#qi-hardware IRC log for Saturday, 2011-06-11

kristianpaulhola Fusin 00:00
kristianpaulyeah SIE v4 is coming 00:01
Fusinhi kris :)00:01
kristianpaulseems no more fpga, shame00:02
kristianpaulbut indeed that could drop prices and move ingenic soc to more general porpuses setups00:03
kristianpaulxburstduino ;) like board, why not? :-D00:04
wpwrakkristianpaul: if they're using the 4725B, i wonder why they didn't throw out the NAND too. the 4725 has two MMC controllers and can boot from MMC00:05
kristianpaul1 bit MMC...00:05
kristianpaulwith no linux suport. wel..00:05
kristianpaulhttp://wiki.linuxencaja.net/wiki/File:SIE_V4.png btw00:05
wpwrak4 bit MMC and 4 bit MMC00:05
wpwrak(if the manual can be believed)00:06
wpwrakNAND and MMC0 share pins. so if you choose NAND, you lose the 4 bit MMC controller00:06
kristianpaulmy sie have 4725, and so far i remennber MMC is 1 bit00:07
kristianpaulbut yeah, dropping nand will be could00:07
kristianpaulmore gpio00:07
wpwrakthe black PCB seems to be an older version. the yellow one is the latest, says rafa00:07
kristianpaulone more chip you can buy everywhere00:07
kristianpauli was confused then :-)00:08
wpwrak(SIE MMC 1 bit) yes, of you have NAND. then you can only use the MMC1 controller, which is 1 bit. see, NAND costs you more than you may think ;-)00:09
LunaFrizzlewpwrak, gotta go to bed now, I'll stick with the ben + ubb for now :)00:10
wpwrakno NAND would also make the layout a lot simpler00:10
LunaFrizzlewpwrak, thank you so much for all theses info, see ya00:11
kristianpauland smaller00:11
wpwrakand if you choose to boot from USB, you could use the MMC0 controller for other things00:11
kristianpaulxburst have internal sram?00:15
wpwrakthey should have a little, for booting00:15
kristianpaulsome kbytes?00:15
wpwraksomething 4 k-ish would be common. i'm too lazy to look it up :)00:15
kristianpaulOn-chip boot ROM 8KB00:21
wpwrak4 kB ROM and 8 kB RAM, according to section 3.5.2 of Jz4725B_pm00:24
kristianpaul4kB Boot ROM memory00:25
kristianpaulwell, having some tiny code running at 336Mhz still intersting :-)00:26
wpwrakoh, that's just for the first stage. then you load whatever else you want.00:28
kristianpaulhum? i dont understand00:28
kristianpauli mean what whatever else i can load? (considering i was thinking in remove SDRAM chip :-) )00:30
wpwrakno, remove NAND. not SDRAM :)00:31
kristianpauli want rip all ! ;)00:31
wpwrakkristianpaul: consider just power, headers, a few buttons, and a piece of paper telling you which buttons to press when ;-)00:37
DocScrutinizerMentorgraphics PADS *free* PCB viewer!!!!02:26
wpwrakfree binary or open source ?02:27
DocScrutinizer"we" actually could give FR PCB to community now02:27
DocScrutinizerof course free binary, for windows ;-D02:27
wpwrakwith the calypso ? hardly ...02:27
DocScrutinizerWTF is calypso??02:27
DocScrutinizernm, Sean will not care anymore02:28
DocScrutinizerand I'm still hesitant to "break the rules"02:28
DocScrutinizerI always like to push as hard as I can, for community, but I'm not the one to ignore the limits02:29
DocScrutinizeras long as it's about (ex-)contractors02:30
DocScrutinizerI don't give a fkying F* about calypso though02:30
DocScrutinizerhell this component isn't even sourceable since almost 3 years02:31
wpwrakit probably still is. dunno if from ti, though02:31
DocScrutinizerplus if the PCB design were of Intel, I'l like to sue them still today, for #102402:32
DocScrutinizerso what's Int'l's royalties in PCB design anyway??02:32
DocScrutinizerthe size of their chip? or what?02:33
DocScrutinizeror did OM agree on never mentioning we are using this chip? ;-P02:33
DocScrutinizernah, forget calypso, forget Intel. It's all about OM and Sean02:34
DocScrutinizerI always said there's pretty limited use in community getting full access to eachof the 8 layers of the PCB, but Wolf been always keen to disclose *all*02:36
DocScrutinizerI'm not really averse02:36
DocScrutinizerand now that we *can* do it, I even like to push an itsy bit02:37
wpwrakwith gta02-core not getting components from sean and openmoko having abandoned technology anyway, the layout has pretty much drifted into irrelevance02:37
DocScrutinizerthat's why I think asking Sean about it is a lost case02:38
DocScrutinizermaybe he even doesn't own the technical stuff of FR anymore02:39
Action: DocScrutinizer wonders idly how Wikireader ended at MEDION02:39
DocScrutinizermakes me rethink my previous statements and my usual practice of checking carefully the destination of sftp transfers ;-)02:41
DocScrutinizerbut nah, I want to stay in business02:41
DocScrutinizerwpwrak: could you asl wolf about it, just what's his take on it all02:42
wpwrakDocScrutinizer: are you afraid to ask him yourself ? ;-)02:43
DocScrutinizernah, he's not online02:43
wpwrakDocScrutinizer: you want me to call him in china ? ;-)02:44
DocScrutinizerit's 4:44 here02:44
wpwrakuntroubled dreams ! :)02:44
qi-botThe build was successfull, see images here: http://fidelio.qi-hardware.com/~xiangfu/compile-log/openwrt-xburst.trunk-full_system-06102011-0026/03:52
tuxbrain_awaywpwrak: when do you think we can have a wpan0 interface on NN? you know use it as easy as we use usb0 now?07:23
tuxbrain_awaywpwrak: did you read my first measures? 07:25
tuxbrain_away250k 12m07:25
tuxbrain_away500k 12m07:25
tuxbrain_away1M    11m if orientation is correct07:25
tuxbrain_away2M    7m if orientation is correct07:25
tuxbrain_awaythis was for an atusb/atben pair I'm also courious about atben/atben and atusb/atusb 07:27
tuxbrain_awaynot willing to flood but ,  wolfspraul have you see the first range results on fab atben/atusb?07:34
tuxbrain_awaywell time to some spanish revolution and then to flash/test some wpan, I'm really happy with the results :) my heartbeats had go lower when I see the test past07:42
wpwraktuxbrain_away: hehe ;-)08:41
wpwraktuxbrain_away: (wpan0) dunno. first, lemme test a few more things about the firmware. to make a version you can flash in all those devices.08:42
wpwraktuxbrain_away: 6lowpan will be harder to use, because you may also have to introduce ipv6 to your network.08:43
wpwraktuxbrain_away: (ranges) have you tried how far you can get with optimal orientation at 250 kbps ?08:44
wolfspraulno I haven't seen results yet, nothing conclusive at least09:04
wpwrakwolfspraul: he got 12 m indoors at 250 kbps and still 7 m at 2 Mbps (if holding the ben at the right angle). that's a bit better than what i obtained with my prototypes. (about 7 m here at 250 kbps with the right angle)09:06
wpwrakwolfspraul: so this looks pretty good. seems that the open-loop manufacturing worked ;-)09:06
wolfspraulthat's for both 8:10 and usb versions?09:11
wolfspraulI read something about trouble with usb, clock stability or what it was (I didn't try to follow the entire conversation)09:11
wolfsprauleverything works, everything can be sold? 100% yield rate?09:12
wpwrakthat's between atben amd atusb for now09:14
wpwrakthe clock stability was a problem of tuxbrain's laptop (some arm notebook) having very noisy timing. with a better machine, the results are stable.09:16
wpwrak(yield) i think he only tested one board from each set yet. so we'll see how that goes. but it means that the yield is > 0%, which was my greatest worry :)09:17
wolfspraulah :-)09:17
wolfspraulthat's open-loop manufacturing indeed09:17
wpwrak(e.g., chip rotated the wrong way. bzzzt, game over)09:17
wolfspraulwell, Tuxbrain needs to understand he is the manufacturer here. He can only sell features (and boards) that he has tested.09:18
wolfspraulI hope he understands that :-)09:18
wolfspraulevery feature has to be tested on every board09:18
wpwraki think he does. no worry there09:18
wpwrakwhat went wrong was that the smt fab went ahead without anyone testing what came off the line. that was high-risk.09:19
wolfspraulsure but that's what they will do if there is nobody to stop them09:19
wolfspraulthe margins are too thin there (by design) to afford any sort of doubts or questions...09:20
