they may just mention the navigation to make it look less like a writeoff
<wolfspraul>
wpwrak: you and me haven't been big meego believers, so yes, we can see it that cold blooded. many meego guys cannot.
<wolfspraul>
I'm using a super cheap mtk 20 usd phone.
<wolfspraul>
can do calls and sms, and alarm clock.
<wpwrak>
wolfspraul: (phone) i still have my fancy little samsung, whose battery now lasts about two hours standby, 30 second talk time. so it's no longer really "wireless". and then i got the cheapest phone i could find here, a nokia for about usd 50 :)
<wpwrak>
wolfspraul: (phone) i was in a bit of a rush. otherwise, i may have tried to bring up one of the many gta02 is have sitting around ...
<wpwrak>
wolfspraul: in any case, you're asking the wrong questions. they should be: 1) does this affect the marketing message of qi-hw/sharism ? 2) does this create risks ? 3) does it create opportunities ? 4) should any action result on our part ?
<wpwrak>
wolfspraul: i perceive most of this as happening in a world that's separate from ours. so perhaps there's little with respect to points 1 and 2. there could be opportunities in the form of engineers leaving and/or managers with some savings looking for a new mission, but i wouldn't know how to steer these our way.
<wpwrak>
wolfspraul: what's your opinion ?
<wolfspraul>
one sec, phone call
<wolfspraul>
or maybe not 'one sec' but more like "10 min"
<wolfspraul>
wpwrak: 1) [qi marketing] not in my opinion. our message is the same as before. I also wouldn't want to make some whatever announcement, because I would feel it's disrespectful to such a large movement/company like Nokia, Meego, even Qt
<wolfspraul>
I leave that to people who like to talk a lot, and I rather focus on improving the NanoNote and Milkymist One.
<wolfspraul>
2) [risks] no, what risks? don't see anything.
<wolfspraul>
3) [opportunities] definitely. the free phone!
<wpwrak>
wolfspraul: (1) oh, a response doesn't have to be an announcement about the situation :) could be a shift in strategy. or affect an on-going decision process.
<wolfspraul>
4) [action] see #1. I don't see which action of ours could possibly make Meego guys feel better. if the action is just an announcement I'd rather pass.
<wpwrak>
wolfspraul: (3) hmm, seems like before then :) the only semi-direct competitor would be meego, and that one's unchanged so far
<wolfspraul>
the thing that bugs me right now is how to position Milkymist One (and the Milkymist SoC) where it belongs. And there is no way at all how I can see connecting that to Meego or Windows Mobile.
<wpwrak>
wolfspraul: (4) action could also be something that makes us benefit from the situation, not necessarily something that helps them
<wolfspraul>
Sebastien's approach is right, make it a good end user product, and market it as VJ, maybe dance gear like visikord. Totally unrelated to phones.
<wpwrak>
(position mm1) what were the responses at fosdem ?
<wolfspraul>
don't know, haven't heard much from tuxbrain, lekernel or rejon
<wpwrak>
nothing we do at present is much related to phones ;-)
<wolfspraul>
it will bubble up
<wolfspraul>
yeah, exactly
<wolfspraul>
so better keep our mouths shut
<wpwrak>
yet they're all here. speak up, guys ! :)
<wolfspraul>
rejon traveled back to US, currently sleeping :-)
<wolfspraul>
tuxbrain posted a blog entry, but no real directions or summary of Milkymist feedback I think
<wpwrak>
that leaves lekernel for the mission debriefing
<wolfspraul>
if you think I managed badly at om with all our changes, they surely beat me!!!
<wolfspraul>
that looks like 90% negative reactions there...
<wolfspraul>
you must enjoy the sarcasm of some of those posts, I think even the great Werner can still learn a few bits and pieces there, no?
<wpwrak>
lemme see ...
<wpwrak>
funny. lots of voices pro symbian there too. well, understandable in a way, but still ...
<wpwrak>
wolfspraul: of course, maybe nokia are just very shrewd. ally with MS for now, pacify the analysts, get some cash injection, meanwhile beat meego into perfect shape, and then take over the world with it in a year ;-)
<qi-bot>
[commit] Werner Almesberger: fw2/Makefile: use the ATmega32U2 configuration (experimental) of avrdude http://qi-hw.com/p/ben-wpan/85612e5
<wolfspraul>
wpwrak: I really cannot remember such a lopsided feedback, ever.
<wolfspraul>
this is Nokia's official developer forum
<wolfspraul>
and the reactions are not just 90%, but more like 99% negative
<wolfspraul>
sure those are existing developers, maybe Nokia wanted to screw them all because they think THEY failed? :-)
<wpwrak>
wolfspraul: i don't think the response is overly surprising. it's been a 100% non-MS camp so far. MS evokes strong emotions. so what do you expect ? :)
<wpwrak>
wolfspraul: plus, many seem to be on symbian. and it undeniably is bad news for these.
<wolfspraul>
ok maybe we would need to look in Microsoft developer forums to see the other side. good point.
