#qi-hardware IRC log for Thursday, 2012-04-12

wolfspraulok, later then00:02
wolfspraulno rush on this anyway00:02
wolfspraulto me, very superficially, metric & one file looks easier00:02
wolfspraulbut I will just change to what you want00:02
wolfspraulroh: hi morning :-)00:18
wolfspraulsorry I missed you yesterday but actually wanted to ask: have you ever experimented with any form of epoxy polymer to seal/enclose electronics?00:18
wolfspraulI'm thinking about my atben/atusb and how I can make them more rugged00:18
wolfsprauljust want to dip them in some epoxy blob, or even superglue if that would work :-)00:19
rohwolfspraul: hey00:31
rohnope... epoxy is quite nasty, heavy and one cannot repair stuff when glued anymore00:31
rohwell.. i used it to glue stuff, but not as case00:31
wolfspraulhow heavy?00:33
wolfspraulwhich specific epoxies did you try with?00:33
wolfspraulabout 'no repair', sure, I understand00:33
wolfspraulif the case is really good, maybe no repair needed :-)00:34
wolfsprauljust kidding, but the one-way aspect of epoxy of course is clear00:35
wolfspraulsame as first-level IC packaging or ultrasonic welding00:35
wolfspraulfor my atben & atusb, I'm sure i just want to seal them and done, don't know what I would ever want to touch on the atben/atusb board00:40
wolfspraulcladamw: good morning :-)00:55
wolfspraulI saw that xiangfu was also starting to edit the board design now, great00:56
wolfspraulthat gives me hope we can finish faster...00:56
cladamwyeah ... pipelines in qi. :-)00:57
rohwolfspraul: kinds... mostly 2-component glue stuff00:58
rohi havent used moldable stuff like 2-component PU or similars00:59
rohits on my list of stuff to try00:59
wolfspraulyeah, that's what I was getting at00:59
rohi was thinking.. fuck cases... lets use whats there00:59
wolfspraulI remember years ago Tully gave me a dvd from some US epoxy/resin manufacturer, that was very impressive to watch00:59
wolfspraulunfortunately I forgot the company name00:59
rohjust use something.. exact form doesnt matter, as long as we can buy it and know how it works. its only mechanics. we can design our own when we do 5-digit numbers01:00
rohmay look a bit 80s, but whats with teko and boppla cases or so?01:00
wolfspraulI agree conceptually, but when you start you still run into things that are really important to you01:00
wolfsprauloh absolutely01:00
rohor are you really thinking nanonote-style?01:00
wolfspraulI am 100% with you01:00
wolfspraulabsolutely not01:00
wolfspraulI think just what is practical and what works, in *our* priority scale01:01
wolfspraulfor example the current atben/atusb boards are a pain, so fragile!01:01
rohi mean... i see droid devices flying by for sub 100E now01:01
wolfspraulbut I have a big tube of cyanoacrylate right here :-)01:01
rohanother thing i think we should tackle01:01
rohthe 'eco'-market01:01
wolfspraulsure, of course [eco]01:02
wolfspraulone sec, the prices don't surprise me at all01:02
rohopensource devices are the only real ecological things.. because you know what they contain (more than usual) and have a better repairability.. etc.01:02
wolfspraulan android wifi phone costs about 20-30 USD now01:02
wolfsprauland it will continue to go down01:02
rohthat may help getting contrast to the cheap stuff in the droid area01:02
wolfspraulthe advancement is in the chips, the last 60 years, no?01:02
wolfspraulyou can essentially think of every IC to be a 1 USD item01:03
rohwolfspraul: sure. but why for example do you have a new keyboard?01:03
wolfspraul(gracefully skipping over billions of investments)01:03
wolfspraulthat's a big mechanical piece customized to fit my hands01:03
roha ibm buckling spring from the 80s still would be 'recent developed' enough. nothing happened since then (besides usb, but there is adapters)01:03
rohthe 'cheap materials' area is about to end. i know we will dig up garbage to mine for gold01:04
wolfspraulthere are different price trajectories01:04
rohone ton of mobile-phones has more gold than the ore from a goldmine.01:04
wolfspraulchips will continue to become more powerful while still costing "a dollar" in raw materials01:04
