#qi-hardware IRC log for Tuesday, 2011-07-05

wpwrakhmm, didn't really pay attention to < vs. > 2000:00
wpwrak(talk link) yeah, seems to take a while to do the censorship ;-)00:00
wpwrak(new things learned) hmm, one thing. do you know the "GPLv2 death penalty" ?00:01
wolfspraulnever heard about that00:01
kristianpaulsort of ritual from fsf guys? :)00:02
wpwrakclose ;-)00:02
wpwrakit's basically that, if you violate the GPL, you lose the right to use the code under that license, and that you have to return to compliance AND get re-licensed by each author/rights holder individually to be allowed to use the code under the license again00:04
wpwrakGPLv3 relaxes that to simply having to return to compliance00:05
wpwrakit's an interesting technicality00:05
wolfspraulsounds good that they removed that00:06
wpwrakyeah. it's quite nasty :)00:06
wolfspraulit may also affect enforcability00:06
wpwraknot that i'd really expect someone to try to enforce it, but ... 00:06
wolfspraulyou shouldn't try to link too many things together, like punishment chains etc. leave that to the legal system case by case, people can make their arguments anyway00:07
wolfspraulotherwise you risk that your entire thing becomes unenforcable because of some higher legal concept like practicality or what not00:08
wolfspraulso a simple "back to compliance" is much better00:08
wolfspraulwhat this means can then be determined in each case00:08
wpwrakwell, there's always the clause that if part is not enforceable, the rest still prevails00:08
wolfspraulthat's just my opinion of course00:08
wpwrakbut yes, "back to compliance" makes a lot more sense. having to find all the rights holders would be quite impossible in many cases00:09
wpwrakperhaps they put it in such that you can't hurriedly return back into compliance once a lot of money has been spent on legal action. i wonder how they solved that.00:11
wolfspraulwhat other projects or talks at fisl did you find most interesting?00:12
wpwrakof course, some of the violators are stubborn enough that they see things through to the bit end no matter how many chances for backing off they are given :)00:12
wpwraki didn't go to many talks. i listened to the one on tor (onion routing for privacy). it's interesting how crazy the powers that be get. and how the cat and mouse game between tor and the censors evolves.00:14
wpwrakbut .. not directly applicable to our stuff. well, except for you personally perhaps :)00:14
wpwraki also had someone with RF experience have a quick look at my atben/atusb boards, and nothing obviously fishy turned up00:15
wpwraki found a few bugs when trying to show things to people :)00:16
wolfspraulChina is big on tor00:17
wolfspraulthe powers of darkness00:17
wolfspraulI would stay away from it only because of that, but it's good that they bind some resources of evil00:17
wpwrakah, and my ubb-vga is more robust than i thought. after days of carrying it around without any protection, it still worked. and produced very nice colors on two occasion - nicer than that i had seen on my test devices00:17
wpwrak(tor) yeah, it only works if you can hide in the crowd00:18
wolfspraulwe have to use more steganography, randomization, etc.00:19
wolfspraulthere are so many attack vectors, like packet delays and what not00:19
wolfsprauloh well :-)00:19
wpwrakthey now also have tor briges that don't show up in the list of nodes00:19
wpwrakand they have tor mail services, so that you can get it via, say, google mail. and don't need to connect to tor itself00:20
wpwraknow sure how much that matters in practice, though. after all, you still need to find the explanation of all this :)00:21
wpwrakmay be easier to just hand out uSD cards. the perfect medium for covert operations :)00:21
wpwrakah, M1 doesn't include a mouse, right ? can it actually be used without one ?00:22
wpwrakjon talked with mozilla about possible common interests. he's thinking of trying to convince them that they need a hardware platform they can control.00:24
wpwrakwe also talked to some google folks. alas, not the ones with their hand on the money. jon will follow up on that in SFO.00:25
kristianpauloh, tor mail, thats new for me00:27
wpwrakreaction to the M1 demos (in the talk) was very good. better than i expected - they really captivated the audience. unfortunately, with the update problems, we didn't have time to show it off to a larger audience in the exposition area.00:27
wpwrakwe didn't do the L19 rework. didn't want to tempt fate ;-) the camera didn't cause problems00:28
wolfspraulwe do not include a mouse00:29
wolfspraulfor two reasons: 1) I have never seen a mouse that would not work (but with keyboards there are many that won't work, such as most/all from Apple)00:29
wolfspraul2) the GUI either is already 100% controllable from keyboard, or almost there00:29
wpwrak2) is good. 1) alone would still mean that your out of the box experience may be limited.00:30
wolfspraulmy biggest worry for including a mouse is that we have so many accessories already, it becomes like a bag of assorted electronics00:30
wolfspraulit's not so much the price, but also the size and weight of a mouse that turns me off00:31
wpwrak(keyboard) also jon had some no name keyboard he said didn't work00:31
wolfsprauloh sure00:31
wolfspraulit took me 4 keyboards until I finally found one that works00:31
wolfspraulthat's when I decided that we had to include one :-)00:31
wpwrak(too many accessories) yeah, also not exactly fun for the sourcing side00:31
wolfspraullots of work, yes, but we are mostly done with it00:32
wolfspraulso the mouse, yes, it's still in the air00:32
wpwrak(mice that work) you can add my little travel mouse from kensington :)00:32
wolfspraulwhen comparing silicone keyboards, we had one with an integrated mouse00:32
wolfspraulbut, as always, there were problems00:32
wolfspraulfirst the current software would not support mouse + keyboard as 2 endpoints on the same usb port00:32
rohheh. integrated mouse means hub doesnt it?00:33
wpwrakdo we know at what layer are the problems with the keyboards ?00:33
rohwpwrak: mm1 doesnt support stuff behind usb hubs afaik00:33
wpwrakroh: of course :)00:33
wolfsprauland second, even more important, the mouse on that silicone keyboard was very imprecise00:33
wolfspraulit was just nearly impossible to position it over a button00:33
wolfspraulit would always jump too far etc.00:33
wolfspraulso that would have meant even more software work00:33
wolfspraulroh: no, that one implements 2 endpoints00:34
wpwrakbetter not to ship with crap then00:34
wolfspraulor 2 'interfaces' something like that00:34
wolfsprauldefinitely no hub00:34
wolfspraulit would be relatively easy to support, but then we still would have precision problems00:34
wolfspraulyes correct00:34
rohweird. but yes.00:34
wolfspraulApple keyboards include a hub, always I believe00:35
wolfspraulthat's why none of them work on m1 right now00:35
wpwrak(no hub) hmm. need to check the spec on whether that's allowed. not that it wouldn't make perfect sense ...00:35
wolfsprauldefinitely allowed00:35
wolfspraulbut anyway, that's the mouse story so far00:36
wolfspraulI wish that silicone keyboard would have had a better precision mouse00:36
wpwraki think the entire USB trouble just has one solution: get Linux to work and move everything that need RT into some little RT bubble under linux00:36