<lekernel>
wolfspraul: (fosdem) not much to say, lots of indifference/passivity from the average fosdem attendee
<lekernel>
we should go for slashdot, engadget, etc.
<wpwrak>
lekernel: no reaction to hearing about the fpga ?
<lekernel>
phew, no, except 2-3 people
<wpwrak>
lekernel: bah, bunch of girls
<lekernel>
the knowledge about electronics/low level programming pretty much sucks there
<wpwrak>
yeah, could be just the wrong forum. hw and open haven't really connected so much yet.
<wolfspraul>
I am already sourcing for rc3, we need to have some fully assembled and fully working boxes in stock when we do the real launch.
<wolfspraul>
lekernel: by then we also need the user sharing forum setup, I will see whether rejon plans anything there. I understand you guys talked but I'm not sure what the plan is.
<kristianpaul>
menos mal mi ultimo telefono nokia se me perdio
<kristianpaul>
ahora tiene un panasonic de hace como 6 años
<kyak>
xiangfu: git stash && git fetch -a && git reset --hard origin/master && git stash pop worked like a charm. Let's see how build would go
<zrafa>
kristianpaul: lo malo es que si meego iba a estar bueno como proyecto open source linux entonces parece que no tiene mucho futuro si usan windows :(
<zrafa>
kristianpaul: y nosotros perdemos a una empresa tratando de hacer un linux open source para celulares :(
<kristianpaul>
zrafa: :/
<zrafa>
kristianpaul: lo cual implica un retroceso en la conquista de ese mercado por parte de linux
<kristianpaul>
zrafa: si es cierto
<zrafa>
kristianpaul: yo diria que con apple, android, y linux el mercado de telefonia esta como muy parejo
<zrafa>
kristianpaul: y si windows viene a cagarl... la vida a linux
<zrafa>
kristianpaul: ganan los otros : apple, etc
<zrafa>
kristianpaul: cuando el mercado se defina yo creo que va a existir un ganador y listo. Ya nadie va a querer probar otra cosa
<kristianpaul>
window ha estado dando patitas de ahogado en este tema... a ver si se pone peor con nokia tal ves diria yo
<kristianpaul>
s/ganador/terminal = para gugol ;)
<zrafa>
kristianpaul: claro, pero el tema estaria es que windows de patitas de ahogado sin joder proyectos linux
<zrafa>
kristianpaul: una onda.. como que hagan arreglos con empresas que no usan linux :)
<zrafa>
kristianpaul: mira en los cuadritos como la guita que estaba destinada a linux va ahora a windows :P
<kristianpaul>
ausente
<M_Rojas>
zrafa: que noticia tiraste ....
<M_Rojas>
menos mal :-S
<xiangfu>
kyak: when compiling full_system, I meet one compile error about package: libxfce4util-4.6.1. cause this package not update to the last NLS strategy. just finish the patch, will send to upstream now.
<zrafa>
M_Rojas: :P
<dvdk>
hi kyak, if your git stash/fetch/reset magic works for you, mind posting a few lines to the mailinglist so other's can take the same shortcut?
<wpwrak>
zrafa_: (nokia) yeah, wolfgang mentioned it already. i'm not sure if this changes much for free software. maybe not even for nokia.
<wpwrak>
zrafa_: symbian was dead already. everybody knew that. in a way, it's good that they're not putting the unfinished meego in the firing line. well, sometimes such things can do wonders to progress, but it would be a huge gamble.
<wpwrak>
zrafa_: the R&D investment slide is scarier, though, depending on when "longer-term" is meant to be.
<zrafa_>
wpwrak: yeah, but meego would seem something nice or not? I do not know it, I just read about. And some projects like ofono and similar were thought to do meego better, and those were free software
<zrafa_>
wpwrak: anyway, red hat or canonical should be some distro for mobile devices... so companies do not need to spend years for one thing made just for them, and also, companies like nokia just can choose: or android or windows if they decide do not invest on making an OS for their devices.
<zrafa_>
should be=should do
<wpwrak>
zrafa_: it seems meego simply isn't ready yet. do it would be very dangerous to gamble the whole company on it. and probably counter-productive.
<wpwrak>
zrafa_: now all the symbian folks will hate windows. and all the pressure on selling things (and the blame for failing) will be on windows, too. so meego gets more time to mature.
<wpwrak>
zrafa_: if they're smart, they'll use it wisely
<wpwrak>
zrafa_: android would have been "safer" than windows, agrees. maybe they expected too much competition. or maybe ms promised them big $$$. maybe they also didn't want something too close to meego.
<mstevens>
is rteading about the nokia deal, so far I think nokia is screwed.
<zrafa_>
wpwrak: what is needed is a linux distribution for business phones, that is why I mention red hat for example. So it is an option at least. So far, there is not. Just android or windows, or some propietary OS
<wpwrak>
zrafa_: the distribution per se is the easy part. the applications are difficult, because phone users expect things to be tightly integrated.