rohso lets make sure we know what we used, where and make modules where it makes sense (also to gain upgrade-pathes on hardware)01:05
wolfspraulI can see them getting a few thousand times more powerful still01:05
rohe.g. i could imagine a computer with a mm-like soc, consisting out of multiple chips on multiple boards plugged into one frame as case01:05
rohmay be some need 3 or 4 fpga in the end, maybe some can use one or 2?01:05
wolfspraulthe cost of raw materials will generally increase I agree, but not explosively I think01:05
wolfsprauljust a few percent each year, inflation01:06
rohwolfspraul: more than explosivly... just see the prices of rare earth skyrocketing.01:06
wolfspraulnah, I think most of that is propaganda and lobbying01:06
rohwolfspraul: well.. china has export limits on that now (new)01:07
wolfspraulthere are lots of discontinued mines all over the world, where people just got too lazy to compete with the chinese01:07
wolfspraulyeah finally01:07
wolfspraulthe chinese are destroying their environment and poisoning people in nearby villages01:07
wolfsprauland the government cannot keep the activities of greedy local miners under control01:07
rohand when the wto mumbled, the chinese argued with protection of the environment.01:07
wolfspraulif I would be in the chinese govt, we would cut down harder against this :-)01:07
rohsome people who did the arguing there must been laughing a day long01:07
wolfspraulbut there are plenty of mines in australia and many other places that could be reopened01:08
wolfspraulwhere workers would be paid australian rates, and australian environmental standards would apply01:08
wolfspraulwhich is a good thing!01:08
wolfspraulthe chinese govt is right on this!01:08
rohmaybe. still limited ressources. means in the end we need to stop using, and start making whole cycles out of it.01:08
wolfspraulchina is a big big country, and there are very greedy local miners that don't care that the entire village nearby their mine is poisoned01:08
wolfspraulall kids loosing their teeth etc.01:09
wolfspraulwhile their own kids (of the factory owners) go to expensive private schools in shanghai01:09
rohwolfspraul: also china has extremely bad performance when it comes to economy of such things.01:09
wolfsprauland the australian mines could easily produce the rare earths as well, they just cannot compete with such gross violations of labor and environmental standards01:09
rohthere are clean processes which do not dirty everything and get more ressources out in the end.01:09
wolfspraulsomeone has to stop this, and it's good that this (seems) to be happening now01:09
wolfspraulit's not "china", it's individual people that are responsible01:10
wolfspraulwhat I try to tell you is that the problem are greedy local miners, who put their own profits above everything else01:10
rohi know. actually i think the chinese planers are much more considerate about the future than our western economies01:11
wolfspraulbut for electronics, the fact is that the amount of rare/expensive raw materials is very small01:11
wolfspraulwhat do all raw materials in the typical smartphone add up to?01:12
wolfspraulif you just look at weight and cost of raw materials per kg/ton ?01:12
wolfspraulmust be a few USD at most01:12
wolfsprauleven less than a dollar maybe :-)01:12
wolfspraulok let's say 5 USD - MAX01:13
rohits availability, not only the per piece price which counts afaik01:13
wolfspraulcompared to the service revenues being driven out of this device, it's negligible01:13
rohsame as for weird chips. can kill your whole product if not avail in time01:13
wolfspraulthere is no shortage in rare metals, just a totally lopsided mining environment01:13
wolfspraulthe chinese got 9x% market share because they were willing to ignore labor and environmental standards01:14
wolfspraul'the chinese' being individual mine owners01:14
wolfspraulI hope this stops soon. I certainly don't want any such mine anywhere near where I live.01:14
wolfspraulwhat were you thinking with 'take any case'?01:15