wolfspraulthen I would have taken that (and added support for the 2 endpoints/interfaces, which looked like not to be that hard)00:36
wolfspraulso yeah, we need to make sure that keyboard only is intuitive, I haven't tested it yet00:37
wolfspraulthe precision problems may be compensated in software as well, I would think. but it's more work...00:37
kristianpaulwpwrak: RT bubble sounds nice :)00:38
wpwrakthe gui seems a bit overloaded. but again, that's comfortably a v2 problem to address00:38
wolfspraulit just needs to move more slowly, and then accelerate when pressed longer00:38
kristianpauland is not usb, also memcard, i confess i miss linux a bit now :)00:39
wpwrakkristianpaul: i think even lekernel would now consider linux to be the lesser evil than having to port a gazillion of usb drivers :)00:39
wolfspraullekernel will think that if someone other than him ports/improves Linux, or adds a MMU, or ..., that'd be awesome00:41
wpwrakwolfspraul: the next step would be to integrate an interface into the M1. e.g., a touch screen lcd.00:41
kristianpaulon top ! yes00:41
wpwrakwolfspraul: (port linux) yeah. i wonder what's taking the guys so long. it can't be *that* hard.00:42
wolfspraulLinux is booting00:42
wolfspraulnommu version00:42
wolfspraulbut I don't think all the different peripherals are supported, and it seems work on the port has stopped in March00:42
rohyeah. nommu is quite a big turnoff for many people00:42
wpwraknommu is suckish00:43
kristianpaulwpwrak: last i heard there were some minor issues with interrupts, you can ask mwalle i think00:43
wpwrakbut why not implement a really simple mmu ? one-way TLB. that can't be so hard.00:43
wpwrakkristianpaul: ah yes, he didn't like the hierarchy. but i think he and lekernel now agree on an approach00:44
rohwpwrak: btw.. i got some people around working on their own small designs.. currently the i.mx233 looks fancy to us.. because its available in hand-solderable packages and still not be totally outdated when it comes to performance00:44
wpwrakroh: sounds nice00:45
rohalso freescale seems nice when it comes to datasheets and the linux ports are maintained by pengutronix :)00:46
wolfspraulsounds like the right approach to think about practical execution first00:46
rohonly downside: no eth mac included on the imx233 ... the imx28 or so has one.. but its bga again00:46
wolfspraulwpwrak: there was serious planning about the mmu, see http://lists.milkymist.org/pipermail/devel-milkymist.org/2011-April/001418.html00:47
wolfsprauland http://lists.milkymist.org/pipermail/devel-milkymist.org/2011-April/001472.html00:48
wolfspraulbut someone has to take it from planning to implementation now00:48
wolfspraulSebastien will surely merge back any good code that comes his way :-)00:48
wolfspraulfor the video synthesizer itself, the mmu is not critical imho00:48
wpwrak4 privilege levels ? who needs that ? :)00:49
wpwrakis anyone porting VAX/VMS ? ;-))00:49
wpwrak(mmu+synth) not for the video synthesis. but if you want to make the synthesizer talk to more things, you need the cornucopia of drivers from linux. and linux is so much happier if it has an MMU :)00:50
wpwrak(mmu arch) quite fancy already. for getting started, you could make something much simpler :)00:53
wolfspraulok but I want to be as real as possible00:58
wolfspraulI am not working on mmu, i have no experience in Verilog at all, no time to do it. I have no resources to fund someone doing it. For Sebastien it's not high on his priority list at all.00:59
wolfspraulafaik nobody else (for example from the people discussing the design on the list) is currently working on it either00:59
wolfspraulso keep your expectations about a mmu 'arriving from somewhere' very low00:59
wpwrakyeah, yeah, the old "if you want it get done you have to do it yourself" ;-)01:01
wolfspraulit's important that we communicate that rather than speculating whether it would be nice, and leaving out that it is beyond the resources of the project (and priority list) to work on one now01:01
wolfspraulI just think with the most clear assessment you eventually succeed best. rather than implying all these wonderful things will, or may, happen.01:01
wolfspraulm1 has a long list of desirable todos01:01
wolfspraulmmu is one on that long list, somewhere01:01
wolfspraulso I communicate that clearly: m1 has no mmu, don't expect one, we don't need it for the video synthesizer, we can drive the video synthesizer to much higher grounds without one, etc.01:02
wpwrakmmu is a de facto prerequisite for linux; linux is a de facto prerequisite for proper driver support. people will quickly get frustrated with USB host having all sort of weird restrictions.01:03
wolfspraulhe :-)01:04
wpwrakso the sooner this gets solved, the better01:04
wolfspraulyou are a good candidate for communication effectiveness tests01:04
wpwraki mean, just think of the horror scenario of someone losing the keyboard ;-)01:04
wolfspraulI never endulge in wishful thinking01:04
wolfspraulI do not see those things on the practical radar01:05
wolfsprauland they are not needed now, imo. at least not for me selling a video synthesizer.01:05
wpwrak(comunication) i hear your "i don't want to think of the problem" ;-)01:05
wolfspraulnot at all01:05
wolfspraulas you can see I am very clear about the status of it, that's why I try to get it across01:05
wolfspraulyou are not the first one to ask :-)01:05
rohhen and egg problem somehow01:06
wolfspraul"community needs to grow" is another way to describe it01:06
wolfspraulbut how can it grow? needs to be attractive?01:06
wolfspraulneeds Linux to be attractive?01:06
wolfspraulwho brings Linux there?01:06
wolfsprauletc. etc.01:06
wolfspraulhen and egg, indeed01:06
wolfspraulmeanwhile I focus on the video synthesizer :-)01:07
rohthere is a clear target market who has no demand for 'vj-ing' equipment with an interrest to use linux on a fully open arch.. the question is how big.01:07
wolfspraulroh: help us attract more devs to the platform01:07
rohto be fair.. i dont have any usecase i'd need vj-ing equipment for. i am interrested in open hw.01:07
wolfspraulwe are all 'in favor' of that01:07
DocScrutinizerno mmu was when I stopped on *nix porting to Amiga100001:08
rohfor real vjs (the ones i meet) we need a non-gui user interface anyhow.01:08
wolfspraulroh: the approach with m1 is to make one use case work very well first, in this case video synthesizer. then break through the hen and egg problem like that.01:08
rohthey all had notebooks with pd or such with em. and used xga or up resolutions. and they asked how to not have a menu popping up ;)01:09
rohi guess they would have liket to have the mm1 as kind of a 'eth controlled pd playout renderer' instead of keyboard, mouse and gui01:09
wolfspraulagreed, but some of that sounds like relatively minor software mods01:10
wolfspraulone by one01:10
wolfspraulplease remember the size of the engineering team that brought it to where it is01:10
rohwolfspraul: i hear you. i am only thinking that there is a balance between trying to reach a self-set spec in reasonable time and widening the target groupt to not end up like openmoko in the end (where we only kept the die-hard-no-binary fraction)01:11
rohsometimes its just 'making stuff work' to make customers happy01:12
rohregardless of politics01:12