wpwraktesting is just a matter of putting things in the right (physical) configuration and then "make usb" or "make ben". there are 1-2 operator inputs but that's all. have you seen this ? http://downloads.qi-hardware.com/people/werner/wpan/prod/09:20
wpwraknot sure if it was an attempt to push the margins or if their customers are just generally the trusting kind09:21
wpwrakfrom the way they work, i have the impression their typical customer doesn't worry very much about production issues. if the EDA system says it's okay, send the files to the fab and they'll take care of everything, right ? ;-)09:23
wpwrakso there's a bit of consulting work in their price. considering that this was a run with a never-before fabbed design, several unproven tools, and probably also tuxbrain's first, probably a reasonable approach. now we can go to the chinese penny-pinchers for the next million boards ;-)09:26
wolfspraulyes many people produce blind, but that's why you have so much crappy hardware nowadays, stuff you take out of the box that just doesn't work09:28
wolfspraulor some features don't work09:28
wolfspraulthat's because you are the first one to run them on that particular device :-)09:28
wpwrakokay, that's even worse09:28
wolfsprauleven for a brand company, it may be cheaper to catch those things at the retail end, with a no-questions-asked return/replace program09:29
wolfspraulyou need this anyway for things that broke after testing09:29
wolfspraulso you can throw manufacturing issues into that program as well, if it's cheaper :-)09:29
wpwraksigh. welcome to the fight club ...09:30
wolfspraulI'm all easy about this. If I order boards from tuxbrain and they don't work, I expect him to replace/fix my boards :-)09:30
wolfspraulbut anyway, results of testing the first 2 boards seem good, and that's great!09:31
wolfspraulhe made 100 each? or more?09:31
wpwrakyeah, and both waste a multiple of the value of the merchandise in logistical overhead and shipping ...09:31
wpwraki think ~130 atben. not sure how many atusb.09:31
wolfspraulyes, but the margins are quite fat there09:32
wpwrakhe also has a ton of pcbs left (making 100 costs almost at much as making 500, so he got ~500 of each). so if there's good demand, he can quickly make more. and actually begin having a margin :)09:33
wpwrakit's fun to watch the crystal drift with temperature changes ;-)10:15
Action: dvdk starts to hate cmake. people complain about autofail tools? it can get much worse than that10:29
wpwrakmeanwhile, in a parallel universe, the eclipse foundation publishes a document titled "The Open Source Developer Report"10:41
wpwrakdvdk: we're doing open source, so i think we should switch ;-)10:42
dvdkwpwrak: cannot parse that :) 10:43
dvdkah, finally plplot compiles again10:43
qi-bot[commit] David Kühling: plplot: make compile again with openwrt trunk; switch to using cmake.mk http://qi-hw.com/p/openwrt-packages/01b5b1510:44
wpwrakdvdk: switch to the eclipse ide10:46
Action: larsc uses MS Visual Studio C++ as an IDE for all his linux kernel work ;)10:49
dvdkwpwrak: well when porting other people's software (cmake is used by plplot), there's just not much we can do to move them over to visual studio :)10:52
wpwraklarsc: ah, i though you were doing it in C# with .net and do some automated conversion to C ;-)10:53
wpwrakdvdk: for me, make is one of these IT honeypots. just like TCP. people who don't understand the problem is solves try to construct "better" replacements that solve a subset. that's a fairly common pattern. what makes "make" and TCP special is they're the target of particularly persistent yet unsuccessful attempts at replacing them.10:56
wpwrakof course, the "i can do a better tcp" fallacy is great entertainment in research. when you get a paper with such a project, it's time to prepare the pop corn while watching the drama unfold ;-)10:57
wpwrakof course, what's even better is if a reviewer doesn't notice the handful of cases where the author actually knew what he was doing. e.g., van jacobson (kinda "Mr. TCP") once got a paper rejected with the reviewer claiming that the (anonymized for the review) author didn't understand tcp ;-)11:00
larscyea, anonymized reviews can be amusing at times11:05
dvdkwpwrak: just getting back.  your IT honeypot thought really makes me smile.  c12:19
tuxbrain_awaywolfspraul: yes I will test all the board, all boards shiped will pass the werner test and a distance test, if there is faulty ones I will agree with werner how to analize. there was 135 atbens 110 atusb, I hope I will test enough(and that they pass the test) over the weekend to start selling on tuesday (monday is hollidays here)12:21
tuxbrain_awaybtw how many do you want? :)12:21
wolfspraulthen we need to make sure the software works as much as possible out of the box in the OpenWrt image (that's what I will go after)12:21
wolfspraulhow are you selling them? pair of ben+usb?12:22
wolfspraullet me see the price and all when it's online :-)12:22
wolfspraulI first need 2 or 3 pairs I guess, then I need to think whether I can bundle it somehow, or sell, but that will require more testing and thinking.12:23
tuxbrain_awaywolfspraul: totally agree, I don't know if with the dirtpan we can do a wpan0 interface that will be the best of the best case!12:23
wolfspraulalso I don't want to give any away because that would devalue the product on your end - don't want to do it12:23
wolfspraulwe give away stuff of other companies :-)12:23
tuxbrain_awaywhatever I will send some to some devels 12:24
wolfspraulmaybe we can make some games where people can play with multiple Bens12:24
tuxbrain_awayok I must leave12:24
wpwrakwolfspraul: (games) you could to the "snake in a tunnel" thing ;-))13:44
Fusinhi qiots15:00
Fusinhi doc15:18
DocScrutinizerwpwrak: you're aware a *proper* QA still is a bit more challenging than a mere sw-based test of functional aspects. Of course you know. I'm just pondering idly where's the point where this kind of open hw manufacturing has to start to think about such things like voltage / temperature stress tests. Given the fact we mostly got away without any such proper QA at OM ;-P15:20
wpwrakDocScrutinizer: i would put most of this into the domain of "design verification". unless you actually see a lot of failures in the field due to parts not meeting specs, there seems little point in testing all this, too15:24
DocScrutinizersome stories for the general background: on really large scale production, you probably go to stress a 5000 sample, then if you e.g find 1% drops out at temperatures <5°C you try to find *other* parameters of the DUT that indicate same problem - like "oscillator takes >3*Tavrg to start" and then you test for these parameters instead15:24
wpwrakDocScrutinizer: i mean, we're not designing nuclear reactors here. well, considering it, that may not be a good example :)15:24
DocScrutinizera simple VDD variation stress test is rather standard though, as it tells you a mad friggin lot about real state of DUT15:26
wpwrakDocScrutinizer: yup, exactly. you need to observe some actual failures. our sample size is way too small for that. maybe we'll discover one or two systematic defects from this run, maybe with the opportunity to do better the next time, but with such tiny quantities, something has to go massively wrong before you even get a blip on the radar.15:27
wpwrakDocScrutinizer: (vdd) that needs equipment ... tuxbrain can't even afford a low-end scope. so ...15:27
DocScrutinizerjust thinking15:28
FusinIknow you know :D15:28
wpwrakFusin: iKnow - quick, before apple grab it ! ;-)15:28
DocScrutinizeryou know I always like to have a complete plan, so you know what you are discarding / postponing, and why15:28
wpwrakDocScrutinizer: my tests try to test everything i can do with on-board means and with a pair of devices. the chips also detect voltage excursions and such, so there's a chance to pick up some of these. also, if there's massive droop, it'll show up in the transmit test.15:29
DocScrutinizerbetter go "that's the optimum right way: A, B, C... - and this is what we do right now: B as we can't do A and C is nonsense"15:30