wolfspraulI agree conceptually, but when you start you run into major inconveniences that will still make you want to customize this or that aspect of the case. that's what I found.01:15
rohwell.. depends on what the target usecase is.01:15
rohfor the nanonote its hard01:16
wolfspraulso I think it's more like "make case v1 asap" with least cost and resources01:16
wolfspraulthen use it, run into the major daily issues01:16
wolfspraulthen proceed to case v201:16
rohbut what hinders us to use some readymade thing like the ben case?01:16
rohcases are nearly always designed with a specific production method in mind for every part01:17
wolfsprauland hopefully someone has the (daily) usage routine in mind as well01:17
rohso using one concept on low volume, and the switch is basically impossible01:17
rohone would rather learn from one case and the manufacturing method, and try others over time, depending on scale, needed detail, strength, volume and on how many01:18
wolfspraulthat's what we do, no?01:19
rohi think the most realistic best mechanics for the buck we can get is what ben has01:19
rohanything we do would be more expensive and bulkier by far01:19
rohso maybe the nn isnt the best device to experiment with new case making stuff and learn?01:22
wolfspraulhmm, hard to say01:23
wolfspraulwe learn, that's the good part01:23
wolfspraulcan't really pin down how we remix what we learnt into something really good yet01:23
wolfspraulfor sure the plastic injection path with expensive steel tools has lots of pros and cons, which we know now, so we can use that when needed...01:24
rohsure. but i have good experiences with only changing few things from revision to next, to make it work at all01:24
wpwrakphew. back.01:25
rohso.. new soc AND new case.. may be too hard?01:25
wolfspraulit comes down to experience and picking the right tool for the job, and designing the case (and even use case) with those things in mind from day 101:25
rohi think we have no chance even replicating what the nn has now in mechanical keyboard quality with a non-injection molding manuf. process01:26
wpwrak(metric/one file) without switch, metric is better, because there's a clean conversion imperial -> metric but not metric -> imperial01:26
rohmaybe we can, but not for sane money.01:26
wpwrak(one file) probably doesn't matter much as far as pos2fab is concerned01:26
wpwrak(atben/atusb) how about acid-free silicone ?01:27
wpwrak(keyboard) do it the apple way. mill it keycaps from some fancy material, then sell it at a high price ;-)01:33
wpwraks/it k/the k01:33
rohwpwrak: we can't compete in any sane way doing something like that01:42
wolfspraulI think more incremental01:42
rohnobody buys keyboards for >100 euros01:42
xiangfudvdk, thanks on mplayer. works just fine. :)  I will release image today. test then release. 01:42
wolfspraulyes agreed, those are products that are part of a huge marketing and design world and experience01:43
wolfspraulit's like saying the candy sold in disneyland has such huge profit margins01:43
wolfspraulsure it does01:43
wolfspraulbut you try to go into disneyland with your own bag of candies and sell them there01:43
wolfspraulha ha01:43
wolfspraulsecurity people will be at your place within, what. 30 seconds? :-)01:43
wpwrakroh: sure ? :) http://www.trustedreviews.com/Art-Lebedev-Optimus-Maximus_Peripheral_review01:43
wolfspraulI think our path is excellent in making incremental experiments and adding more tools, processes and materials to our arsenal01:44
wolfspraulit just has to be documented well, reproducible, so later we can build upon that, refine parts, next iteration01:45
wpwrakroh: the one i'm typing on at the moment also cost something like USD 250. of course, it'll probably outlive me :)01:45
wpwrakwolfspraul: yeah, we'll eventually have to crack the injection molding nut01:46
wpwrakwolfspraul: for now, i'd be happy with STLs and functioning mechanical prototypes. then someone who knows the process details can re-capture the critter for the plastic fab.01:47
wolfspraulit's not a big black box01:48
wpwrakand you could probably print STLs of you care to01:48
wolfspraulso that just won't happen01:48
wpwraks/of /if /01:48