wpwrakit's pretty huge. an EE, a verilog designer, a part-time fpga tool designer, a GUI infrastructure designer, a GUI designer, a driver developer, an AVR chip designer, a AVR USB firmware author, ...01:12
wpwrakoddly enough, they're all called "Sebastien" ;-)01:12
wpwrakso you just need to find more sebastiens :)01:12
wolfspraulthere's more people that made significant contributions, and I include soft contributions and those on the edges as well01:13
wolfspraulbut still, if we get lost on the long todo list now, that'd be a mistake. at least I won't do it.01:13
wolfspraulI focus on video synthesizer, and need to reach our sales targets there.01:14
wolfspraulwhich is 200 for 2011, and I'm 31 into it, 169 missing01:14
rohwolfspraul: the question is: is that market that big?01:14
rohwith the given current spec01:14
wolfspraulrunning a little late with rc3, which is hopfeully another 75, hopefully sells in 2-3 months (?), and hopefully leaves enough room for an rc4 to reach the target01:14
wpwrakwhat would be the changes from rc3 to rc4 ?01:15
wolfspraulroh: the next challenge is a marketing challenge01:15
wolfsprauljust look how wpwrak described audience reaction at fisl01:15
wolfspraulnow... the question is: can we effectively get the capabilities of m1 across to a large audience? in the next few months?01:16
wolfspraulcan we quickly and effectively sell to people around the whole world?01:16
wolfspraulbig question marks01:16
wolfspraulif an established company like Roland, M-Audio, etc. would put their marketing muscle behind it, I am 100% sure they could sell 5000 or so in the first year01:16
wolfspraulwhy? because they throw the same technology into an existing marketing machine01:17
rohwolfspraul: thats the part where i am unsure. is the market with the given spec really big enough? i only meet people who have higher demands.01:17
wolfspraulso it will show up in specialized stores all over the world01:17
wolfspraulno way. the people I meet never once ask about 'specs'01:17
wpwrakyou basically need to find a group that loves it enough to forgive its quirks. many of its limitations will be regarded as bugs. if the users love the concept enough, they will forgive virtually everything. so it's important to make them fall in love :)01:18
wolfspraulthis product does not need to wipe out everything else in that space, and it won't01:18
rohmy guess is that roland or so wouldnt have used that versatile hw but something much more limited to one usecase, still speced higher in some domains (dvi/hdmi ports, hd ready etc)01:18
wolfspraulthe difference whether we sell 50 in the next 6 months or 2000 depends entirely on marketing effectiveness01:18
wolfspraulwhich is a big problem, but alright, we have to meet that challenge01:19
wpwrakroh: you always get a lot of people who want more. let them carry that hecta-core mainframe with two suitcases full of peripherals around ;-)01:19
wolfsprauljust count the number of nightclubs in China alone that could put one of these boxes into their place01:19
wolfspraulit must be in the tens of thousands :-)01:19
rohfrom my pov these matrox 'make 3 out of one vga/dvi port' devices are a bigger contender than roland or so01:19
wpwrakroh: these people will also be extremely vocal, because they're desperate :)01:19
wolfspraulI think that card costs more than the entire m1, including camera01:20
wolfspraulno no01:20
wolfspraulthis is not how sales function01:20
rohwpwrak: mainframe? huh? nah. they all just carry that one macbook.01:20
rohwpwrak: also they have multiple beamers with them anyhow, to everything smaller than a notebook is fine.01:21
wolfspraulgo to any middle-class hotel in the world, check what projector they have in the conference room. is it the latest and greatest? no01:21
wolfspraulwhy is it there? because they chain bought it for all hotels, or some company setup the stuff01:21
wpwrakroh: in most clubs, they'll just what is already installed01:21
wpwraks/just/just use/01:21
wolfspraulwhy did that company choose it? because they know it, because it's cheap and gives them a big margin? many reasons01:21
rohwolfspraul: low-end beamers are XGA now. high end is fullhd.01:21
wolfspraulif someone doesn't like m1 that's fine01:22
wolfspraulyou think too black & white01:22
wolfspraulwe need to communicate what it can do01:22
wolfsprauland it needs to work flawlessly and out of the box, dumbed down01:22
wolfsprauland then some people will buy, and hopefully be happy01:22
wolfspraulso... my sales target remains at 200 for 2011 :-)01:24
wolfspraulrc3 is late, unfortunately01:24
wolfspraulwe loose 1-2 valuable months of 2011 there01:25
wolfspraulroh: one big issue on the marketing side is also that with m1, there are several new things at once01:25
wolfspraulnew technology, new brand, maybe new category, new business model01:25
wolfspraulno sane company would risk all those things together, but we have no other choice right now01:26
wolfspraulif it would be an existing brand, and new technology, that would be easier01:26
wolfspraulnew brand, old technology - also easier01:26
wolfspraulnew business model, old technlogy - all fine01:26
wolfspraulbut all of those together - NOT GOOD01:26
wolfspraulit creates a huge marketing challenge01:26
wolfspraulbrand = trust01:27
wolfspraulmany people will trust a company to sell them something good if they remember that brand positively from the past01:27
wolfspraulwe cannot make use of that01:28
wolfspraulthey have to trust a new brand, new technology, new category, all at once01:28
wolfspraulalso my ability to sell into NanoNote brand is limited, because the product is 5 times more expensive01:29
wpwrakwolfspraul: ah, by the way: have you registered any trademarks in the qi-hw universe ? (qi-hw, nanonote, M1, etc.)01:29
wolfspraulso even though I have a sizable number of customers who trust me, the practical outreach is limited01:29
wolfspraulthe would buy something for 99 USD from me, many01:29
wolfspraulnot even look at the specs for 1 minute01:29
wolfspraulbut 499 USD? many will say "too expensive"01:29
wpwraki don't think it's so much a question of trust. a few nice reviews will settle that.01:30
wolfspraulit's a big wall to climb01:30
wolfspraulI hope we can find some marketing parent corp to help us with global marketing of m101:30
wpwraki mean the trust issue. they may still dislike the price, but not because they don't trust you01:30
rohwolfspraul: sure.01:31
wolfspraulthe price is great, everything is included in the box like camera01:31
roh500us$ is nicely above the 'expensive geek-toy' class of 299-399E01:31
wolfspraulI agree prices are coming down in electronics, fast01:31
wolfspraulmany people don't even closely look at anything above 150 EUR anymore01:32
wpwrakroh: OQO sold quite a few units at a much higher price point. and with a rather interesting list of flaws.01:32
wolfspraulbut the downside is that there are only a handful of products left, all the same (big surprise)01:32
wpwrak(oqo) of course, they still didn't make it ...01:32
wolfspraulwpwrak: that's long ago01:32
rohwpwrak: not around here. not seen anybody with one01:32
wolfspraulthose times are gone01:32
wolfspraulas always there is a downside01:32
wolfspraulproducts become extremely boring01:32
wolfspraulbecause the companies need huge volume, cannot take risks01:33
wolfspraulthe first thing many people admire about the m1 is the case01:33