<wpwrak>
zrafa_: so you need to write a whole set, and it must be good. difficult and time-consuming, if the story of openmoko is any indication :)
<zrafa_>
wpwrak: yes, I know. I mean distribution as a whole thing
<zrafa_>
like android, but more linux :)
<zrafa_>
and some company like redhat or similar behind
<wpwrak>
zrafa_: of course, you could try somethings radical, and have the "shell gui" i half-jokingly suggested. that may actually be a success for a while due to novelty value.
<wpwrak>
zrafa_: yes, a solid company with a compatible mission behind it would help a lot.
<zrafa_>
wpwrak: I did the shell gui, now people need to write the scripts to use the phone :)
<kyak>
dvdk: hi! i've got another magic for you: GIT_AUTHOR_DATE=$(date -R -d '23:59:59') git commit :)
<kyak>
the commit will be timestamped midnight this day
<kyak>
dvdk: regarding the git stash/fetch/reset, sure, i don't mind you posting it to mailing lists.. in fact, it was suggested by xiangfu, so you should credit him
<dvdk>
is looking at the date manpage
<dvdk>
kyak: not bad.  will add a shortcut to my bin directory :)
<dvdk>
kyak: looking up important 'posts' in the irc log is even more annoying then searching my emil :)
<dvdk>
kyak: if it works for me tomorrow then I'll post it
<kyak>
dvdk: it's not that much talking :) but if it is really big, usually grepping for your nick helps
<kyak>
oh, btw i just saw an e-mail from xiangfu
<kyak>
he explained the git stash/fetch/reset thing
<dvdk>
is checking his email
<dvdk>
did not find any new mails from xiangfu
<dvdk>
remembers that the qi-hw mailserver usually works at a snail's pace
<kyak>
the e-mail subjected "[UPDATE openwrt-xburst] rebase on last upstream backfire", arrived 1 hour ago
<dvdk>
seems i overlooked the paragraph "notice:" when i looked at the maila
<wpwrak>
dvdk: can't complain about the server's turnaround time. maybe a minute or so.
<rjeffries>
Nokia + MS on some but n ot all Nokia hardware is prolly a smart move. They have NOT killed Meego, so we shall see
<rjeffries>
they need to keep their options open. killing Symbian had to be 1)a very difficult decision 2) obviously required
<rjeffries>
Keep a close eye on HP now. WebOS could becone very sucessful.
<viric>
rjeffries: what is that webos?
<kristianpaul>
argg damn isp !
<dvdk>
wpwrak: for me it seems to take hours to get an echo from the mailinglist
<rjeffries>
viric that is the new software/OD that Palm designed por Palm Pre etc abouy a year or so ago. They were bought by HP
<rjeffries>
dvdk agree mail list feels slow. At least for me, I never see my own messages. Known bug if I recall correctly.
<dvdk>
rjeffries: i *do* see my own mails.  just not immediately
<viric>
rjeffries: what is it for?
<dvdk>
has the theory that every mail send to the mailinglist has to wait for approval by invisible censors.
<dvdk>
or mybe it's just to lure people into using irc instead
<rjeffries>
dvdk I have wondered myself. wolfspraul can comment maybe he moderates the list
<roh>
if one gets its own mail is config
<roh>
same goes for moderation.
<roh>
often mail ends up in moderation when people dont send it correctly or make it look like spam.
<roh>
then the listbot holds it till a human decides what it is
<wpwrak>
that human should then complain if the people make a habit of sending things the wrong way ... (well, or let mailman add them to the address filter, if that is the problem)
<roh>
wpwrak: believe me if i say: one tends to get a thick skin of ignorance after some time.
<wpwrak>
roh: i see no reason to belive this, particularly not coming from you ;-)))
<roh>
some people just dont learn. thats what i learned... e.g... we had people not even able to keep their mailclient working a week without needing us to reset the pw
<wpwrak>
roh: names ! names ! :-)
<roh>
not in public
<wpwrak>
i was just kidding ;)
<roh>
M)
<wpwrak>
sigh. i think the time has come to replace freakusb. i think their handling of SETUP with large OUT data is irreparably broken :-(
<rjeffries>
wpwrak I gave a Ben preso to 12 older guys at Linux meeting last night
<rjeffries>
they were polite. ;)
<wpwrak>
rjeffries: ;-)
<rjeffries>
one guy said he'd like one, but would need to ger a few extra hours a day (he is a bust consultant)
<rjeffries>
another guy may want one for his grandkid.
<wpwrak>
rjeffries: i think you really need to show a device. the form factor needs to be experienced in real life
<rjeffries>
I am arranging to borrow my friends Nanonote. It is a "drawer comuter" (not in use
<wpwrak>
rjeffries: (bust consultant) hmm, if he's based in the LA area, that could be a nice business :)
<rjeffries>
s/bust/BUSY!/ smile
<roh>
freakusb?
<wpwrak>
roh: the usb stack (from freaklabs)
<roh>
wpwrak: why should one use that?