wpwrakwhat won't happen ?01:48
wolfspraulbut maybe someone starts experimenting with plastic and manual moulding01:48
wolfspraulif you really want to do plastics01:48
wpwrakfor manual molding, i see some potential in heating the whole system, with the mold. they probably don't do that in industry, because then they'd have to wait for it to cool. but if we don't care about time, we could test fairly accurate molds this way01:50
wolfsprauleverything is heated afaik01:50
wolfspraulthe 'industry' (which is also just individual companies and people) optimizes as to their own needs01:50
wolfspraulthere is no big 'strategizing' going on, really01:51
wolfspraulyou know the optimization potential right when you have it in front of you01:51
wolfspraulroh sits at his workbench, and spends hours and hours doing some step X01:51
wolfspraul(which is already profitable, assuming that he works for a paying customer)01:51
wpwrakwell, a similar set of goals and capabilities often leads to similar solutions01:51
wolfspraulwhile doing this manual repetitive thing, he calculates and says "if I invest 5 hours into making this tool, I can then make back those 5 hours and save another 3 hours on top compared to continuing with the current process"01:52
wolfspraulthose 3 saved hours increase his hourly rate01:52
wolfspraulso he may take the risk (the tool making may still fail)01:52
wolfspraulthere is a nearly infinite variety of optimizations in mechanical01:52
dvdkxiangfu: cool.  01:53
wolfspraulI just want to learn more, and I think as I've said initially I want to try with some epoxies :-)01:53
wolfsprauldvdk: is the ssl problem gone now? (just double-checking)01:54
wolfspraulthanks a lot for the heads up! that deadline somehow bypassed my attention01:54
dvdkwolfspraul: seems to work here (sslpatrol tells me about new certificate, but no info, no warning nor error)01:54
wolfspraulroh: one detail. a while ago we talked about the heads of the screws on the m1 case01:56
wolfsprauland I kept saying I prefered phillips over the hex keys you seemed to like better01:56
wolfspraulwell, after some time I have to say unless you changed again, I am now in the hex camp as well :-)01:56
wolfspraulI rather explain people that they have to get this relatively rare driver, because the phillips heads just wear out too fast01:56
dvdkwolfspraul: maybe you shouldn't use HTTPS based CSS references on the non-HTTPS entry page to the shop?  This way it'd fail less fatally on the next cert timeout.01:57
wolfspraulnah, if anything I will move everything behind ssl01:57
wolfspraulon all sites01:57
wolfspraulmakes the life of the various traffic shapers and deep packet inspectors and friends a bit harder01:57
dvdkwolfspraul: that's a great idea I think.  With all the recent deep packet inspection, java-script injecting proxies etc. that can subtly break pages01:57
wolfspraula new term I heard there was "revenue extractor"01:58
wolfspraulwhich are people that are inserting extra ads into html streams! :-)01:58
wolfspraulI know ssl is no ultimate solution, at all. but it ups the ante a bit.01:58
wpwrakthe "revenue extractor" is pretty evil. well, if it sounds like a pickpocket who went to school, what can you expect ? :)01:59
wolfspraulyeah, nice term, eh?01:59
wolfspraulbut I can see how that sells01:59
wolfspraul"do you want to install our revenue extractor at your hotel/coffee shop/whatever?"02:00
wolfspraulwho would say no??? :-)02:00
wolfspraulwe have to be very clear that 99% or more people really have no understanding at all how this internet and protocol and computer stuff works, it just does work. so when someone comes with a message like that, it really *is* hard to say no.02:01
wolfspraulyou don't want to be the last idiot refusing some good new thing... and you cannot tell whether there is any problem with this or not.02:01
wolfspraulthat's why I rather see the responsibility on my side for example in moving everything behind ssl.02:02
wpwrakand who would be able to tell it's the hotel's fault anyway02:02
wolfspraulI think the hotel staff (in this example) making the decision has no way to tell whether this is good or not.02:02