wolfspraulwho has seen a computer/box like this EVER?01:33
wolfsprauloh sure. simply because it is different01:33
wolfspraulshow me something like that? go to any store? everything looks the same there, and getting even worse.01:33
wolfspraulthat's the downside of prices going down (and in the background: volumes going up)01:34
rohgazillions of devices on a fair.. the 'messe-version' ;)01:34
wpwrakat FISL, the reaction to the price was that people didn't think it too expensive for what it can do. they may still consider it too expensive in relation to their individual spending power, of course.01:34
wolfspraulroh: what do you mean? interesting stuff at fairs?01:35
wpwrakalso, they may change their opinion once all the taxes are added, which they may or may not be aware of01:35
rohwolfspraul: no.. the frequency of custom acryllic showcase-case on fairs01:35
wolfsprauldo they make it to the store?01:35
Action: kristianpaul argghh taxes01:35
rohmostly (atleast around here)01:35
wpwrakkristianpaul: ;-)01:36
rohto be fair.. most people dont buy hardware in stores anymore. loads of sales shifted to mailorder01:36
wolfspraulI do not believe we can escape the price spiral down, and low prices are good01:36
wolfspraulbut I look at this as a function of volume and price, and then I think we do pretty good. Now my focus is sales, then bring price down.01:36
rohwhy go somewhere to order it then because its not on stock in that version you want?... and stockpiling isnt possible for shops mostly because of the small margin on sales of such stuff.01:37
rohon computer articles you usually have <3% margin for sales to end-customers compared to the recommended sales price. so either you are more expensive or need to push through really many boxes01:37
wolfspraulsoon we will have ad-supported hardware that is free01:38
wolfspraulAmazon is making some smart experiments right now, which I believe will succeed01:38
wolfspraulI'm fairly relaxed about all this01:38
wpwrakwolfspraul: or the ingenic computer in playboy and some other magazine ;-)01:39
kristianpaulyeah :-)01:39
wpwrakroh: recommended sales price would still include some margin ...01:40
wolfspraulm1 has some fairly unique connectors like dmx, rca video-in, midi01:40
rohwpwrak: some. which is really small in mass market CE01:41
wolfspraulthat may also help drive some sales, if marketing is effective in message and reach01:41
wpwrakwolfspraul: can it actually *do* anything with DMX ? :)01:41
rohwpwrak: dmx is widely used for lighing control.01:41
wolfspraulI haven't seen it, but sebastien has some dmx patches, I would think they do something01:41
rohbasically its 512channel 8bit value control.01:41
kristianpauli think it reads dmx01:41
wpwrakroh: yes. but does M1 do something with it ?01:41
DocScrutinizerad-supported free hw. Hell they must earn their money with some other service then01:42
kristianpaulactually code used for osc was derivated from dmx, i think xiangfu did?01:42
rohwpwrak: i think you can control is as well as control lamps from the mm from within patches01:42
wpwrak(midi) another issue is that it can't do USB-midi. again, drivers.01:42
DocScrutinizergiven the rationale ads are to make people *buy* stuff01:42
wpwrakDocScrutinizer: let's try perfume ;-)01:43
rohwpwrak: usb-midi? huh? ofcourse you can connect a usb-midi if to the midi of the mm101:43
wolfspraulI think it makes no sense to compare our hardware that is made and sold in the hundreds or thousands to hardware that is made and sold, or at least anticipated to sell, in the millions or tens of millions01:43
wolfspraulI am very happy that I'm not an investor in that other side, it's at least as risky as our stuff if you think about it01:43
wolfspraulno risk no fun, no risk no profit01:43
kristianpaulDocScrutinizer: may be is just marketing budget anyway.. like when you gave free cloth accesories with brands etc..01:43
wolfspraulso I need to sell 200 m1 boxes this year, end of story.01:43
wolfspraulSamsung needs to sell 5 million tables, or more, who knows. that's their problem, and the gamble they accepted when they did some new piece of hardware.01:44
wpwrakroh: i mean a usb-midi device to usb01:45
kristianpaulyes, thats a very common stuff right now01:45
rohwpwrak: luckily we dont have usb client on the mm;)01:45
wpwrakroh: there's now a ton of small stuff that only talks usb. but it's still midi on top.01:45
wpwrakroh: no, usb host is fine01:45
kristianpaulactuallu i tried to source a midi keyboard for m1, and just founded cheap midi-usb stuff01:46
rohwpwrak: ah.. now i get what you mean. like connecting a usb-midikeyboard to the mm101:46
wolfspraulI know what's next :-)01:46
wpwraklinux ? :)01:46
rohkristianpaul: the same models are avail with midi too. just some euros more01:46
rohwpwrak: well.. one could use the driver layer... or some 'similar' api to make porting much easier01:47
kristianpaulroh: did  i said i need it _cheap_ ? ;)01:47
rohkristianpaul: i can guess so.01:47
rohit was like 50E usb-midi only, 60 or 65 with midi ports01:47
wpwrakroh: sure, just do that ;-)01:47
wpwrakroh: if they import the "real MIDI" version at all ...01:48
wpwrakroh: not every country is a high-value high-volume market like germany01:48
rohnaaah.. its not that nice here01:48
wpwraki knew you'd say that :)01:49
rohits getting better as the us goes under ;)01:49
wolfspraulhey btw, one small thing01:49
rohasia somehow realises it shoudnt sell the nice stuff there but here.01:49
wpwrakyeah. no us market, need to find others :)01:49
wolfspraulif m1 is successful, there is chance we can build up a reasonable strong free pool/market of visual works, that are freely shared among users01:49
rohwolfspraul: like a 'patch-bay' ?01:50
wolfspraulwe are working on that slowly now with the update automatically downloading new patches01:50
rohsorry for the patch vs piratebay pun ;)01:50
wolfspraulthat's just the first step but the direction is clear01:50
wolfspraulif uploading/downloading (syncing) is easy, and the quality of that pool reaches a certain critical mass, then that alone may drive sales quite well01:51
wolfspraulattribution of creators needs to work well, maybe even rewarding them (donations, flattr, whatever)01:51
wolfspraulwe'll see01:51
wolfspraulso let's not forget that m1 is not just a piece of plastics and physical goods, but also connects to a 100% free world of bits and bytes01:52
DocScrutinizeranother appstore?01:52
wolfspraulI said 100% free01:52
wpwrak(patchstore) yup. that's the new way :)01:52
wolfspraulthe strengh of the visual works that are either included when sold, or freely and easily downloadable later, is quite important01:53
wolfspraulit creates value01:53
wolfspraulnot just higher vga resolutions create value (though of course they do too)01:53
wpwrakyes. create a user community platform. not just some places for developers to hang out with like-minded freaks.01:54
DocScrutinizerfor risk see Nokia's N950 (formerly N9) desaster01:54
wpwrakyou also want the ability to comment/discuss, have a forum where you can link to/from the patch archive.01:55
wpwrakDocScrutinizer: Nokia is the new Openmoko. don't do what they did and you'll be fine ;-)01:55
wolfspraulit is my firm belief that 90%, even more, of tech buyers use only a fraction of the features of the products they buy01:56