<rjeffries>
he is leader of the group and put Nanonote stuff on web page:
<wpwrak>
roh: well, i tried to use a "proven" stack first. at least see that the basics work
<wpwrak>
roh: among the available stacks, freakusb is the leanest that supports this chip
<roh>
wpwrak: lufa is proven.
<wpwrak>
roh: and way too big
<roh>
wpwrak: huh? isnt that a question of compiling it with the right featureset?
<wpwrak>
roh: does tons of things that i don't need, that don't fit, etc. adapting it would probably be more work that writing what little i actually need
<roh>
i hear of freakusb the first time now.. and it cant do much, and doesnt even have a repo available... thats really bad style
<wpwrak>
roh: heh, even compiling with the right feature set would be interesting :) have you looked at just how many options are in there ? :)
<roh>
doesnt take people who distribute source as zipfiles serious
<wpwrak>
roh: yeah, their distribution is a little odd
<roh>
wpwrak: yes. thats what makes lufa great. its used in real world. even commercial applicances.
<wpwrak>
roh: lula has nice stuff on top of the basic. none of which is any use to me. all i need are control transfers and an interrupt endpoint.
<wpwrak>
errr, luFa
<roh>
eeh.. what do you need an usb stack for then?
<wpwrak>
roh: but if someone wants to adapt lufa, i'm not stopping them ;-)
<roh>
i mean.. isnt that what the hw gives you after some very thin layer?
<wpwrak>
roh: register/buffer read/write, a few device commands, send one byte when an interrupt has arrives. that's it.
<wpwrak>
roh: and enumeration
<roh>
so you dont want a usb stack, only a lowlevel driver
<wpwrak>
roh: yes. ijust need the bottom layer
<rjeffries>
roh: the guy behind Freaklabs is a cool, clever person. very creative. his idea of hooking up Wireshark to allow analysis of 802.15.4 is sweet
<roh>
btw: i hope you only see the 'control the radio from the pc' thing as a intermediary solution
<wpwrak>
roh: (intermediate) not really. the higher-level driver would be identical for atusb and atben.
<roh>
rjeffries: maybe. still people without public svn or git repos... how can one take them serious or interact with them?
<wpwrak>
roh: you could of course write alternative firmware that does other things
<roh>
wpwrak: sorry.. but that will kill any acceptance in real world.
<wpwrak>
roh: our real works is linux. there it will work very well :-)
<roh>
no.
<roh>
it would mean that you get support requests about compiling it in all kinds of shitty os and environments.
<wpwrak>
roh: no. kernel driver. one kernel to rule them all :)
<rjeffries>
roh if you say we need Windows based dariver to talk to atUSB I would agree
<roh>
it means it doesnt work on every openwrt by default but needs code to be compiled
<roh>
rjeffries: even that.
<wpwrak>
roh: it will need code anyway
<roh>
wpwrak: you know that there already IS a driver for the radio you use in linux?
<wpwrak>
roh: ieee 802.15.4 isn't just another ethernet. there are substantial differences.
<wpwrak>
roh: yes, for the 230. very primitive, though.
<roh>
wpwrak: sure. doesnt mean it doesnt need to go into the kernel somehow.
<roh>
my point is: abstracting the 'usb end' of it to something like cdc eth would make real-world compatibility true. what you are designing is 'open but proprietary'
<wpwrak>
roh: oh, it has to go into the kernel. along with other things. for example, 6lowpan has header compression.
<roh>
i thought all that code was already written?
<roh>
by someone else
<wpwrak>
roh: abstracting it to look like ethernet in a meaningful way may be quite tricky. you may basically have to implement a router.
<roh>
true.
<wpwrak>
roh: (code written) not sure. there's some stuff in the contiki os, but i don't know how much and how this connects to linux. please bear in mind that the most common use of 802.15.4 is to talk to sensors. so many people don't need a "real" stack.
<roh>
even abstracting it to a serial would be ok.
<wpwrak>
roh: the problem doesn't change if you change the link layer
<wpwrak>
(virtual link layer, in a sense)
<roh>
wpwrak: my point is: if you need to install extra code everywhere to use it. it looses even more acceptance we dont have
<wpwrak>
roh: i take it as given that you need extra code. so there's no point in trying to fight it.
<wpwrak>
roh: you may be able to avoid it in the long run, but that may be many months of development.
<roh>
wpwrak: also you are killing your lateny with the 'dumb dongle' approach i presume.. or do you handly anything on the mcu at all?
<roh>
irgh. handle
<roh>
this keyboard sucks.
<wpwrak>
roh: the mcu will do very little. basically just execute register accesses. btw, there are sources ;-)
<wpwrak>
roh: what i don't handle yet are interrupts. for now, i just poll, but that's of course not nice. my idea is that the mcu will read the interrupt status register and send it over a bulk endpoint.
<wpwrak>
roh: latency ... yes, usb is slow. but that only matters for a few applications
<steve|m>
wpwrak: maybe using an interrupt transfer containing the interrupt status register to signal the host there was an interrupt?