wolfspraulno chance, zero02:02
wolfspraulthe salesmen could use something like "this is like twitter, it will take off. get on it early before everybody else does. don't be stupid"02:03
wolfspraul"the other hotels are all installing it now"02:03
wolfspraulssl :-)02:04
wolfspraula small step02:04
wpwrakwell. let's hope the message makes the round02:10
wolfspraulwpwrak: so you think an acid-free silicone may be able to enclose atben/atusb without affecting rf performance?02:29
wolfspraulacid-free so that the board remains undamaged02:29
wpwrak(acid-free) yes. seems that the regular silicone is a bit acidic. though they still don't mind recommending it for metals. but not for electronics.02:40
wpwrak(rf performance) good question :-)02:40
wpwrakit will almost certainly affect rf performance. but at least there's hope it won't be overly bad because it's thin. but then, i don't really know02:41
wolfspraulok, got it02:41
wolfspraulmore experiments02:42
wpwrakyeah. pretty heavy RF experimenting02:45
wolfspraulwell, it could start easy as well, just by doing high-level performance tests and seeing whether something gets worse02:46
wolfspraulthat doesn't tell us whether the lower layers are struggling with more noise though (for example), at least not precisely02:46
wolfspraulbut have to start somewhere02:46
wpwraksure. if it's a total failure, that would be easy to detect02:46
wpwrakyou can actually measure the signal strength with atben/atusb alone02:47
wpwraksee "Spectrum" under http://downloads.qi-hardware.com/people/werner/wpan/prod/test.html#procedure02:48
wpwrakthis does a frequency sweep and measures what arrives on the other side02:49
wpwrakstill extremely coarse. nothing like these http://downloads.qi-hardware.com/people/werner/wpan/20110306/02:49
wpwrakyou could also try acrylic paint. that should be thinner than silicone. drawback: not nice to remove02:52
qi-bot[commit] Werner Almesberger: genex/: added support for multi-part components (master) http://qi-hw.com/p/eda-tools/0f0e73203:06
qi-bot[commit] Werner Almesberger: genex/sym2xps: also translate and pass on text fields (master) http://qi-hw.com/p/eda-tools/e52c3cf03:06
qi-bot[commit] Werner Almesberger: genex/sym2xps: also check that the unit parameter is present (master) http://qi-hw.com/p/eda-tools/3bbd0e703:06
qi-bot[commit] Werner Almesberger: genex/sym2xps: use @ as regexp delimiter, to accommodate symbols names with / (master) http://qi-hw.com/p/eda-tools/dcee39003:06
dvdkwolfspraul: I see you already made sharism.cc https-only?04:10
dvdkwolfspraul:  there are two side effects: the (totally outdated) blog is gone. BUT: the Developers link to qi-hw.com is also gone!04:10
wolfspraulforgot, when I work on a server I tend to move things behind ssl whenever I can04:10
wolfspraulI want to move the shop back to qi-hw anyway :-)04:11
wolfspraulbtw today I am having terrible trouble with the gfw (great firewall of china), so I may drop out a lot04:11
dvdkI mean, people might not buy if they can't find a reference to firmware upgrades and manual?04:11
wolfspraulsomething is acting up, they are probably testing more aggressive intrusion schemes04:11
wolfsprauldns interception, attacks against all sorts of protocols, ssl, ssh, non-standard ports, artificial delays, you name it, they try it :-)04:12
wolfsprauldvdk: yes I agree this all needs to be united better04:12
dvdkbtw, on the nanonote's shop page https://sharism.cc/shop/product_info.php?products_id=104:13
dvdkif i click on "please visit this products webpage.", i get here:04:13
wolfspraulhe he. you are right! spring cleanup needed... :-)04:14
dvdkand suddenly SSL is gone but the Developer link is back.  This link graph has too many (directed :) nodes04:14
dvdklooking forward to the shop being moved to the qi-hw wiki.  so we can edit the price at will :)04:15
dvdk(plus all the bots finally get a chance at increasing nanonote sales)04:16
wolfsprauldvdk: is the qi wiki down right now?04:25
whitequarkwolfspraul: SSL is the way to go. On my sites I use HSTS, that is, require an user agent to always use SSL, even when the user has typed http://04:48