wolfspraulfor the sake of this argument we can say they use 1% of them01:56
wolfspraulbecause stuff does not work, plain and simple. it's too difficult, they have no time. they press a few buttons, it doesn't do what they want, they move on.01:57
wpwrakthere are also more features than they individually need01:57
wolfspraulthe approach is exactly the opposite approach from ours (mentally)01:57
wolfspraulwe look at the specs first, then we think (with our understanding of software), what is theoretically possible with those specs01:57
wolfspraul90% or more of people do it exacty the other way round01:58
DocScrutinizerplus you usually pick a lot of functionality for "just in case"01:58
wolfspraulthey press a few buttons, and see what happens01:58
wolfspraulthen that's what they will use of this product01:58
DocScrutinizerlike aircrack & monitormode on N900 ;-)01:58
wolfspraulmany companies in tech don't care about this, and slowly start to believe that all the great features they are adding is what sells their products01:58
wolfspraulthen they ignore the stats that are saying that 90% of 'smartphone' buyers only make phone calls and sms01:59
wolfspraulwell then :-)01:59
kristianpaulwpwrak: so, we linux support, just in case we lost the factory keyboard that came with MM1 :-)01:59
wolfspraulso the main thing for something like m1 is that what it can do is super easy to reach01:59
wolfspraulgive the m1 box to a newbie in a test room01:59
wolfspraulrecord with a camera what happens01:59
wolfspraulif we improve those kinds of things, we can sell tons01:59
wolfspraulI'm sure01:59
wolfspraulnot if we dwell over feature lists and think what more could be added02:00
wolfspraulat least that's what I will do, and I took the risk to manufacture them, he he :-)02:00
Action: DocScrutinizer has strange visions of green cards with lao tse sayings on them02:00
wolfspraulI like that card, but it's unrelated to the point I just made.02:01
wolfspraulwhich is that it's far more important that some basic features are very easy to use, than adding more features on top.02:01
DocScrutinizerit's loosely related02:01
wolfspraulbut I need to proove it in sales, not in chat02:01
wpwrakkristianpaul: (linux support) well, it's just one of those scenarios that can easily happen if you have too narrow a known to be good path. another is the power supply that has to be exactly 5.000 V (exaggerating :)02:05
kristianpaulwpwrak: (power suply) i think thats fixed for rc3, but is a fair point02:07
wpwrakwolfspraul: you basically have to match reasonable expectations. add too many unpredictable restrictions and people will get frustrated. add too many predictable restrictions and they'll think the product is weak. so you need to put it all in a good story.02:07
DocScrutinizerwpwrak: (the new OM) the part that really puzzles me is that literally everybody can see clearly how Elop is a mole sent by M$ to kill the company, just they seem to ask for it and like it02:16
wpwrakDocScrutinizer: makes you wonder what kind of deal he struck with the board of directors, eh ?02:17
DocScrutinizerI dunno if at OM there was an parallel after I left :-)02:17
wolfspraulI don't believe the mole story. I think it's just exceedingly hard to turn the company around now.02:22
DocScrutinizerprobably OM could have sold a gta03 and gta04 and 05, if only we wouldn't have started from square one basically 3 times for one product. a gta03 with s3c2442 would have sold another 5000 units02:22
wpwrakDocScrutinizer: i had proposed to make a cleaned up gta02 ... but of course, nobody listens to me :)02:23
wolfspraulthe Microsoft deal is totally US centric and ignores 'traditions', yeah well. don't fight with mother nature, is the US expression.02:23
wolfspraulElop decided to fight with mother nature, and I believe it will not work.02:24
rohwho cares. one big contender less.02:24
wolfspraulom was unrescuable since day 102:24
DocScrutinizerand probably Nokia could have sold millions of N9 when they just had rolled out their N950 design last year as they could've done02:24
wolfspraulthe mother bug...02:24
rohi guess android has >50% market share on smartphones already when counting all the chinese knock-off-iphones02:24
DocScrutinizerwpwrak: cleaned up gta02 hadn't sold >100 devices02:25
wpwrakDocScrutinizer: correct, because it was never made ;-)02:25
wolfsprauldo we think that Nokia might have had a chance with an all-Meego strategy?02:25
wolfspraulit would have been risky, but not less than what they do now02:26
wolfspraulit's so obvious to me that they will be following Palm's lead now02:26
DocScrutinizerI'm sure they would've had a chance with all-maemo strategy02:26
DocScrutinizerbtw they are in fact back to maemo now, even while they call it meego ;-)02:26
wolfspraulok, I didn't want to distinguish between the two, I just meant a 'homegrown Linux/free software' strategy02:26
wolfspraulNokia will proudly go where Palm was before, oh well02:27
wolfspraulmaybe HP can buy the rest of it in a few years :-)02:27
DocScrutinizerthe N9 might well fly, just Elop killed it in advance by telling it won't be followed even if it was a success02:28
DocScrutinizeryou *don't do that* on a new product02:28
DocScrutinizerthat's why makes the mule state obvious02:28
DocScrutinizernothing about turning around the company02:29
wolfspraulwhat can Nokia do better now that HTC cannot do already?02:32
DocScrutinizerthey had a semi decent strategy for trans-platform development with Qt - Elop says Qt won't ever be supported on WP02:32
wolfsprauland Nokia carries expensive legacy with it, staff, pensions, facilities, support obligations, etc.02:33
DocScrutinizerbasically Nokia nuked their smartphone segment for good now, with a triple stroke02:34
wolfsprauldo they have a content strategy?02:34
DocScrutinizerno more02:35
wolfspraulwell. the Nokia boots are still selling in Finland I read. so there is a fallback base :-)02:35
wpwrakmaybe they can also make guided tours through the ruins :)02:37
DocScrutinizerlike in chernobyl02:37
DocScrutinizerhell, they're selling one million symbian featurephones a day, and aren't able to sustain a proper smartphone strategy02:39
DocScrutinizerthey even have a product that's well on par with iPhone UX wise, and still opt for nuking all that for going elopocalypse-now02:41
kristianpaula bit OT here but, maddog's new aventures http://www.projectcaua.org/comment/29#comment-2902:41
Action: roh still doesnt get why you care about other companies suicides02:42
wpwrakkristianpaul: he's currently trying to get the government to make it easier to start small companies. he'll need thousands of them ;-)02:42
wpwrakroh: indeed. nokia are irrelevant.02:43
wpwrakroh: well, mostly. losing the batteries will be a pain. they were an easy reference for all sorts of other devices.02:43
rohwpwrak: i dont see china stopping building them02:44
wolfspraulroh: trying to understand which pieces of tech are reusable for a future copyleft hw phone02:44
wolfspraulthat's why I follow Nokia02:44
wpwrakroh: once the nokia phones disappear, so will the nokia batteries.02:44
rohwolfspraul: sure. concepts stay. but details change02:44
rohe.g. i'd not use any of the soc we used then etc.02:45
rohwpwrak: there are gazillions of devices using these batteries. lets care when we must (in 10 years or so)02:45