<roh>
nah. it also matters for transfer speed.
<roh>
i dont think you will have more than a few kbytes/sec if you keep usb latency for every irq.
<wpwrak>
roh: (efficiency in general) it's always possible to add combined operations if necessary
<wpwrak>
steve|m: err yes, bulk = interrupt, more or less :)
<roh>
it will wait more for the usb than to transmit/recieve data
<wpwrak>
roh: let's see. you can always optimize when you encounter problems.
<steve|m>
huh, can the device send bulk data if the host hasn't asked for it?
<wpwrak>
steve|m: depends on what level you consider ;) it can enqueue bulk data, but it has to wait for an IN transfer asking for it.
<steve|m>
"However under USB if a device requires the attention of the host, it must wait until the host polls it before it can report that it needs urgent attention!"
<steve|m>
ah, damn
<wpwrak>
roh: the most timing-critical part will be back-to-back frames. where back-to-back still includes some quiet time, because frames are acked, have a preamble, and so on.
<steve|m>
i thought the interrupt transfer was a 'real' interrupt
<roh>
wpwrak: btw: i am not completely confident yet that the 6lowpan will work out.. so we'll see
<wpwrak>
roh: so if we're too late, new frames would be lost, which would be quite bad for performance. at some point in time, this will need examining. if necessary, the mcu could signal arrival, and then copy the frame to a buffer.
<roh>
work out  in matters of performance, latency etc
<wpwrak>
roh: we could also run ipv4 over ppp-over-ieee802.15.4 ;-))
<roh>
frames seem to be much too small to get serious throughput.
<wpwrak>
steve|m: yup. usb is all about polling. that's also why its power efficiency isn't exactly stellar.
<roh>
it really seems to be a technology intended for gas gauges and powermeters and not for using the interwebz
<kristianpaul>
<3 ppp
<wpwrak>
roh: (small packets) remember atm ? that had only 53 bytes :)
<roh>
steve|m: interrupt in usb means'polling'
<roh>
and interrupt transfer just means 'highes prio, before all bulk'
<roh>
+t
<wpwrak>
roh: packet size is never a problem. duty cycle could be, though. also this needs to be seen.
<roh>
wpwrak: atm was made by stupid people. you know how it got to it?
<wpwrak>
roh: oh yes. i know atm very very well ;-)
<roh>
it was a 'compromise' which made both sides unhappy. the 'telephony guys' and the  'packet guys'
<roh>
from my pov is complete and utterly desaster. thats why technolgy doesnt progress. stupid decisions made to compromise for getting stuff unified where it shoudnt be.
<wpwrak>
roh: if you like scary stories, familiarize yourself with the concept of an "atm adaption layer". start with AAL5, which is kinda nice, made by "data" people. then look at AAL3/4 to see what the telephony side has come up with.
<roh>
would rather live in a world with both systems. line AND packet based. they both have their specific usecases and pros and cons. and none can replace the other.
<rjeffries>
wpwrak: I lived the ATM nightmare. Made some good money off my newsletter; ATM USER
<roh>
our telecomunications systems on this planet would only eat half the power and need half the space and cost half as much money if we would just stop billing by useage.
<wpwrak>
roh: then you may want to direct your view to the signaling. Q.2931 to be precise. there, you may want to savour the structure of the Broadband High Layer Information Element and the Broadband Low Layer Information Element.
<roh>
flat per port. it would be so much cheaper.
<rjeffries>
actually ATM was not made by any stupid people. It is connection-based, but connextion-LESS won the day
<wpwrak>
roh: (billing) of course. billing is crazily complex. also inter-carrier billing is a complete branch of mad science.
<rjeffries>
MPLS is conncetion oriented over IP, and serves a useful purpose
<rjeffries>
roh you are tolting after yesterdays (actually, last decade) telecom issue. thanks to Internet, a huge amount of voice is now essentially FREE
<wpwrak>
rjeffries: oh, there were a lot of interesting thought patterns in atm ... e.g., remember how desperately some people tried to dethrone TCP ? every few months, they hatched some new crazy scheme. i liked the one where they suggested to just replace TCP with SSCOP. shows some deep insights into the matter ;-))
<wpwrak>
rjeffries: MPLS came very very late, IETF-driven, when they realized that their telco madness didn't go anywhere.
<roh>
rjeffries: the internet didnt free voice. it broke it.
<roh>
voip is CRAP.
<rjeffries>
Ok roh i sure enjoy my FREE Skype calls and my FREE Google voice YMMV
<rjeffries>
nop VOI is far from crap. I worked for 8 years at a company that made good money with a system for TELCOS that took advantage of VOUP, then had gateway to standrad telephony
<roh>
rjeffries: i can imagine that for 3rd world countries like the us when it comes to the phone system voip is interresting. for its a step into the wrong direction. NONE of the features, less availability. no serious quality. no thanks
<roh>
i got sip myself here.. i still use my mobile for most of the calls. why? IT WORKS. sip does only in 2/3 of the cases.