Fusinlol @ wpwrak15:30
DocScrutinizerI'm well aware this device isn't properly ESD tested and whatnot15:31
wpwrakDocScrutinizer: we what can slip through the tests are parasitic effects and slight performance degradations. e.g., you may only get half the range. but you'll probably never know if this is because of a poorly performing device or because of your environment.15:31
Fusinwhat about an iBen?? :-> and iNano, iOpenMoko? 15:31
Fusinfood time15:31
wpwrakDocScrutinizer: atusb has TVS at least ! ;-)15:32
wpwrakdigi-key 445-2559-1-ND15:33
DocScrutinizertoo lazy15:33
DocScrutinizerwill be fine15:33
DocScrutinizergood job!15:34
wpwrakyeah. they're designed for such things. ought to work :)15:34
DocScrutinizeralso on antenna? ;-P15:34
wpwrakthe antenna is supposed to be isolated15:34
DocScrutinizerindeed, I think tehre's a balun or sth iirc15:35
wpwrakotherwise, you'll eventually etch it away just with your hands :)15:35
wpwrakthere's a balun, yes15:35
DocScrutinizergood enough for ESD-prot15:36
wpwrakkewl. one worry less :)15:36
wpwrakthe boards are designed such that you should be able to solder an antenna connector (U.FL type), so it's nice if this doesn't cause an esd hazard15:37
DocScrutinizeroh, yu gotta remove a 0R then?15:37
wpwrak(antenna connector) with a bit of scraping and it's pretty much irreversible. but can't hurt to have that option.15:37
DocScrutinizeryes, good job again15:38
wpwrakthanks :) your praise honors me ;-)15:38
DocScrutinizersorry, my straight honest notion about it, nothing implied15:39
wpwrakheh :) still good to know that my hack passes scrutiny15:40
DocScrutinizerjust listening to my inner "Ahh varistors :-) AAAHH an option for connector :-))"15:40
DocScrutinizeranother such detail and I'm sold to the device, no matter if I need it or not :-)15:41
DocScrutinizerwhat's the TX power? Your bandwidth figures seem pretty decent15:43
wpwrakhmm, does dfu-upgradeable firmware count as a "nice" feature ?15:43
wpwrak+3 dBm = 2 mW15:44
DocScrutinizerI think is counts, yes15:44
DocScrutinizer(2mW) :nod: would consider that a design well in bound of expectations15:45
wpwrak(take that, whitequark, with your 1 W ghettoblaster !)15:45
DocScrutinizernow for next version of device which comes with 4GB flash mass storage integrated ;-)15:46
wpwrak(2 mW) that's the nominal tx power. i don't have the means to measure the really emitted power. it seems that there's nothing significant at the 1st harmonic. then the range of my equipment ends, too.15:46
DocScrutinizer10m range for 2mW is nothing to worry about anyway, pretty ok15:47
DocScrutinizerthough with a proper antenna and good RX you should reach 100m easily15:48
DocScrutinizerwhere proper==external here15:48
wpwrakyeha, i think it's also a range for real-life use. particularly when i compare it with wlan.15:48
wpwraksome chip makers claim that sort of distance, yes15:48
DocScrutinizer"B-52s-The_Discography-9albums: Download complete" \o/15:51
wpwrakbad boy :)15:51
DocScrutinizernah, after my records got stolen 3 times in my life, I feel not any bad with it anymore15:52
wpwrakDocScrutinizer: 3 times ? maybe you should consider more upscale neighbourhoods ;-)15:52
DocScrutinizermakes a total of like 4000 stolen vinyl and CD, some 300 of them bootlegs15:52
wpwrakin backups we trust :)15:53
tuxbrain-mobtesting the irc on the mobile, seems that all works ok :)15:53
DocScrutinizergor pissed on music somewhat, even since the 2nd time15:53
wpwrakbut yes, i had some CDs stolen from my car once, too. naturally, most of them not yet backed up :-(15:53
DocScrutinizernow rethinking my life w/o music15:54
wpwraktuxbrain-mob: ben + dirtpan ? :)15:54
DocScrutinizeralas a lot of the 80's stuff never will come back15:54
DocScrutinizere.g B-52s15:54
DocScrutinizerhard or impossible to get all albums, even for gold15:55
DocScrutinizerso FSCKYOU sony, EMI, whatever the name15:55
tuxbrain-mobhehehehe create the wpan interface and then we can talk about it15:55
wpwrakDocScrutinizer: let's call it the distributed global cultural conservation project ;-)15:56
tuxbrain-mobI will recoonect later see you15:56
wpwraktuxbrain-mob: with dirtpan, you have IPv4. so ...15:56
DocScrutinizerindeed, and you're free to laugh at me but I considered to start an "official" such project15:56
wpwraktuxbrain-mob: no support for atusb yet, though. so you need one ben as gateway.15:56
DocScrutinizermodulo the "distributed"15:57
DocScrutinizerjust for music15:58
DocScrutinizeras it's so goddamn volatile you'd never expect in times of ubiquitous copy technology15:58
DocScrutinizerjust the general notion of "ther's for sure a copy out there" makes good music vanish into oblivion it seems15:59
DocScrutinizerthere *IS NO* mastertape!16:00
wpwrakDocScrutinizer: (project) good luck ! :) i think you need a few paradigms shifts to happen before such a thing can fly, but it'll probably take a few martyrs to get there.16:00
DocScrutinizernah, write-only petabyte store in a bunker would be a simple easy approach16:01
wpwraknow ... there's that critically important interrupt. if it leave it on, it fires all the time (which is expected in that case), wasting system resources.16:01
wpwrakif i turn it off completely, the thing just works. this makes no sense :-(16:01
DocScrutinizereverybody (incl all artists) could get a nuke-safe backup of their artwork, given they agree everybody can "watch" the backup after tedious filling of papers etc16:02
DocScrutinizerthe duty of the archive managers (and tools) would be to assure the readability of content over the decades, and keep doublettes out, and process read-inquiries16:05
DocScrutinizerplus of course manage world-readable state after expiry of (C)16:05
DocScrutinizerwhenever that is16:05
DocScrutinizersuch an archive also would be a nice way to claim (c) by mere timestamp and signature of the artwork16:07
wpwrakyou'll never get this past the labels. they'll just claim that they keep the backups. their milk cows are probably already forbidden by contract to let anyone outside maintain a backup.16:07
DocScrutinizerthe archive won't care :-)16:07
wpwraknice. if i make the transfer large enough, the interrupt starts to matter. the world makes sense again ;-)16:08
DocScrutinizerexistence of backups isn't covered by any (c)16:08
DocScrutinizerjust distribution is16:08
DocScrutinizerthe global art archive isn't world readable in a simple way, so any sharing concerns are moot16:09
wpwrak(c) covers copying. you can fight it out with their lawyers whether external but non-public backups constitute fair use. and if your backups go public within the 70 or whatever years, the case is pretty clear.16:10
DocScrutinizerthink of it like the US congress library16:10
wpwrakcall it a vault. to be opened by your grand-grand-childen ;-)16:10
DocScrutinizerjust for digital and digitized artwork16:11
wpwrakand good luck convincing anyone your business will still be around in 70 years. note that you either need to laser the data into gold or such, or make regular refresh copies.16:11
DocScrutinizerinstead of printed books16:11
DocScrutinizerrefresh copies is the duty of the employees there - you'll of course need sponsors, or a foundation16:12
wpwrakconventional digital media only lives for a few years, decades at best. your cd-r will be unreadable in 70 years, probably even if you add a lot of redundancy16:13
DocScrutinizeralso you might consider to ask for fees for reading their own backups, just in case. Maybe even fees for allowing them to access their own backups any time16:14
wpwrakwell, you can probably offer a number of add-on services, yes :)16:14
DocScrutinizerthis way it becomes kind of a mega-remote-backup facility16:14
wpwrakas long as Anonymous don't pay you a friendly visit, making a little download :)16:15
DocScrutinizerwhich you may use for relatively little fee, given you agree on the terms of semi-scientific public use16:15
wpwrakit would of course be safer if you really took the data offline until the public domain data.16:16
DocScrutinizerI admit I haven't spent more than maybe 15min on that idea, just found it compelling when it occured to me16:17