rohwe will see movement in that area anyhow. li-po and li-ion will get replaced gradually as lifepo gets cheaper and more dense (more secure chemistry)02:46
rohalso nokia sells much more non-smart-phones as smart ones.02:46
rohmy guess is they also earn more on 3g licensing than on building smartphones themselves02:47
kristianpaulwpwrak: yes, i met the project like 3 years ago in a talk he gave for a local FLOSS event, i wonder that it still alive at least :)02:48
wpwrakkristianpaul: oh yes, he's still on it. at least he was last wednesday :)02:49
wolfspraulahh, good point02:49
kristianpaulwpwrak: first was 1GB wifi tech or something that he predict today will be avaliable, but seems not, so now ethernet.. well.. i want to see how all this will end02:49
wolfspraulwpwrak: did you catch up with maddog on all of those many wild free phone plans of the past?02:49
wpwrakwolfspraul: we held that service already some months ago, when he was visiting buenos aires for some other conference.02:50
DocScrutinizerroh: I care simply because I need a handheld device that works and meets my requirements. Nokia produced such devices for some time. Seems they stopped on that, and probably won't pick up this product line any time soon again02:53
DocScrutinizerI'm honestly considering android02:55
wolfspraulDocScrutinizer: Android now :-)02:56
Action: DocScrutinizer heads out for the bathroom02:57
wolfspraulI mentally skipped over all this, that's why I build this NanoNote and Milkymist stuff...02:57
wolfsprauland why I am excited about the ben-wpan project, or kristianpaul's gps baseband02:58
DocScrutinizerbeen there 6 years ago, when I realized it's a PITA to have a phone separate from the PDA where your contacts are stored02:58
wolfspraulbut of course, nothing that comes even close in practical use today to N900, N whatever, Android anything. I know that.02:58
wolfsprauloh of course, we need to integrate it02:59
wolfspraulbut as you noticed I didn't follow on the N900 path, and I don't regret02:59
wolfsprauland I won't follow on the Android path02:59
Action: roh is still on featurephone an real computers. for light traveling i have a eeepc. would switch to lenovo X120 if i had money now03:00
qi-bot[commit] Werner Almesberger: COPYING mentioned COPYING.GPLv2, COPYING.LGPLv21 which weren't included. Added them. (master) http://qi-hw.com/p/ben-wpan/298f08403:19
qi-bot[commit] Werner Almesberger: AUTHORS: werner@openmoko.org is now werner@almesberger.net (master) http://qi-hw.com/p/ben-wpan/e549ee403:19
qi-bot[commit] Werner Almesberger: tools/lib/atusb-common.c (atusb_open): give more detailed error message on EPERM (master) http://qi-hw.com/p/ben-wpan/161537503:19
rjeffriesAsus announced a THIN $200 Netbook that runs Meego. 06:43
stefan_schmidtwpwrak: ping07:07
wpwrakstefan_schmidt: pong09:15
wpwrakis it working ? :)09:15
stefan_schmidtwpwrak: kinda09:16
stefan_schmidtwpwrak: At least I'm getting the manid and chip id infos during probing now09:16
stefan_schmidtwpwrak: only read1 so far09:16
wpwrakcongratulations !09:16
stefan_schmidtwpwrak: thx09:16
stefan_schmidtwpwrak: well, mostly your patience to bring me back to the right path. :)09:17
stefan_schmidtwpwrak: I have a strange porblem with that though and I wonder if there is a bug in the firmware09:17
wpwraki told you you were almost there :)09:17
wpwrak(firmware bug) not impossible :)09:17
stefan_schmidtwpwrak: I have set the size to two bytes instead one in the urb and then I get the bytes back swapped09:18
stefan_schmidtwpwrak: had to change at86rf230 to make it work. This will surely break something else09:18
wpwrakhmm. read1 should only return one byte09:19
stefan_schmidtwpwrak: yeah, that's what I expect from the name as well09:20
stefan_schmidtwpwrak: maybe I just misinterpret something else09:20
stefan_schmidtmoment, going to push my last changes09:20
qi-bot[commit] Werner Almesberger: spi_atben: removed interrupt redirection (ben-wpan-stefan) http://qi-hw.com/p/qi-kernel/9e362c809:23
qi-bot[commit] Stefan Schmidt: Merge remote-tracking branch 'origin/ben-wpan' into ben-wpan-stefan (ben-wpan-stefan) http://qi-hw.com/p/qi-kernel/bd5e2bd09:23
qi-bot[commit] Stefan Schmidt: spi/atben: Remove sysfs files for at86rf230 regs (ben-wpan-stefan) http://qi-hw.com/p/qi-kernel/7539f4409:23
qi-bot[commit] Stefan Schmidt: spi/atben: Start with write (ben-wpan-stefan) http://qi-hw.com/p/qi-kernel/b5f9b7309:23
stefan_schmidtwpwrak: if you take a look at  atusb_read1() you will see the usb_fill_control_urb function09:24
stefan_schmidtwpwrak: I have req->wLength = cpu_to_le16(0x02); and also the size in the fill function to 0x0209:25
stefan_schmidtwpwrak: Also changing a *data = buf[1]; to *data = buf[0]; in the __at86rf230_read_subreg made it "work"09:26
stefan_schmidtwpwrak: there is something wrong on my side for sure as the 230 driver must stay as is09:27
wpwrakpulling ...09:27
Action: stefan_schmidt hopes its understandable what he writes. :)09:27
wpwrakstefan_schmidt: easy :) you need to receive into rx+1 :)09:30
wpwrakstefan_schmidt: read1 isn't a "real" SPI, so it doesn't return anything on the byte being sent09:31
stefan_schmidtwpwrak: ah, damn09:31
stefan_schmidtwpwrak: that makes sense (why does this insight only always came after you told me whats wrong?) :)09:32
Action: stefan_schmidt tests09:32
stefan_schmidtwpwrak: how was the conference btw? I read your talk was well attended.09:34
wpwrakthe conference was quite good. people liked what we showed them, particularly M109:36
stefan_schmidtwpwrak: (test) you are right :) rx+1, size to 1 and the hack in 230 can go09:36
stefan_schmidtwpwrak: And brazil gov is going to fund big open hardware research now? ;)09:36
wpwraksigh. i wish ...09:37
qi-bot[commit] Stefan Schmidt: spi/atusb: Use correct size and buffer location (ben-wpan-stefan) http://qi-hw.com/p/qi-kernel/133a8cb09:38
wpwrakrejon made contact with people from the mozilla foundation and we also pinged google. all this needs following up, probably in person. so rejon again :)09:38
stefan_schmidtwpwrak: the problem of the pioneers...09:38
stefan_schmidtwpwrak: yeah, sometimes I wonder which conf he actually misses. :) Met him at guadac back in 2008 and at FOSS.In 2009 09:39
Action: stefan_schmidt is up to his next victim: write209:41
wpwrakhe's everywhere :)09:41
stefan_schmidtwpwrak: he and his clones09:42
wpwrakthe multiverse interpretation of quantum physics is probably right. and people like rejon and lekernel can actually make use of their parallel selves :)09:43
stefan_schmidtwpwrak: cheater09:43
qi-bot[commit] Werner Almesberger: install/INSTALL-Ben: switch linux-zigbee tools source from (old) tarball to git (master) http://qi-hw.com/p/ben-wpan/17f7b6611:29
qi-bot[commit] Werner Almesberger: tools/lib/atben.c: check if the 8:10 slot is available and fail if it isn't (master) http://qi-hw.com/p/ben-wpan/f535a8011:29
qi-bot[commit] Werner Almesberger: TODO: removed EPERM, added usb_read_bulk vs. signal, toolchain, DFU on U1010 (master) http://qi-hw.com/p/ben-wpan/e610a9e11:29
viricwolfspraul: thank you for appearing nanonixos on the news, btw :) without any screenshot hehe12:30
viricxiangfu: offrss 1.1 out!12:30
xiangfuviric: thanks for notice. I will update our package.12:31
viricthank you very much :)12:32
viricxiangfu: you are the only one in the world doing anything about offrss ;)12:32
viricof course I think the program rocks, but some curtain keeps people blind.12:33