qi-bot[commit] Xiangfu: nanonote files: we still need this gforth workaround (master) http://qi-hw.com/p/openwrt-packages/0b9b2b305:39
kyaki revisited linux usb gadget in RNDIS mode and Windows and it still doesn't work06:56
kyaki was hoping it would work in 3.206:56
kyaki suspect it might be not a software issue06:56
kyakis Ben's USB port is capable of RNDIS? i;m asking because not all usb gadgets work - for exmaple, mass storage doesn't work as well06:57
kyakmaybe it's still the software issue, but it must be fixed not in RNDIS driver (on Linux or Windows side), but in linux/drivers/usb/gadget/jz4740_udc.c ?07:04
kyakfrom what i read, it seems that dingoo has no problems being a mass storage device in Windows07:10
kyaki guess RNDIS would work as well07:10
kyakAzbuka owners claim that their device also connects fine with Windows07:13
whitequarkkyak: xz0032 has been verified to work with rndis08:23
whitequarkand it has the same 4740_udc gadget08:23
whitequarkkyak: what's in dmesg (on the host) and lsusb -vv ?08:24
qwebirc1900I'm trying to flash, but when on usbboot y try to run "nprog 2048 openwrt-xburst-qi_lb60-root.ubi 0 0 -n", it gives me an error08:29
whitequarkwhich one?08:31
qwebirc1900the nanonote08:32
qwebirc1900with the latest image (2012-04-09)08:34
whitequarkwhich error does it show?08:35
qwebirc1900not enough argument08:35
xiangfuI haven't release the 2012-04-09 yet.08:37
xiangfuit will release today. sorry I am slow. I found a small bug. so it will release in next few hours. 08:37
xiangfuqwebirc1900, can you paste all output from usbboot? and the command you using.08:38
qwebirc1900only the one from the error?08:38
qwebirc1900usbboot :> nprog 2048 openwrt-xburst-qi_lb60-root.ubi 0 0 -n  not enough argument.  Usage: nprog (1) (2) (3) (4) (5)  (1)start page number  (2)image file name  (3)device index number  (4)flash index number  (5) image type must be:  -n:no oob  -o:with oob no ecc  -e:with oob and ecc08:38
xiangfuwhat is the usbboot version?08:40
xiangfuqwebirc1900, ok. there is a bug on that one. 08:42
xiangfuI think it because there is space at the end of command. if I don't remember wrong. 08:42
xiangfuqwebirc1900, please update to recently xburst-tools.http://projects.qi-hardware.com/index.php/p/xburst-tools/downloads/77/08:42
xiangfuit have a new command "reset" for reboot nanonote from host.08:42
xiangfuand a lot of bug fixed.08:43
xiangfuthe latest version is 20110508:43
qwebirc1900ok, thanks08:43
qwebirc1900flashing now, thanks again. Is there a way of making a splash screen for gmenu2x?08:48
xiangfuyou mean the boot splash ?08:52
qwebirc1900yes, but  not the Qi Hardware Logo one, I mean another one, for example, made with python that requires to press Enter08:53
xiangfuqwebirc1900, check the /usr/bin/gmen2x.08:55
qwebirc1900ok, i'll give it a look08:56
xiangfuit's a script wrap the real gmen2x. I think you can add some lines before "exec /usr/bin/gmenu2x.bin"08:56
kyakwhitequark: by "host" you mean Ben, of course?08:59
kyakwhitequark: hard to copy-paste something from Ben when i'm on Windows. Basically, im using linux.inf frmo kernel.org to install rndis driver. In windows xp, Ben says something "chosen config #2: RNDIS". After 2-5 seconds, Windows shows error code 10 "unable to start device"09:01
kyakon Windows 7, Ben shows the same "chosen config #2: RNDIS", and after 2-5 seconds Windows also shows error code 10, but Ben also displays "config #0: unconfigured"09:02
kyakWIndows 7 is kinda smarter than WIndows XP i guess and informs linux side that something went wrong09:02
kyakthe error code 10 with regards to rndis is mentioned here and there in Internet09:04
kyakbut no real solution09:04
kyakand i also tried different cables/ports/and even laptops09:04
xiangfukyak, have you try this one : http://wiki.openmoko.org/wiki/Neo1973_and_Windows#USB_Ethernet_emulation09:10
xiangfutry some other Rndis.inf.09:11
xiangfunot sure if this works fine with Ben Nanonote 3.2 kernel.09:11
kyakxiangfu: yes, i tried all possible linux.inf files i could find09:23