wpwrakand you want to have a minimum "keep it closed" interval. otherwise, people will just use you as public web space, and send you their warez collection.16:18
wpwrakso, better to have the minimum near 70 years anyway. that way, even pirated content will be safe :)16:18
DocScrutinizerdata offline was the idea, "others" can access only by paying insanely high fees or by going thru a real nasty paperwork that takes months16:18
DocScrutinizeror just go there to *listen*16:18
wpwrakif you let "others" access it, then you open yourself to the usual litigation16:19
wpwrakmake it a vault. buried treasure. let generations to come marvel at your work.16:20
DocScrutinizerthat's the basic idea16:21
wpwrakthe problem is finding a way to ensure this kind of longevity.16:21
DocScrutinizeryou nevertheless need candy to lure folks in16:21
DocScrutinizerthe candy is a) they can access it themselves any time though for some fee maybe16:21
wpwrakthen you have the authentication problem16:22
DocScrutinizerwe're hackers! we don't have such problems, we got the solution to them16:22
wpwrakthe candy would be the eternal fame. make good PR. it may work.16:23
DocScrutinizerb) for a really high fee forensics, scientists etc can search the *content*16:24
wpwrakyou still have the issue tht labels will try to kill you on general principles. after all, the artists sell their right to the labels, so they have no right to let you have an archival copy, unless the label agrees. and why would they ?16:24
DocScrutinizerc) everybody can search the *directory*16:24
wpwrakthat is, unless you get them to agree to your plan in the first place. so plan on some budget for travel, a few suits, and long meetings with IPR lawyers.16:24
DocScrutinizernah, you sell the (C) but you're always the owner of your own artwork and may make copies for your own purpose as many as you like16:25
DocScrutinizeryou just mustn't *COPY* (means "distribute")16:25
wpwrakyou could sell it to the labels as a protection against laws constraining them. e.g., laws that weaken the protection for apparently abandoned works.16:25
wpwrakread the laws. they talk about copies, not distribution.16:26
DocScrutinizerthere's a lot of potential benefit in such a global un-nukable write-only/no-delete/no-general-world-access lib16:26
wpwraksure. you just need to pull it off ;-)16:27
DocScrutinizer(c) isn't a problem of the GAA though, if labels want to sue somebody, they have to start at their artists who claim they are owners of that stuff when putting it into that archive16:28
wpwrakmaybe talk to google. they love to collect data. and they can talk eye to eye with labels. unlike you, who starts with a legal budget handicap of a few megadollars.16:28
DocScrutinizerand depending on whether submitter makes description world-readable or not, labels never may get to know about it16:29
DocScrutinizerwpwrak: exactly, why didn't I think of google first instance16:29
wpwrak(sue the artist) good point. this means that the index shouldn't be public. otherwise, you're exposing your customers.16:29
DocScrutinizer"the eternal youtube"16:29
DocScrutinizerthat's up to the customer to set their index entry public or not16:30
wpwrakwell, that's choice, yes16:30
wpwraks/choice/a choice/16:30
DocScrutinizerthe only immutable precondition is they agree on never-erase policy and disclosure after $random time depending on what they declare the stuff to be16:31
DocScrutinizermax maybe 100 years which would cost them quite some money, 10 years and you go for free16:32
wpwrakhmm, you'll probably get lawsuits on erasure16:32
wpwrak10 years means that you'd be at risk of distributing pirated material16:32
DocScrutinizerwe don't guarantee non-erasure, we just say there's no way for *user* to ever hit the delete button16:33
wpwrakwell, they'll send their lawyers ;-)16:33
DocScrutinizersure we get problems with pirated material. Given the never-really-fileshare concept we me get away with it nevertheless16:33
DocScrutinizeror extend the period by another 10 yeras on a simple dispute16:34
wpwrakpornstarlet X makes some movie at age 18. things it's work conserving and sends it to you. ten years later she finds god and wants all copies destroyed. including yours. maybe she has a rich husband by then who pays for her legal extravaganza.16:34
DocScrutinizerwho's sending their lawyers to facebook to make sure their private data got erased? ;-P16:35
DocScrutinizerstarlet -> extend non-disclosure to 10016:35
DocScrutinizerset index to hidden16:36
DocScrutinizercan't erase sorry -our technology doesn't even allow that 16:36
wpwrakscenario 2: same pornstar, but it's fan Y who sends you her movie, claiming he's got the rights. then she'd go after you too, and you wouldn't even have a usable contract holding you free from erasure requests in this case.16:36
wpwrak(facebook) oh, i'm sure they get such things16:37
DocScrutinizerand just in case we get a legal order to erase, we still can do. Rmind we never *guaranteed* never-erase16:37
DocScrutinizerthough SOP for erase request is "set disclosure time up to max, set index and description to hidden"16:38
DocScrutinizeroh and "deny disclosure requests, unless they are by law enforcement"16:40
wpwrakyeah, you could do that. dunno what the legal situation would be. this may be something without precedent. can you be forced to destroy an illegal copy you vow to never touch before it enters the public domain ?16:40
DocScrutinizerprolly not16:40
DocScrutinizeras there's no way for foe lawyer to prove you even are in posession of that copy anymore16:41
wpwrak(law enforcement) sure. they'll just subpoena what they need. add it to the damages they sue you for.16:41
DocScrutinizerhey, that's a Schroedinger's cat :-D16:42
DocScrutinizersorry I can't destroy that allegedly existing copy, as I promised I never will touch that tape, so nobody can know if there's that copy on it16:43
DocScrutinizernobody can force me to touch that tape unless they got sound clue about their copy being on it16:43
wpwraki don't think it works like that ;-)16:44
wpwrak"sorry, you can't search for my meth lab unless you can have proof it exists" ;-)16:44
DocScrutinizermaybe not in US - for sure in sme northern EU countries :-)16:44
DocScrutinizeronly police may search for my meth lab, and even those need a judge's order to do so16:45
DocScrutinizeryou can't sue the mayor for the morphine in their town's bunker16:46
DocScrutinizerthough the mayor for sure isn't allowed to "own" morphine16:47
DocScrutinizergenerally to get sued for posession of sth, you need to have *access* to it16:47
wpwrakyou can start a lawsuit and ask the court to order a search. no problem.16:47
DocScrutinizerlet's assume there'd be a technology that doesn't allow to unzip/untar the tapes prior to e.g 2100. Do you think anybody could ask for deletion of the tapes for the mere possibility there's some material on them?16:50
DocScrutinizerlet's assume I got two chemicals that - when mixed - react to give meth after 50 years. Would this be illegal?16:51
DocScrutinizereven when I mix them, when could somebody reasonably start a lawsiut against me for it?16:52
DocScrutinizerin 1 year, in 25 years? immediately? what if I destroy the whole set after 49 years?16:53
wpwrak(meth) probably not16:54
wpwrak(copyright) who knows ?16:54
DocScrutinizernah, each brick in my wall holds the encrypted latest (c) songs of $random_artist - I just had to find the right decryption method and I'd get sued for posessing the brick16:55
DocScrutinizeror could I only get sued for *processing* it?16:56
wpwrakin some places, you can even go to jail for not being able to provide the key ;-)16:56
DocScrutinizerI, I'm happy to share the key with $WORLD, it's simply "idiot!" - alas I have no clue about the *16:57
wpwrakyou'll discuss these fine philosophical points with your cellmates :)16:58
DocScrutinizercan I go home now? ;-D16:58
wpwrakand don't forget - never reach for the soap ! :)16:58