kyakviric: btw, i'm myself using newsbeuter on my PC :)12:34
virickyak: nothing comparable to offrss+elinks12:34
wolfspraulfree software never dies12:36
wolfspraulviric: if you have a nanonixos screenshot, that'd still be cool12:36
viricI'll open a ticket about that12:38
viricI've troubles remembering this.12:38
wolfspraulyou rememberd to thank me for including it in the news, which is over 1 month ago!12:43
wolfspraulso I'm not worried, one day you will have the screenshot :-)12:43
stefan_schmidtwpwrak: write2 works now. At least I can see the 230 go through hw_init setting the state to STATE_TRX_OFF and reading the correct value back13:01
stefan_schmidtwpwrak: it stops while setting the SR_CLKM_SHA_SEL register13:02
stefan_schmidtwpwrak: wow, no it went through the whole registering13:21
stefan_schmidtwpwrak: something is still fishy as sometimes the async urb are failing and sometimes not13:22
stefan_schmidtwpwrak: http://pastebin.com/47fw3ehx13:55
stefan_schmidtwpwrak: setting an address still fails though. Need to trace this down.13:56
qi-bot[commit] Stefan Schmidt: spi/atben: Fix write function. (ben-wpan-stefan) http://qi-hw.com/p/qi-kernel/cddec7a14:01
wpwrak*yawn* morning siesta terminated14:27
wpwrakstefan_schmidt: looks good :) are you also enqueuing an URB for the bulk EP 1 ? that would then give you interrupts14:28
stefan_schmidtwpwrak: not yet. Need to understand why not always all urbs are coming through. And I need to implement READ2 and getting address setting etc working14:33
stefan_schmidtwpwrak: but at least some real progress here.14:33
wpwrakyeah, looking better every minute :)14:37
wpwrakrejon: heya ! had some good flights ?15:09
rejonwpwrak, yep15:09
rejonbiz class15:09
rejonto chicago, to miami15:09
rejonnow here chillin15:09
wpwraksounds comfty :)15:12
wpwrakwhy ORD-MIA ? seems like a bit of a detour15:12
rejonyeah, united15:13
rejoni had biz class seat on that flight15:13
rejonvs. to dulles, back of the plane15:13
wpwrakah :) you wealthy man15:13
qi-botThe build was successfull, see images here: http://fidelio.qi-hardware.com/~xiangfu/compile-log/openwrt-xburst.full_system-07042011-1513/19:14
Action: stefan_schmidt is back at the computer19:35
Action: parolang just got his NanoNote :)20:06
parolangHow long does reflash_ben.sh usually take?20:09
parolangNevermind, now it's doing something :)20:11
kyakparolang: congratualtions :)20:20
parolangThanks guys.  I assume you guys are in China, which I imagine it's late there.  Didn't expect anyone one :)20:22
wpwrakparolang: we're everywhere ;-)20:29
wpwrakparolang: and welcome to the club ! :)20:29
parolangHonestly, I have gratitude towards those who took the risk in manufacturing the device.  I don't know if you've made your money back yet, hope so :)20:33
wpwrakstefan_schmidt: how's the battle going ? :)20:37
stefan_schmidtwpwrak: struggling to understand why some urbs are not going through sometimes.20:38
stefan_schmidtwpwrak: reading about how to handle such cases and how to re-submit, etc20:38
wpwrakstefan_schmidt: what does "not going through" mean ? do they come back with an error ? do they just never complete ?20:39
stefan_schmidtwpwrak: What I'm also missung to understand right now is how we would handle the INT with USB20:39
stefan_schmidtwpwrak: come back with error20:39
stefan_schmidtwpwrak: going to collect the numbers now to see if they stay the same or if we have different errors20:40
wpwrakstefan_schmidt: (int) easy: enqueue an URB on EP1 to receive 1 byte. when the URB completes, fire the interrupt and enqueue another URB.20:40
stefan_schmidtwpwrak: btw, did you have time or motivation to put one of your atusb into your pc and test the driver?20:40
wpwrakstefan_schmidt: (int cont'd) when shutting the driver down, simply cancel the last pending EP 1 URB20:41
stefan_schmidtwpwrak: so it waits there and the firmware does only react if a irq is needed?20:41
wpwrak(react) yes20:41
wpwrak(install) not yet, but i'll do so in 1-2 hours. just catching up on silly little things to do20:42
stefan_schmidtwpwrak: I also do only hit read 1 and write 2 from the classifier20:42
stefan_schmidteven when doing an address set via ip20:43
stefan_schmidtThat one still fails though20:43
wpwrakthe address set shouldn't do much down there anyway20:43
stefan_schmidt(testing) cool20:43
stefan_schmidtwpwrak: ah, right. That will only happen when doing auto ack and therefor hw address filtering20:43
wpwrakwhat should produce a longer read/write are izcoordinator and iz assoc ...20:44
stefan_schmidtwpwrak: I was thinking in terms of cc2420 and forgot that 230 does not have this yet20:44
stefan_schmidt(iz) not done this yet20:44
wpwrak(nb: read2 is not used so far. that would be for SRAM access. don't know if we'll ever run into a case that needs it.)20:45
stefan_schmidtwpwrak: ok. What would be a case for write 2+n then?20:45
stefan_schmidtmaybe we really only have read 1 and write 2 yet20:45
wpwrakyou could izchat for longer read/write20:46
wpwrakor else izcoordinator/iz assoc20:46
wpwrakand then dirtpan and whatever :)20:46
stefan_schmidtso nothing I'm doing right now :)20:46
stefan_schmidtmeans there hopefully is no bug with not hitting this cases right now. Just not the right use cases :)20:47
wpwrakso all works exactly like expected :) well, except the errors20:47
stefan_schmidtThe urb erros are what I'm looking into now20:47
wpwrakthe errors may actually be good. perhaps they tell us something useful. e.g., whether the USB controller has tried to get data. that could be very handy for synchronizing20:48
stefan_schmidtdoing some more tests and noting down the error codes20:49
stefan_schmidtwpwrak: so far it seems only to be -32 (EPIPE) and happens for both read1 and write221:01
stefan_schmidtwpwrak: another case is that is something just stalls without error on write221:01
stefan_schmidtwpwrak: That also leads to a "stuck" atusb which I have to unplug and re-plug to make it work again21:02
stefan_schmidtneed to check if the write2 stall and the "stucked" atusb are the same case21:02
stefan_schmidtwpwrak: two more times write2 stall and "stuck" atusb afterwards. Seems indeed related.21:11
stefan_schmidtwpwrak: Stalled endpoint (-EPIPE) from Documentation/usb/URB.txt21:16
wpwrakhmm. that usually means that your request wasn't understood. could it be that you're not initializing something ?21:20
stefan_schmidtwpwrak: could be21:24
stefan_schmidtwpwrak: strange is that is does not happen all the time21:24
wpwrakmaybe i should add a debug log ...21:25
stefan_schmidtwpwrak: maybe just luck in the working cases21:25
stefan_schmidtwpwrak: Am I supposed to do a ATUSB_RESET?21:25
wpwrakthat would reset the AVR21:25
stefan_schmidtwpwrak: I know21:26
stefan_schmidtwpwrak: so far I only reset the rf23021:26
stefan_schmidtwpwrak: No AVR reset needed I get it?21:26
wpwrakno AVR reset needed21:26
wpwrakby the way, do you know that you can snoop USB with wireshark ?21:29
stefan_schmidtwpwrak: yeah, seen something like this. Just never used.21:30
stefan_schmidtwpwrak: wanna me to a trace?21:30
Action: kristianpaul think that logs are an essential part for every application21:30
stefan_schmidtkristianpaul: doing the log on the micro controller fimrware is not that easy :)21:30
kristianpaulhum :)21:31
stefan_schmidtkristianpaul: the kernel driver side is easy enough with printk and dmesg :)21:31