<rjeffries>
roh you are confusing VOIP with internet voice which uses VOIP. VOIP can deliver great quality voice when it is used in a well engineered network.
<rjeffries>
the comany I was emplyee #6 at just sold to another comany for $186 Million
<roh>
rjeffries: please come back when you used a proper telephony system. ss7 based.
<rjeffries>
they sell equipment to traditional telcos and NOBODY complains about voice quality
<roh>
and money doesnt impress me.
<roh>
i guess some billards of flys eat shit every day without complaining to the cow. does that make the shit good?
<rjeffries>
roh gaod bless you. I know about SS&, d'oh and am well versed in telephony. Your opinons are your opinions, but there are solid existence proof that connectionless voice works very well
<rjeffries>
the INTERNET as tarsnport can be dicey but VOPI over an engineered network is first rate
<rjeffries>
s/gaod/God  ;)
<roh>
rjeffries: whe i do a _landline_ call here in .de i get a uncompressed, 8bit clean, basically latency free 64kbit channel. and lots of additional features. with ip i got more latency on the first hop than with isdn to my peer.
<roh>
there is a reason the packet and the line people could not find a packetsize which is good for both. THERE IS NONE:
<rjeffries>
s/ss&/SS7Â Â look, I am well versed in telco world. don;t assume I am ignorant. ;) I have written tens of thousands of words that peopel paid me for on these topics.
<roh>
try sending a fax via voip.
<rjeffries>
roh you miss the point. ARM and it's 53 byte cell size was indeed a compromise.
<rjeffries>
ATM is no longer where the action is. It is used in limited cases for trunking
<roh>
it was a BAD compromise. none would be better
<rjeffries>
IP is used maybe 85 percent of time and IT WORKS
<rjeffries>
i will now eat breakfast. a late brekfast.
<roh>
my point is: there doesnt need to be a compromise always. sometimes you need 2 technologies.
<roh>
ip would be better off with bigger packets. voice with smaller.
<rjeffries>
I am not ANTI circuit switced at all. but it is a more expensive technology, and is fading FAST
<rjeffries>
roh I gues syou are not aware that VOIP uses small packets over IP.
<rjeffries>
biab
<roh>
rjeffries: its only expensive if you have 'pots' as your default.
<roh>
overhead per packet is what kills the cow.
<roh>
.oO(hey.. why am i discussing this... the us is also the country with 56k isdn lines m-)
<roh>
sorry for the OT. i should have stopped when i realised i talk with somebody from the us.
<rjeffries>
roh you are a smart and even funny guy. we also has ISDN but it is pretty much irrevelant.
<rjeffries>
and yes let's move on.
<roh>
rjeffries: i know. sorry if i sounded a bit harsh. but us telecom markets.. thats something for me to laugh about, when not cry. not to celebrate
<roh>
rjeffries: i still dont have proper clip, clir, and all the isdn feats on sip.
<rjeffries>
roh no problemo, since I have worked in telecom and data com for maybe 20 years or more, these topics (and arguments) are not new to me;)
<roh>
and sip is really bad when it comes to reliability. i know that from experience. thats why i use my mobile. gsm works. very well.
<rjeffries>
SIP does have issues, including complexity and interoperability (because of teh complexity)
<rjeffries>
that is why MGCP is used in some teclo VOIP to Telco gateways\
<rjeffries>
did wpwrak go to sleep? or is he just hiding?
<roh>
on a landline (even an analog one here) i cannot remember the last outtage. not even when the power grid fails for a second ( and havent seen any real outage for more than seconds, and once minutes on the power grid in my 30 years on this planet)
<roh>
dsl over which in the end the sip must run, fails even when the weather changes.
<wpwrak>
rjeffries: just watching the show :)
<roh>
i dont wonder. its up to 30mhz over unshielded copper.. i can accept that for ip. (when it comes back in seconds). when my phone doesnt work because of that or doesnt properly come back up. thats fail. and thats how i know sip.
<roh>
i dont blame it purely on sip. it IS complicated (see the amount of specs for isdn)
<roh>
same happens with skype and all others. especially when nat is involved
<rjeffries>
my previous company has maybe a million lines of telephony. POTS from house to a cabinet (this is typically a drop of a few thounad feet
<rjeffries>
the POTS is converted to VOIP and backhauled over an engineered IP network
<rjeffries>
in some cases it hits an IP base sofswitch, which thenmay ytansport the call over an IP trunk
<rjeffries>
or... if person is calling his neaighbor, it remaiins VOIP all the way
<rjeffries>
easy peasy lemon squuezy. nit easy to tune up but when it is
<rjeffries>
people are delighted
<rjeffries>
anyway I am boring myself. ;)
<roh>
no way sending a fax works
<roh>
fax doesnt like jitter or compressing codecs at all.
<Jay7>
btw
<Jay7>
I have idea to create UPS-server
<Jay7>
to convert usb-connected UPS into network one
<Jay7>
I can do it from software side but need some platform under with usb-host and ethernet
<Jay7>
any suggestions?