bartbesDocScrutinizer: god forbid you carry a piece of paper into the us with on it the decimal representation of a cryptographic key used for drm!16:58
DocScrutinizerthat's exactly what this GAA basically is about16:58
DocScrutinizerI know, I got a t-shirt. Will this work better? ;-P16:59
bartbeswpwrak: instead, claim that the soap contains a song16:59
larsci have all my data encryted in /dev/zero with xor. just need the right key ;)16:59
DocScrutinizerlarsc: \o/ :-P17:00
Action: DocScrutinizer considers to make another t-shirt with larsc 's nice statement17:00
Action: DocScrutinizer recalls the cop with t-shirt >>if you'd say "PLEASE SOMEBODY SHOOT THE PRESIDENT", I'd have to arrest you17:02
DocScrutinizeronly existed in my fantasy17:02
DocScrutinizeror - very simple - get a firewall in front of GAA that doesn't allow access from any US location - done17:04
DocScrutinizerChina and others are going mega-LAN rather than internet. The only problem with USA is they think they *own* the global internet17:05
DocScrutinizerhell, I bet some of them still think they own the globe17:08
DocScrutinizerwhile on the other hand their citizens not even own the hardware they paid for17:11
wpwrakthere's a tendency of going "inside". e.g., also facebook works that way.17:16
DocScrutinizer>tear down all churches immediately - they serve for some rogue individual to pin papers with illegal ideas written on them t the doors<17:18
Action: DocScrutinizer starts to wonder if he's going silly17:18
wpwraki think you know that the powers that be had a rather keen interest in arresting luther. if not, wikipedia is your friend ;-)17:19
DocScrutinizeror maybe it's some politicians going utterly silly17:19
DocScrutinizerwpwrak: what do you think why I made up this silly example17:20
wpwrakgood :)17:20
DocScrutinizerI don't recall having heard about a singe church or even church door getting problems due to that though17:21
DocScrutinizernowadays the common sense is "sue the door, make it reject pins"17:22
wpwrakthey probably took care of their internal undesirables discreetly back then17:22
DocScrutinizerooh, they do today as well, on all levels. But they extended their PR to accuse the door to be a "Mit-Stoerer"17:24
DocScrutinizeras the door did nothing to avoid that paper oinned at it17:24
DocScrutinizernext step: accuse the sun for providing light needed to read the paper17:25
wpwrakyup. they've learned. the "sudden unexpected death after quick and severe illness" doesn't work so well anymore these days. but you can neutralize your enemies just the same. don't even have to leave the office :)17:25
DocScrutinizerfinding somebody who "got raped" is a simple and rather inexpensive and effective way to neutralize enemies nowadays17:27
DocScrutinizerand there's just so much you can do against such weapons17:28
DocScrutinizerway smarter than Polonium ;-P17:29
wpwraksimple rule: never set foot into any fundamentalist abrahamic country17:29
DocScrutinizerbut what to do when the country where your foot is goes fundamental?17:30
wpwrakrun :)17:31
larscthey'll shoot your before you reach the border17:33
DocScrutinizerthat's what I always fail to get -what's the use in forcing to stay those who disagree?17:36
DocScrutinizerI'd see the pint in shooting those who stand up17:37
wpwrakkeep those from leaving who would if it was easy. that particularly includes well-educated people who could easily find work abroad.17:38
wpwrakanyway ...17:38
qi-bot[commit] Werner Almesberger: atusb/fw: revert most of the user_setups change and use simple callback instead http://qi-hw.com/p/ben-wpan/bbe27fc17:38
qi-bot[commit] Werner Almesberger: atusb/fw/: moved common elements of DFU processing to dfu_common.c http://qi-hw.com/p/ben-wpan/254400417:38
qi-bot[commit] Werner Almesberger: atusb/fw/usb/: DFU_ITF_DESCR also needs a protocol argument http://qi-hw.com/p/ben-wpan/5f5061117:38
qi-bot[commit] Werner Almesberger: atusb/fw/: added transition from runtime to DFU mode according to DFU spec http://qi-hw.com/p/ben-wpan/2d7ee9117:38
qi-bot[commit] Werner Almesberger: atusb/fw/: moved timer initialization to board_app.c as well http://qi-hw.com/p/ben-wpan/2acdaca17:38
qi-bot[commit] Werner Almesberger: atusb/fw/: make timer code use interrupts instead of polling http://qi-hw.com/p/ben-wpan/a69916d17:38
qi-bot[commit] Werner Almesberger: atusb/fw/usb/atu2.c: use implicit race-free "and" when writing to UDINT/UEINTX http://qi-hw.com/p/ben-wpan/f9681e517:38
qi-bot[commit] Werner Almesberger: atusb/fw/: changed USB stack to use interrupts instead of polling http://qi-hw.com/p/ben-wpan/d03beb217:38
qi-bot[commit] Werner Almesberger: atusb/fw/usb/: we were busy-looping on TXINI. Mask the interrupt when idle. http://qi-hw.com/p/ben-wpan/376236917:38
qi-bot[commit] Werner Almesberger: atusb/fw/boot.c: adjusted the delay loop and don't race with dfu-util http://qi-hw.com/p/ben-wpan/b42577b17:38
qi-bot[commit] Werner Almesberger: atusb/fw/atusb.c (main): sleep (idle mode) while waiting for interrupts http://qi-hw.com/p/ben-wpan/ce16a1617:38
wpwrakphew :)17:38
DocScrutinizer(well-educated) that's idiotic as those are the ones you have the most problems with when they stay17:40
DocScrutinizermaybe "better shoot them than deal with their rant when in exile"17:40
DocScrutinizeralso kinda silly17:41
Action: DocScrutinizer getting some coffee components for the long weekend17:41
DocScrutinizerwell, fundamentalism *is* idiotic17:42
wpwrakDocScrutinizer: (most problems) naw, think of engineers and such. many people may hate the system yet have too much to lose for open rebellion (or sabotage, etc.)17:43
DocScrutinizerif those dudes/folks would act smart, it'd be even way more scary17:43
Action: DocScrutinizer glares at USA17:44
Action: DocScrutinizer also glares at Hungary17:44
rjeffriesis wpwrak awake?17:57
rjeffrieslooks like maybe he is17:57
rjeffriesthis idea will take a few lines. pls. bear with m.17:58
rjeffriesGoal: be able to use mouse and external keyboard with Ben Nanonote.17:58
rjeffriesPossible to leverage cleunt USB on Ben as follows:17:58
rjeffriesa little board prolly AVR based that has Two USB Host ports. (moer on this later)18:00
rjeffriesone USB host port attaches to Nanonote18:00
rjeffriesthe 2nd USB host port goes to a USB hub so it supports 1x USB mouse and 1x USB keyboard18:01
rjeffriesthere is software on Ben that "understands"  a simple protocol that the external USBx2 board uses to18:02
rjeffriestell Ben it is plugged in, and that it has one or two USB devices typically a mouse and a keyboard18:02
rjeffriesfor keyboard, keystrokes come into the USB2x board, and are sent to the "driver" on  Ben18:03
rjeffrieslikewise for the mouse18:03
rjeffriesTHE END18:04
rjeffriesyes, compared to a future Nanonote with USB host this is the long way home18:04
rjeffriesand it is a hack and is Not Pretty18:05
rjeffriesbut i think it is doable, somewhat similar to Android ADB http://developer.android.com/guide/developing/tools/adb.html18:06
DocScrutinizerrjeffries: there are such contraptions, called USB-network or USB-sync-cable or whatever, usually used to link two windows desktop PCs via USB18:10
rjeffrieswpwrak those would connect two usb hosts together, correct?18:12
DocScrutinizerI'd like to ponder if I might go thru "make USB hostmode work" part 2.5 :-)18:12
rjeffrieswhat do you mean DocScrutinizer ??18:12
DocScrutinizerprobably it's not completely out of sight to use Ben's USB B connector to simulate a USB A host18:13
DocScrutinizergiven you use some additional gear, like a hub with upstream VBUS, or a Y-cable18:13
DocScrutinizermaybe some resistors18:13
rjeffriesDocScrutinizer to my mind, there are two use cases, but are Human Interface Devices: mouse and keyboard18:13
DocScrutinizerrjeffries: believe me, there's a lot of usecases for USB hostmode you'd not even think about18:14
DocScrutinizerI was amazed what users are going to do with USB hostmode when I enabled it on N900 ;-)18:15
rjeffrieslittle AVR computers are ridiculously cheap now. the PCB for this ultra tiny one is about $1 $SD. plus BOM and assembly18:15
rjeffriesdocScrutinizer remin me what SOC the N900 uses18:16