kristianpaula serial port i will think, coluld be usefull, even only with  TX, but i'm not aware of the details of wpwrak atben/atusb implementation21:32
stefan_schmidtkristianpaul: so far we should be able to work with the tools we have. Actually I don't expect to much trouble on the firmware side.21:34
kristianpaulgood plan :)21:34
wpwrakkristianpaul: serial would be nice to have but messy to implement21:35
kristianpaulman, ietf is like wikipedia every page link to another rfc..21:36
kristianpaulwpwrak: or a blink led :)21:36
kristianpaultwo blink leds and you're done ;)21:36
methril_workwpwrak, i`ve implemented some serial/usart features in some devices21:43
methril_worknot so hard..21:43
methril_workbut so arch dependant21:43
methril_worktoday i fix one bug in ISR for RX when RX is not enabled !! and i could debug again!! /)21:43
Action: methril_work bbl going home21:44
wpwrakmethril: adding a uart is very easy. connecting it can be messy.21:48
qi-bot[commit] Werner Almesberger: atusb/fw/: added optional debug log with request ATUSB_READ_LOG (master) http://qi-hw.com/p/ben-wpan/39e3d4e21:53
qi-bot[commit] Werner Almesberger: tools/atrf-id/atrf-id.c: new option -D to read atusb debug log (master) http://qi-hw.com/p/ben-wpan/391c37f21:53
wpwrakstefan_schmidt: if you set DEBUG_LOG and build the new firmware, you should be able to see the last operations with atrf-id -D21:54
wpwrakback in a sec21:54
wpwrakthese are all the things that aren't standard USB control transfers21:58
wpwrakoutput is a bit ugly. prettifying it ...21:58
stefan_schmidtwpwrak: done, you want the output when it hangs?22:02
stefan_schmidtwpwrak: hmm, problem22:04
stefan_schmidtwpwrak: stefan@excalibur:~/Studium/Diplomarbeit/ben-wpan/tools/atrf-id [master]$ sudo ./atrf-id -D22:04
wpwrakhangs ?22:04
wpwrakor errors ?22:04
stefan_schmidtthe problem is that I only can connect it when the kernel driver releases the device22:04
stefan_schmidtand after that I don't get any more debug log22:05
wpwrakcan you somehow convince the kernel to release the device ?22:05
wpwrakeasier than me writing code to log to EEPROM :)22:06
stefan_schmidtNo method I would know about when it hangs22:08
stefan_schmidtshould we not be able to release it with libusb22:08
wpwrakprobably not on behalf of the kernel ...22:09
wpwrakmeanwhile, does wireshark show anything useful ?22:10
Action: stefan_schmidt fires it up22:11
stefan_schmidtah, I need usbmon loaded22:13
wpwrakafk for ~10 min22:14
stefan_schmidtwpwrak: wireshark shows urb traffic in the hanging state22:16
stefan_schmidthmm, it shows -EPROTO (71)22:17
stefan_schmidtas protocol error in status22:17
stefan_schmidtsure is that there is a loop bouncing urbs back and forth22:18
qi-bot[commit] Stefan Schmidt: spi/atusb: A little cleanup (ben-wpan-stefan) http://qi-hw.com/p/qi-kernel/2bcae8122:25
qi-bot[commit] Stefan Schmidt: spi/atusb: More cleanup (ben-wpan-stefan) http://qi-hw.com/p/qi-kernel/6a293db22:25
qi-bot[commit] Stefan Schmidt: spi/atusb: More cleanup (ben-wpan-stefan) http://qi-hw.com/p/qi-kernel/54f9bfd22:25
qi-bot[commit] Stefan Schmidt: ieee802154/at86rf230: Be more verbose and disable irq handling for now (ben-wpan-stefan) http://qi-hw.com/p/qi-kernel/88bf9af22:25
wpwraknow. let's have a look at your code ...22:33
wpwrakread1 and write2 are affected ?22:35
stefan_schmidtwpwrak: write2 only it seems22:35
stefan_schmidtwpwrak: just waited for you to come back22:35
stefan_schmidtwpwrak: getting to tired here. I think in 10 mintes I have to go to bed22:36
stefan_schmidtwpwrak: all changes are commited. Feel free to commit directly into my branch or do your own on top of it22:36
wpwrakerm ...22:36
stefan_schmidtwpwrak: but I can spare some minutes for code review22:36
wpwrakyou're freeing the URB immediately after submitting it ?22:37
stefan_schmidtwpwrak: I do what?22:37
stefan_schmidtlet me check22:37
wpwrakusb_submit_urb .. usb_free_urb ... kfree(req)22:37
wpwraki'd rather do the freeing in the callback :)22:38
stefan_schmidtthat one survived from the wait_for_compe.. scheme I had before22:38
Action: stefan_schmidt mumbles something about getting blind while reading own code22:39
stefan_schmidtit must be blazing fast when it sometimes works out well before freeing :)22:39
stefan_schmidtlets see if this fixes all issues22:39
stefan_schmidtoh, I'm not passing atusb as context but the spi message22:40
qi-bot[commit] Werner Almesberger: Revert "atusb/fw/: added optional debug log with request ATUSB_READ_LOG" (master) http://qi-hw.com/p/ben-wpan/ead7ade22:40
qi-bot[commit] Werner Almesberger: Revert "tools/atrf-id/atrf-id.c: new option -D to read atusb debug log" (master) http://qi-hw.com/p/ben-wpan/505959722:40
stefan_schmidtno handle22:41
wpwrakseems that we don't need this after all :)22:41
stefan_schmidtmaybe I should add the msg to the atusb struct as we need both in the callback22:41
wpwrak(blazing fast) well, depends how soon the memory gets overwritten :)22:41
wpwrak(add msg) sounds reasonable. there can be only one message in flight at a time anyway22:42
stefan_schmidtwill do that22:42
wpwrakyou also need a pointer to the transfer22:42
wpwrakthen you can make it all very nice and compact22:42
wpwrakerr, no ... silly me. the second transfer has to be handled at the same time as the first22:45
wpwrakso you just pass the rx/tx buffer as transfer_buffer/buffer_length in usb_fill_control_urb22:48
Action: stefan_schmidt hopes the first try works out. Bedtime.22:49
wpwrakif you want me to continue experimenting ... i still have about 4 l of fuel (coke) left 22:49
stefan_schmidt(compact) There are a lot things that needs to be cleaned up and can be done more compact. I often gone the lazy path.22:50
stefan_schmidtwpwrak: If you have time and motivation I would love you to experiment22:50
stefan_schmidtwpwrak: I'm just trying my changes for the free in the callback now and then I need sleeping22:50
wpwrakkewl. wiring up my old "pc for experiments"22:50
stefan_schmidtwpwrak: And I'm pretty sure you will spot many more bugs from me in the code :)22:51
stefan_schmidtwpwrak: as I said, I'm fine you you push it directly into my branch (or anywhere else you want)22:51
stefan_schmidtgetting up tomorrow and have some more fixs in would be great after /me battling that long with this "simple" driver22:52
stefan_schmidtThis free() thing makes me a bit angry wasting your and my time. But shows clearly that I got blind on my own code22:53
wpwraktiredness can do this sometimes :)22:54
stefan_schmidtwpwrak: hanging again in write2 with my change...22:55
qi-bot[commit] Stefan Schmidt: spi/atusb: Freeing the urb in the callback instead directly after submit. (ben-wpan-stefan) http://qi-hw.com/p/qi-kernel/c1c221322:55
stefan_schmidttentative fix but does not help22:55
stefan_schmidtsleep now or I'm just doing more stupid things22:56
stefan_schmidtwpwrak: have fun with it for the next hours.22:56
stefan_schmidtwpwrak: I should be back in 8 hours or so.22:56
wpwrakcan you push the latest version ?22:58
wpwrakah, here it is22:58
stefan_schmidtwpwrak: all in now, no more local changes. You have the whole mess. :)22:59
stefan_schmidtwpwrak: night22:59
wpwraknow .. to find a vga cable that hasn't been ubb-vga-modded ...23:01
--- Wed Jul 6 201100:00

Generated by irclog2html.py 2.9.2 by Marius Gedminas - find it at mg.pov.lt!