<steve|m>
a usb device-server? ;)
<Jay7>
kind of :)
<Jay7>
may be combined with printserver
<Jay7>
one HW, different 'firmware'
<steve|m>
I'd use something like a router/seagate dockstar, supported by openwrt
<Jay7>
I have one suggestion about something like D-Link DIR-320
<xMff>
ah another so called opensource router
<Jay7>
looking good but have only 4Mb of flash.. may be hard to debug first time :)
<xMff>
wouldn't use it
<steve|m>
still regrets taht he bought only _one_ dockstar when they were cheap..
<kyak>
owns old good linksys wrt45gl
<roh>
steve|m: is it good hw?
<roh>
i heard its quite flakey
<roh>
Jay7: check out something which runs openwrt with >=32mb ram and >=8mb flash.
<roh>
availble from 20E on
<roh>
new.
<Jay7>
ServerPlug is looking good
<Jay7>
but about $100
<steve|m>
roh: the hardware is quite good.. what do you mean with flakey?
<steve|m>
well, it isn't really fast enough for real gigabit-ethernet, but apart from that..
<roh>
steve|m: temperature problems like on the sheevas
<steve|m>
hm, I ran mine several weeks without any problems, it stayed quite cool.. also, we have one in the CCC FFM hackerspace, which is already running since months
<rjeffries>
wpwrak your spam assasin prevent me sending you an email about the PCB Fab process no biggie but scoring is broken
<rjeffries>
wait i am wrong here is real issue
<rjeffries>
werner@almesberger.net
<rjeffries>
  (generated from werner@openmoko.org)
<rjeffries>
  retry time not reached for any host after a long failure period
<rjeffries>
stebe| I am also a Dockstar fan Still own two. semt one to a freind in Sweden
<rjeffries>
13:02 <rjeffries> steve|m I am also a Dockstar fan Still own two. semt one to a freind in Sweden
<rjeffries>
Dockstar does not run hot
<viric>
Jay7: $100 is not much for someone from the first world
<wpwrak>
rjeffries: oh, the server doesn't quite work at the moment. i'm surprised it accepts any mails ... lemme stop that
<rjeffries>
whatever
<wpwrak>
rjeffries: (retry time) ah, that sounds more accurate. yes, the server is down. was replaced after hardware failure and needs manual restoring.
<roh>
viric: depends. not for gbit hw. true. i find it too expensive for 'eth-usb
<wpwrak>
rjeffries: in any case, whatever almesberger.net might say, openmoko.org mails are also stored at openmoko.org, which is where i actually read htme
<wpwrak>
s/htme/them/
<viric>
roh: it even has a cipher coprocessor
<rjeffries>
ok maybe I should just use openmoko email addr then
<roh>
viric: maybe.
<viric>
roh: I enjoy a normal gnu system there. There is a huge step between 16MiB of RAM or 512
<rjeffries>
viric: cipher processor in dockstar is basically useless there is no source of randomess. other flaw is no real time clocl but neitehr matter for my humble uses
<viric>
rjeffries: what is that dockstar?
<roh>
hrhrh.. no rtc? have heard that joke before today
<wpwrak>
no rtc at all ? that's a bit cheap :-)
<viric>
hm I think it has a RTC
<roh>
some idiots at nxp left the rtc out on some of their arm series lpc
<rjeffries>
it is from Segate google Segate Dockstar real cheap was $20 at several places includding WOOT
<rjeffries>
is discontinued but still available
<viric>
rtc-mv rtc-mv: rtc core: registered rtc-mv as rtc0
<viric>
rtc-mv rtc-mv: setting system clock to 2010-12-19 21:32:12 UTC (1292794332)
<roh>
my guess is it was subventionizedy by hdd sales
<viric>
ah ok
<viric>
I was not talking about the dockstar
<roh>
such they make no money but loose money on selling it for 20E
<rjeffries>
uses the Marvell refernce design more or less has 4 USB ports 1xGbe no display no keybaord has 128MB Ram and 2MB flash
<rjeffries>
really cool
<rjeffries>
rohL yes that was CLOSEOUT price still there are several similar Marvell based little machines at around $100 USD
<roh>
i just dont believe you can sell something that cheap and deliver proper quality.
<rjeffries>
it is the old NSLU on steroids with a LOT more RAM and multiple USBs
<roh>
same goes with 'ip' or telephony.
<rjeffries>
roh you  miss the point
<rjeffries>
at $100 they made some money (there's almost nothing inside)
<rjeffries>
yes their objective was ti sell HDD
<rjeffries>
BUT........ you do Not need to use a seagate HDD
<roh>
nlsu somehow bedazzled me.
<rjeffries>
by the way it also has JTAG (slightly hacky)
<roh>
i mean.. it sucked for the usecase it was originally sold for ;)
<rjeffries>
and I2C
<roh>
still it seemed successful.