Action: wpwrak shall return after feeding18:16
DocScrutinizerrjeffries: yes, but why going this tedious awkward path of a micro-host with all the non-standard drivers etc pp, when you can mod the USB on NN to play host role? 18:17
DocScrutinizerrjeffries: OMAP3430, but that's mostly irellevant for USB hostmode (modulo the fact the OMAP3430 MUSB core made by mentorgraphics was a real bitch to convince to do hostmode)18:18
DocScrutinizeron NN I guess USB is via bitbang or a way simpler USB hw interface inside SOC, so it's even simpler to implement hostmode there than it was for OMAP18:20
DocScrutinizergeneral rule: the difference between USB gadget and USB host are 2 resistors, a 5V line, and another driver in software18:21
DocScrutinizeron the PHY protocol level, there's almost no difference between a host and a gadget, means the data lines remain the same and no hw patches needed18:24
DocScrutinizerthink of it like ethernet, where the host does also PoE18:25
DocScrutinizerthere's no real difference between your NIC in PC and the webcam you attach via ethernet18:26
DocScrutinizerexcept for the host may provide Power over Ethernet (which you can fake, like you'd do for USB hostmode), and with USB you also got different plugs on each end, which you can deal with by using an adapter18:28
DocScrutinizersee the posts about Y-cable and USB on openmoko(-kernel?) ML and wiki, and also on talk.maemo.org (http://talk.maemo.org/showthread.php?t=65232 and previous lengthy threads)18:30
whitequarkDocScrutinizer: duh? is the USB hostmode really so easy?18:39
DocScrutinizerwhat can I say?18:39
rjeffriesDocScrutinizer I have ZERO desire to add an external MCU. Ben useability would improve a LOT by having external keyboard and mouse, and doing that your way would be Most Excellent18:39
rjeffrieswhitequark: USB Host mode allows external (good) keyboard and a mpuse18:40
whitequarkrjeffries: I know what USB host may be used for :)18:40
whitequarkbut I've heard a lot of times that it is quite resource consuming to implement it correctly18:40
rjeffriesdo you question how valuable a more useable keybaord is?18:41
whitequarkrjeffries: not at all. I do question whether an USB host done entirely in software is possible at all18:41
whitequarkif it (somehow) is, then it would be incredibly good, as the device I'm working with also misses a hardware host and it is much worse without it18:42
DocScrutinizerso if there's really interest in supporting this, I'm willing to give it a try and have a look at schematics and drivers. The sw bits need to be done by somebody else though, I can only give advice what to do and what to "mod" on hw side (cable with 2 resistors or whatever we may need)18:42
kristianpaulmoment, what will happen with usb-boot with that hw mode?18:43
DocScrutinizerthe resources needed on sw side are not much different for host and gadget18:44
whitequarkDocScrutinizer: do you know any working examples of bitbanging host on any MCU?18:44
DocScrutinizerkristianpaul: what should happen when there's a modded cable in your drawer? ;-)18:44
kristianpauloh boy :-)18:45
DocScrutinizerI'd guess there are zillion of examples of bitbanged hostmode on MCU out there in the internets18:45
DocScrutinizerI bet wrt has such drivers, as there are plenty of routers that allow plugging in a USB stick for mass storage file server functionality18:47
DocScrutinizeron all those fritzbox and whatnot DSL-LAN-WLAN-roters18:47
larscbut all of these have hw host mode18:47
whitequarkyeah, most of them have EHCI18:48
DocScrutinizerwhat's hw hostmode, after all?18:48
larscwell, not sw18:48
kristianpauljust a tranceiver i guess18:48
DocScrutinizerwell, for sure they don't all have a MUSB-HDRC core on their SOC18:49
whitequarkDocScrutinizer: I always thought it's just several FIFO's, an NRZI encoder and a state machine for packet handling, right?18:49
whitequarkall these things are quite costly when implemented purely in software18:49
whitequarkand via bitbanging18:49
DocScrutinizerwell, whatever they are implemented - they are IDENTICAL for host and gadget18:50
larscyou might do this via an coprocessor but certainly not on the main cpu18:50
DocScrutinizeryou'te alrready doing it on Ben, whateve way18:50
DocScrutinizerthe USB data transmission is symmetrical bidir18:51
DocScrutinizerhost/gadget is decided on a way higher usually sw level18:51
whitequarkDocScrutinizer: I've found just 1 (one) bitbanging implementation. it uses four Parallax Propeller chip cores overclocked to 96MHz and working simultaneously. I don't really think that Ben's cpu can handle this load18:51
DocScrutinizerso it doesn't matter if Ben is doing bitbanging, or has hw FIFOs or whatver, we'll simply re-use them18:52
larscwhitequark: there is a usb stack for avrs18:52
larscDocScrutinizer: no18:52
whitequarklarsc: V-USB? it's device-only18:52
larscit's not that easy18:52
larscwhitequark: ah ok18:52
whitequarklarsc: I meant "just one host implementation"18:52
larscDocScrutinizer: because you can't turn the usb gadget ip core into an usb host ip core18:53
DocScrutinizerfor all I know there's ot that much difference in PHY and 2nd layer, between host and gadget18:54
whitequarkDocScrutinizer: ahem. if it's that simple, then why no one has implemented this? an ability of using hw usb gadget as hw usb host is definitely a killer feature. but can you point me to any Linux implementation of host-via-gadget?18:54
DocScrutinizertzzz, so you got another friggin IP core there?18:54
DocScrutinizerwhitequark: I'n not arguing on such level. I offered to look into schematics, datasheets, and whatnot. If you like to give up on a "hasn't been done yet" assumption, OK. I'm out18:55
whitequarkDocScrutinizer: okay, dump that message. I'll get a better explanation then18:56
DocScrutinizerI made hostmode for N900 work, though *Nokia* told 100 times "It can not work, will never work"18:57
larscbut the ip core actually supports host mode18:57
larscthe jz4740 gadget ip core supports only device mode18:58
DocScrutinizerI feel this discussion is a bit discouraging, and I have no interests in Ben (not even own one), so you're free to implement the micro-host adapter approach, it's known to work18:58
DocScrutinizerbtw oops, the connect-2-PCs adapter obviously is the othr way round, you'll need the "male" variant when that adapter is "female"19:01
wpwrakhmm, not sure if some facts would be appropriate for this discussion ... :)19:02
DocScrutinizerwhatever, I wonder what makes you think the MCU on such an adapter is better siuted to do the hostmode job than Ben's SOC19:02
Action: DocScrutinizer clears a topic on his todo list19:03
whitequarkwpwrak: may we hear them?19:05
wpwrak1) almost everyone who implement usb host or device uses hardware support (NRZ, bit-stuffing, CRC, usually up to and including automatic ack, flow control, EP demultiplexing, etc.)19:05
wpwrak2) the one exception i know of is v-usb, which provides a low-speed device.19:06
wpwrak3) it's correct that at the PHY, device and host are very similar.19:06
wpwrak4) at the higher protocol layers, they differ. you still see that they're related, but USB is an symmetric protocol, so their roles are not easily reversed. e.g., any transfer is always initiated by the host. a device that has data to send waits until the host asks for it.19:08
wpwrak5) the jz4720 has a host controller and a device controller. unfortunately, they use separate pins, and the pins of the host controller are hidden under that black epoxy or whatever splotch.19:09
wpwrak6) i see no reason why you couldn't implement usb host in the v-usb way, i.e., via GPIOs. you may violate the standard by not getting all the timings right, but only few devices may notice that.19:11
wpwrak7) whether you can use the D+/D- lines of an existing USB device implementation depends on whether you can disable the pull-up and on whether you can properly bit-bang them (patterns and speed)19:13
whitequarkwpwrak: hm. you only need 1.5MHz to plug in a HID device. that's probably less than atben requires19:19