kyakopenmoko, dingoo, beagleboard, some pocket readers and whatever else09:23
kyakthey are mostly the same09:23
whitequarkkyak: of course no09:45
whitequarkI mean a linux host09:45
whitequarkthat is, a PC09:46
whitequarkthere is a rndis_host driver for that09:48
kyakwhitequark: that's a good idea to try that. At least it will prove there is no hardware problems10:31
kyakbut even if it works Linux<->Linux, there is no point10:31
kyakthere is a CDC-ECM driver for Linux, i don't know who would use RNDIS10:32
kyakit must work with Windows10:32
qwebirc31535has the new software imagen for nanonote been published?11:03
whitequarkkyak: well, there are at least three USB ethernet drivers12:12
whitequarkCDC Ethernet, CDC-ECM and whatever RNDIS uses12:12
whitequarkand AFAIK CDC-ECM won't work on NN due to endpoint count or something like that12:12
LunaVoraxHello everyone!12:16
JCGsphi everyone12:18
LunaVoraxHey JCGsp 12:18
JCGspxiangfu, did you published new image for nanonote?12:19
xiangfuJCGsp, downloading and test. I fix small bug today. 12:20
xiangfuJCGsp, please wait ~ 1 hour :-)12:20
LunaVoraxNice, where's the changelog xiangfu?12:20
xiangfufixed all 'FAILED PACKAGES'12:21
xiangfumplayer back. QT/GTK program back. etc.12:21
JCGspthats great. Sadly, there's no SD boot yet?12:22
kyakwhitequark: cdc-ecm does work on Ben12:37
kyakin fact, it's the main and the only operating mode as of now12:42
kyaksince RNDIS was explicitely disabled to be able to use cdc driver on Windows (yes, there is such driver, but it's not free of charge and can only work 4 hours in trial mode before it disables)12:43
whitequarkmaybe that was CDC-Ethernet and not -ECM12:43
kyakand frankly speaking it works just perfect.. too pity it's not free12:43
kyakit's called Ethernet Gadget (with CDC Ethernet support)12:45
qi-bot[commit] Xiangfu: reflash_ben.sh: remove useless -t option (master) http://qi-hw.com/p/openwrt-packages/891cce113:40
JCGespHi everyone again14:13
xiangfuJCGesp, Hi14:18
xiangfuJCGesp, I just finished the test and release the new image.14:18
xiangfuJCGesp, changlog updated: http://en.qi-hardware.com/wiki/OpenWrt_Software_Image#2012-04-0914:19
xiangfusorry. correct URL: http://en.qi-hardware.com/wiki/OpenWrt_Software_Image#Image_2012-04-0914:19
JCGspxiang, http://downloads.qi-hardware.com/software/images/Ben_NanoNote_2GB_NAND/2012-04-09/ is where the image is?14:43
mthwhitequark: endpoint count might be a driver rather than a hardware issue14:59
mthis CDC-ECM supported by Windows out of the box?15:00
mthif it is, it would be worth investing some time in to get it running15:00
whitequarkmth: from the times when I read jz4740-udc docs and sources, it's a driver issue15:01
mthme and a friend tried for several hours to get CDC Ethernet working with a netbook running Win7 and various .inf files, both against a kernel with and without RNDIS support in the CDC Ethernet driver15:04
mthand we didn't succeed in making it work15:04
mthwhile with Linux and OS X, it's a matter of plugging in and accepting an IP address via DHCP15:05
mthI think OS X only works when RNDIS support is disabled in the kernel though15:06
whitequarkyou didn't do a ritual sacrifice15:06
whitequarkthat's why15:06
mthI'm not a religious man, maybe that's why I don't get along with Windows15:06
JCGsphttp://wiki.openmoko.org/wiki/Neo1973_and_Windows could one of these .inf work on windows? (maybe changing the IDs...)15:10
DocScrutinizerwpwrak: ping16:17
DocScrutinizerwpwrak: what's the recommended tool nowadays to relabel a vfat partition?16:18
DocScrutinizerwpwrak: please don't kill me answering "mtools", see `less /etc/mtools.conf' to stare in shock16:21
wpwrakyeah, i was thinking of mlabel :)16:30
qi-bot[commit] Maarten ter Huurne: Revert "MIPS: JZ4740: reset: Initialize hibernate wakeup counters." (jz-3.3) http://qi-hw.com/p/qi-kernel/b386be616:37
qi-bot[commit] Maarten ter Huurne: MIPS: JZ4740: reset: Initialize hibernate wakeup counters. (jz-3.3) http://qi-hw.com/p/qi-kernel/ba6ccf416:37
qi-bot[commit] Maarten ter Huurne: Revert "MIPS: Enable vmlinuz for JZ4740" (jz-3.3) http://qi-hw.com/p/qi-kernel/d1c6dd716:37