<viric>
I would not buy anything without JTAG
<viric>
(apart from the nanonote :)
<roh>
basically ever dsl router for 20E has jtag and serial
<viric>
and the TPs are there, I think :)
<rjeffries>
roh I set up NSLU as a small ftp server for my gransdson who uses Macs he has used it withoout a single touch for maybe 2 years\
<viric>
roh: jtag working with openocd?
<roh>
viric: depends on the arch and your free time
<rjeffries>
roh that lottle server waS nslu FWIW
<roh>
viric: but i find the usual 14 and 20 pin headers in >80% of the devices i check
<roh>
rjeffries: for me the nlsu was never interresting. much too low-end hw.
<roh>
maybe thats because i got shiploads of embedded crap around. but i dont like stuff which is 'that slow'
<viric>
it's nice if the sellers provide all needed for the game
<rjeffries>
i recently found out that Dockstar also can talk the SPI (??) that the microSD aka 8:10 spits out
<roh>
cant even watch a hd movie via network from a disk connected to the nlsu.
<rjeffries>
roh it is all about user case
<roh>
sure. it can be used for lots of other stuff. but that can be also other devices. it wasnt better than a regular dsl router with usbports from a performance pov.
<rjeffries>
who in god's name would chose to use NSLU ti view HD movie? that makes zero sense
<rjeffries>
so I also have powerful computers. they cost more that $20 to $100 USD and do not draw only 5 watts\
<roh>
rjeffries: loads of people. who share files in a flat/house? everybody has notebooks.
<rjeffries>
but why would they give that task to a slooooooooooooow NSLU?
<rjeffries>
Dockstar is 1.2Ghz processor for example
<roh>
from my perspective: if you want a nas. use something serious, not arm or mips with no caches or ram.
<rjeffries>
oh well, why are we discussing this. too funny
<viric>
a friend pointed at this:http://www.dealextreme.com/p/standalone-bittorrent-bt-client-usb-nas-ftp-samba-printer-upnp-sharing-network-lan-server-26320
<roh>
viaC7, ulv-pentiumM .. such stuff. serious performance. ethernet and hdds are slow enough as it is.
<roh>
the hdds will eat more power than you cpu can anyhow.
<roh>
rjeffries: slowness isnt a question of clocking alone. its also about ram bandwith, cachsizes etc.
<roh>
also a lot of speed is lost with 'usb and its latencies' and 'samba and similar bad protocols' ;)
<rjeffries>
ok roh got it
<rjeffries>
wpwrak I am moving along in terms of getting quoutes for UBB
<rjeffries>
a PCB aware engineer will review package this weekend
<rjeffries>
I'll then modify or whatever
<rjeffries>
then I have a serious PCBFAB dude (a friend f a friend) and he will then quoyte
<rjeffries>
tuxbrain when will your UBB production run be complete? becasue then I can go and do something else
<Jay7>
hm..
<Jay7>
viric: NS-K330 is looking very interesting
<viric>
Jay7: what is that? the link I sent?
<Jay7>
viric: yes
<viric>
ok. I don't own it, so I know little.
<viric>
    CPU clock at 250MHz, CPU: FA526id(wb)(ARMv4) Memory: 32MB = 32MB total PCI clock at 33M
<wpwrak>
rjeffries: (review on weekend) yes, i saw the mail
<wpwrak>
rjeffries: i still think you could just have sent the stuff directly to a pcb house and just see of they can make heads and tails of it, but okay, your approach is safer :)
<wpwrak>
rjeffries: tuxbrain is probably busy cleaning out the backyard where they'll airdrop the containers with ubbs ;-)
<wpwrak>
ah, mubarak has left. completely overlooked it amidst the hacking and slashing at nokia :)
<roh>
viric: quite low end
<roh>
viric: most recent hw has about 300-500mhz. even in simple cheap routers. 250mhz sounds like 5 years old?
<kristianpaul>
Textmode: Had you founda good use of lua in the nanonote?
<Textmode>
yeah, got some of my projects running, though I hit a snag in that they were written with a larger text-console in mind. lots and lots of scrolling...
<kristianpaul>
xiangfu: I want try the latest owrt image, as i dont know why my current image got compile with no gmenu2x and related, i think i did follow wiki instruccions.. but anyway..
<kristianpaul>
I need my gmu !!!
<kristianpaul>
;-)
<xiangfu>
kristianpaul: the last config.full_system in git repo have one option "CONFIG_ALL=y", this option tell openwrt compile all packages in feeds, this need ~30 hours in buildhost,
<xiangfu>
I am updating the wiki tell people, remove this option before run yes "" | make oldconfig
<kristianpaul>
well my setup dint took that long in compile..
<kristianpaul>
xiangfu: when is goint to be next image release?
<xiangfu>
kristianpaul: now only three packges have problem when build, plplot-octave plplot-tcl and climm,
<xiangfu>
so I would like wait plplot-octave plolot-tcl this weekend,
<xiangfu>
so maybe next Monday,
<kristianpaul>
good
<xiangfu>
my origin plan is the end of January, but there are some big update in openwrt backfire. so we rebase our source code on last backfire. and delay the release date