whitequarkbut on 4725B (and 4740 has the same USB gadget), USB core has dedicated pins, which are not connected to any GPIOs, so you'll have to use other pins anyway in GPIO scenario19:21
wpwraksome USB controllers have the option of treating D+/D- more or less like GPIOs (for testing). but i'm not sure the 4720/etc. is among them. so far, it doesn't look good19:22
whitequarkwpwrak: can you remind me, how long it takes for jz47xx core to set a gpio?19:23
wpwrakthe "speed of light" is about 12 MHz19:26
wpwrak(read the description of the device controller) doesn't look good. there are test modes but they don't do what you need. you're better off just bit-banging GPIOs.19:27
whitequarkso 1.5MHz should be more than doable within software19:27
wpwrakDocScrutinizer: so, get ben+ubb and hack away ! ;-)19:27
wpwrakwhitequark: yes19:28
wpwrakwhitequark: compared to the AVRs, you also have the advantage that your core runs a lot faster. so you have more time for bit-level things like NRZ and CRC19:29
whitequarkbut I see another problem: unlike atben, which uses SPI (am I right?) where you can just switch the clock with any scheduler-altered timings you like, in USB you has to do the switches and readings in a very timing-precise way19:29
whitequarkI wonder if the stock (not rt) linux scheduler can handle that19:29
wpwrakwhitequark: one problem is the LCD. if you don't turn it off, then you can't access memory. not sure if you can manage to run in cache without ill effects. the cache has a few special features that may help.19:30
DocScrutinizerwpwrak: not on my todo list19:30
whitequarkwpwrak: huh, running from cache. that's way more MIPS magic than I really understand :/19:31
DocScrutinizerI'd think all USB cores allow the H and L trsts for D+/-19:31
wpwrakwhitequark: you would do this like i did ubb-vga: turn off everything that could get in the way, then write compact code to bang the bits just at the right time19:31
DocScrutinizeralso setting either line to H or L19:31
DocScrutinizerI'd need to eat the SOC's datasheet which I'm not going to do as I lost interest in the whole topic19:32
wpwrakDocScrutinizer: the 4720 has high-speed J and K (and Z) output, plus two higher-level tests (NAK and test packet)19:32
wpwrakDocScrutinizer: there's no input, though19:32
DocScrutinizeraaah J and K, not H and L19:32
whitequarkwpwrak: what's the point of connecting HID devices without an LCD?19:33
wpwrakwhitequark: well, you could just disable the LCD, get a quiet system, get your code to work, then see how you can make it survive concurrent system activity19:34
DocScrutinizerUSB ransmission isn't usually a 100% of time job19:34
wpwrakDocScrutinizer: and being host-driven works in your favour there19:35
DocScrutinizeryou may get away with stopping LCD for transmitting a packet over USB19:35
DocScrutinizerwpwrak: indeed19:35
DocScrutinizersimplifies things a lot19:35
DocScrutinizermy NASCOM-I also couldn't execute a single opcode on the CPU during display of a video line. Either had to live with the Hsync/Vsync timeslices, or you switched off video output19:37
DocScrutinizeriirc, or was that ZX81?19:38
wpwrakthe ZX81 had this behaviour SLOW and FAST mode19:38
DocScrutinizerprolly ZX81, with the "fast" and "slow" cmd19:38
DocScrutinizerstill ZX81 could do supertape ;-D19:39
DocScrutinizerok, USB is a tad faster19:40
DocScrutinizerwpwrak: you're sure there's no diagnostics register here you can read back?19:44
DocScrutinizerdoesn't make much sense to have J and K then, I guess19:44
wpwrakthe USB standard requires that a device can enter those test modes on the host's request. seems that it's the host that does all the reading19:46
DocScrutinizerso what's that friggin IP core?19:51
DocScrutinizerplease don't say Mentor ;-)19:51
wpwrakhow do you tell ?19:52
DocScrutinizerhonestly, what's the driver.ko's source header saying?19:54
wpwraknot sure if you understood my question. how do you find out what the origin of the IP core is ? except by looking at the data sheet, which i know you can't do19:54
wpwrak(looking at the data sheet and looking for a familiar pattern)19:55
wpwrak(if it just said "mentor", i would have told you :)19:55
DocScrutinizeruhm, lost me19:55
whooohi, have anyone been able boot with an initramfs? or rather, to which address should I load it?21:31
rjeffrieswpwrak one reason to find a way to leverage Ben USB cleint port as opposed to something on UBB 8"10 port is one wouold like to have atben radiop at same time as an external USB keyboard21:52
rjeffriesso did someone say SIE may awaken, sans FPGA? that could be interesting in a hackerish way21:53
rjeffriesif and only if it supports linux21:54
DocScrutinizerwpwrak: btw if I have had a look at Ben prior to "MP" then you bet we had those host pins on 2 pads next to the 0R bridges that connect USB receptacle to gadget pins21:55
DocScrutinizerunless somebody would have come up with "We don't need that" again and started arguing about the cost of 2pcs 0R21:56
Action: DocScrutinizer still wonders if wikireader still has footprint for a USB, or they "optimized" it out... "don't need that"21:59
DocScrutinizerwpwrak: watch out they don't optimize away your ext-ant footprint ;-)22:00
DocScrutinizer(bridges) might even have gotten footprint to get a twin-2way mux chip there22:04
wpwrakDocScrutinizer: (missing usb host signals) yeah, that's one of the sadder parts of the ben's hardware. i don't know if this was just overlooked or if there was a reason (such as lack of space)22:18
DocScrutinizerlack of space? c'mon22:25
DocScrutinizerI never bought the "lack of space" foo even for our GTA02 mods that got rejected using this lame argument22:26
DocScrutinizeralways got parsed at my side as "we don't have enough space unless we would dare to touch the routing that we put so many days of work in"22:28
wpwrakwell, i don't know the details of COB. if you say it can't create space issues, then they probably just overlooked it for no good reason22:29
whitequarkDocScrutinizer: aren't complex things like openmoko or other phones routed automatically?22:30
DocScrutinizerremember A8? A7 even? No problem at all to move all the birdseed away from mic into the codec can. But when I asked for two testpoints: "Not enough space"22:30
DocScrutinizerwhitequark: sure, like 80%22:31
DocScrutinizerstill some days of work, but once you got the basic layout, small mods are no problem at all22:31
DocScrutinizerthose router tools are pretty smart and mighty tools22:32
Action: whitequark has routed all his boards manually. they were quite simple, through22:32
DocScrutinizerdunno ho wmuch of that power is usually really used by the average chinese layouter22:33
DocScrutinizere.g we got zero design rule checks on GTA0x22:34
DocScrutinizerpretty bad22:34
wpwrakDocScrutinizer: you don't know half of what is possible on the darker side of chinese engineering ;-)22:34
DocScrutinizersure it takes a bit longer to define DRC, but once they are defined, your duty becomes a lot easier22:35
DocScrutinizerwpwrak: I bet22:35
DocScrutinizerI'm even aware no 10 years at Taipei would suffice to learn all about that22:35
tuxbrain-mobwpwrak : what was the command to test the timings on the host?22:36
DocScrutinizerwpwrak: I just say "8*8 LED matrix with 64 series R"22:36
wpwrakDocScrutinizer: in hxd8, the folks from the US were complaining that someone had adjusted a bit of the layout by moving it ("drag and drop") to the desired location. what the didn't do was make sure things still connected where they should elsewhere and didn't make new connections where there should be none. naturally, no DRC was run. all those errors would have been just too confusing. and there goes another failed prototype run ...22:37
wpwrakDocScrutinizer: i think that was done by the fine folks fiwin had outsourced the layout to (before fiwin were killed and the project was merged into openmoko). so we never saw that sort of stuff at openmoko.22:38
wpwraktuxbrain-mob: atrf-xtal 10000   ?22:39
wpwrakDocScrutinizer: (8x8/64) that could be correct ... the real problem there was the absence of any specification of what that array should do. with the guy who "owned" the design refusing to explain. i don't blame the engineers for anything that happened there.22:41
wpwraktuxbrain-mob: regarding the troublesome gpio test: there are two similar tests, one in P_ON state and the other in TRX_OFF. the one in TRX_OFF should be more predictable, but can test fewer things.22:43
wpwraktuxbrain-mob: that's why i put most of the testing in the P_ON block. not that this one doesn't seem to work reliably, i'd propose the following:22:43
wpwraktuxbrain-mob: 1) we ignore those problems in P_ON for now and i'll skip this test in the "atusb" script.22:44
wpwraktuxbrain-mob: 2) i'll go over the TRX_OFF test and see if i can test more things, so that we're sure we don't have issues like shorted pins22:44
wpwraktuxbrain-mob: 3) when i get my boards, i'll have a look at the P_ON test and see if i can figure out what went wrong there, and adapt the test if that makes sense.22:46
DocScrutinizerwpwrak: (with the guy owning design) LOL, indeed22:47
wpwraktuxbrain-mob: i.e., the failure probably just indicates me misunderstanding some property of the system that's not relevant for operation.22:47
tuxbrain-mobwpwrak just tellme what to do and i will22:48
wpwraktuxbrain-mob: for now, just don't run the test on all the boards yet, because i'll probably make a few small changes22:48
wpwraktuxbrain-mob: meanwhile, you can set up the spectrum test and see if the sugru works22:49
tuxbrain-mobremember i will start the mass test tomorrow, no sugru here now :( , sorry for the push but I'm really scheduleballsclamped22:51
tuxbrain-mobif i do not test this weekend i don't know when i will have another big time window aviable to retake it22:53
wpwraktuxbrain-mob: tomorrow = monday ? :) that's plenty of time ;-)22:53
tuxbrain-mobok not tomorrow, today after a few hours of sleep22:54
wpwraktuxbrain-mob: ah, and there's new firmware. i also changed the boot loader. lots of things have changed inside: it now uses interrupts instead of polling, i fixed a number of small races in the USB driver, and - most importantly - i extended the boot loader delay from ~2.0 s to 2.5 s.22:55
tuxbrain-mobwell i have not reflashed anymore but the first one so no problem22:56
tuxbrain-mobjust sendme the links by email pleease22:57
wpwrakthe current boot loader delay could become a problem when using dfu-util without prior reset. perhaps you remember the discussion with Stefan ? i've implemented that mode now. the old boot loader usually worked, but that was more by luck than by design (probably because some other changes made the loop take a little longer than 2.00 seconds)23:00
wpwraki'll play with things a little more and then make the "final" firmware images and announce them on the list23:00
wpwrakah, but in case you want to play ... lemme upload them ...23:05
tuxbrain-mobok I hope this final one comes along the week i will reconnect with you tomorrow when i have a bunch of atben atusb depanelized and ready to flash to know if the firmware is also ready23:05
tuxbrain-mobweek  weekend23:06
wpwrakcurrently the latest versions: http://downloads.qi-hardware.com/people/werner/wpan/bindist/boot-ce16a16.hex23:08
wpwrak126a8db91e06180dff4676fdb60b26fe  boot.hex23:08
wpwrak2e43af32a0002cb78ddd794dfe7ece11  atusb.bin23:08
DocScrutinizerwpwrak: the 64 was a joke, no? You're aware you need series Rs only in either the col or row lines23:08
wpwrakatrf-id: #158 Sat Jun 11 13:05:42 ART 201123:08
wpwrakDocScrutinizer: only if you really multiplex :)23:09
wpwrakDocScrutinizer: if you're making a fancy flashlight, you may need one R per D ;-)23:10
DocScrutinizererrr, yup23:12
tuxbrain-mobmmm  i think my toshiba will have event wort timing results than the efika :(23:12
tuxbrain-mobfrom +36.65 ppm to +99.99 ppm23:13
tuxbrain-moball diferent values avobe where the min max23:14
wpwraktuxbrain-mob: hmm. remove anything non-essential that shows up on top.23:15
wpwraktuxbrain-mob: also, killall npviewer.bin nspluginviewer23:15
wpwraktuxbrain-mob: stop music players, video, etc. don't load web pages or move windows while testing23:15
tuxbrain-mobI have killed all  X to obtain this results, just the console 23:16
wpwraktuxbrain-mob: if this still doesn't help, i can add some more filtering to atrf-xtal23:16
wpwrakokay, that's bad then23:16
DocScrutinizerwpwrak: though there are brilliant dudes that need only one R for 36 diodes :-o operated from a 3.6V LiIon23:16
wpwrakyou can try collecting more samples. but that will also make the test slower23:17
wpwrakor defer the test until you're back at a proper pc :)23:17
wpwrakyour laptop probably doesn't have a cleanly ntp-disciplined clock either now, so the measurements would be questionable at best23:18
wpwrakDocScrutinizer: oh, if all the LEDs are on all the time, that ought to work, too. that is, unless something goes wrong. after that, tepco :)23:19
DocScrutinizerhehe, yeah23:19
tuxbrain-mobntp is installed will the 3g connection affect this ntp thing?23:20
Action: DocScrutinizer just adopted tepco for a generic descriptive term of a class of events23:20
wpwraktuxbrain-mob: probably. i don't know how much, but i can't imagine ntpd will stabilize very quickly in this scenario.23:20
whitequarkDocScrutinizer: will that really work as intended? all LEDs have a bit different VAC, AFAIK, and the ones with lowest forward voltage will be brightest (and hottest), and all others will be quite dim23:21
DocScrutinizerI think it's BS design23:22
wpwrakDocScrutinizer: the russians will be angry. now nobody honors the skills demonstrated at chernobyl anymore.23:22
whitequarkwpwrak: "tepco" as adjective is brilliant!23:22
whitequarkDocScrutinizer: what's BS? that abbreviation may have a dozen of meanings23:23
DocScrutinizerBS == tepco here23:24
DocScrutinizerhttp://talk.maemo.org/showpost.php?p=872446&postcount=3 this one23:25
DocScrutinizer35 diodes, 10R 1W series, 7.2V(sorry)23:27
whitequarkno one needs step-down when you can just put a resistor23:29
wpwraktuxbrain-mob: if you want to experiment a bit, you can try to set up dirtpan. if you generate a lot of traffic, your connections will eventually hang, because there's no CSMA and you just drive up TCP timeouts. but it should be good for a few tests.23:31
DocScrutinizerI'd rather use step-up here, to something like 30V, then use 3..4 chains with one series R each23:31
wpwraktuxbrain-mob: ah, and in the commercial description, the available flash size will go down a little. all the code improvements ate a bit of space. i'll update this in a bit as well.23:32
whitequarkDocScrutinizer: is that for minimizing the DC-DC converter current?23:32
DocScrutinizeror use unlimited step-up regulated vie current-feedback, one chain, no series R23:32
DocScrutinizerthat's for minimizing the loss on mandatory series R23:33
DocScrutinizerthere actually are "designs" that come with no (visible) series R at all, but that's really poor stuff23:34
DocScrutinizerfrequently keychain crap with one CR2032 and one LED is of that class23:35
wpwrakthe battery has an internal R, so ... :)23:36
DocScrutinizeryou're exploiting inner parasitary ESR23:36
wpwrakjust because it's parasitic doesn;t mean it's not there ;-)23:36
DocScrutinizerwell, it works23:37
DocScrutinizerfor a 1300mAh 7.2V LiIon pack you run into trouble with that concept23:38
wpwrakmake it pocketable. instant birth control ;-)23:38
DocScrutinizeryou need a series R for each chain of series LEDs23:38
DocScrutinizerthe one shown at that picture? definitely!23:39
whitequarkhow hot it is?23:39
DocScrutinizerI don't need to repost those other links?23:39
DocScrutinizer(I probably not even have them anymore)23:40
DocScrutinizersmall version of the balcony demolition23:40
DocScrutinizeror your trousers look like that airmail package23:41
whitequarkat least it won't be cold in winter then23:42
tuxbrain-mobgood night dudes23:45
wpwrakwhitequark: two boiled eggs and some crisp fried ham make a hearty breakfast :)23:48
--- Sun Jun 12 201100:00

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