qi-bot[commit] LluĂ­s Batlle i Rossell: MIPS: Enable vmlinuz for JZ4740 (jz-3.3) http://qi-hw.com/p/qi-kernel/c1baa3616:37
qi-bot[commit] Maarten ter Huurne: ASoC: jz4740: A320: Add .owner to struct snd_soc_card (jz-3.3) http://qi-hw.com/p/qi-kernel/032d0e316:37
qi-bot[commit] Maarten ter Huurne: MIPS: JZ4740: Use round robin DMA channel priority mode. (jz-3.3) http://qi-hw.com/p/qi-kernel/4bbb1ee16:37
mthlarsc: can you review the round robin DMA commit?16:38
Ayladoes that fix the audio glitches?16:43
mthno, it doesn't seem to have any immediate effect16:54
mthbut it is more useful this way than have more-or-less random DMA channel priorities16:54
qi-botThe build was successful: http://fidelio.qi-hardware.com/~xiangfu/build-nanonote/openwrt-xburst.minimal-20120411-1055 17:15
whitequarkDocScrutinizer: hexdump and sed17:39
DocScrutinizerwhitequark: LOL, exactly what I suggested ;-D17:40
wpwrakdd perhaps ? :)17:41
larscmth: looks good19:43
larscthough i would have made prio an actual bool and called it low_prio or something like that19:43
larscbut that's just a minor nitpick19:44
mththat was how I thought about it originally, but the PM mentions priority groups and then numbering them makes more sense19:46
larsci'd like to switch to dmaengine at somepoint, but the dmaengine framework isn't quite there yet19:46
mtha future SoC might have more than two priority groups19:46
larscok, makes sense19:47
mthI'm assuming the groups have 3 channels each: the PM states 4 channels each but there are only 6 channels19:47
mthhmm, looking at the other priority schemes it could be 4-2 instead of 3-3 as well19:49
mth012345, 023145, 201345 have their permutations in the first 4 channels19:50
mthoops, I've got 2 as the value for round robin instead of 319:50
mthwhich is strange, because I checked the reg contents and it was 0x30019:51
mth...because I swapped mask and value19:52
mthI don't know if you're reading #dingoonity, but weird things happen if I set RDIL to a non-zero value19:53
mtheven though in block mode RDIL should be ignored19:53
DocScrutinizerwpwrak: actually my suggestion was to use dd and/or sed20:34
DocScrutinizera bit of tongue-in-cheek though20:35
wpwrakDocScrutinizer: it's either that, mtools, or roll-your-own :)21:11
DocScrutinizerseems also buggy21:11
DocScrutinizerIroN900:~# dosfslabel --version21:12
DocScrutinizerdosfslabel 2.11, 12 Mar 2005, FAT32, LFN21:12
DocScrutinizerso we updated to version 3.sth, and then...21:13
DocScrutinizer[2012-04-12 20:01:01] <freemangordon> FUUUK, seems this dosfslabel has bugs too21:14
DocScrutinizer[2012-04-12 20:01:23] <freemangordon> it does not update root directory label, only bootsector :(21:14
wpwrakyou went to seek DOS and you found suffering. have your desired thus been fulfilled ? :)21:24
whitequarkDocScrutinizer: I recall there was a tool which performed the inverse of what hd does21:42
DocScrutinizerwpwrak: the pity is uSD comes with VFAT usually21:43
whitequarkDocScrutinizer: hexdump -C21:47
whitequarkand, well, what for do you want that?21:48
DocScrutinizerwhat? relabeling uSD VFAT partitions?21:58
DocScrutinizeralas it seems it's a thing quite a number of lusers want to do21:59
DocScrutinizermkfs.vfat can take label as an parameter, but after you created the fs it seems there's no really good tool to tweak it22:00
dvdkDocScrutinizer: mlabel from mtools can do it22:01
DocScrutinizerwhich is quite a bit weird, from a general perspective22:01
wpwrakergo the tool fits the purpose :)22:02
DocScrutinizerdvdk: yeah, but that tool is so terribly dos-ified22:02
dvdkisn't FAT dosified by design?22:02
DocScrutinizer(me just wonders why wpwrak wasn'T able to invent a way more fsckdup UI for fsck.vfat then, to meet "tool fits the purpose"22:03
DocScrutinizerwpwrak: I'd expect you'd use driveletters for fsck then as well22:04
dvdk(for reference) sudo mlabel -i /dev/sdXXX ::NEW_LABEL22:04
DocScrutinizerfor reference:22:05
DocScrutinizerIroN900:~# mlabel --help22:05
DocScrutinizermlabel: invalid option -- -22:05
DocScrutinizerMtools version 3.9.10, dated March 2nd, 200522:05
DocScrutinizerUsage: mlabel [-vscVn] [-N serial] drive:22:05
Action: dvdk is still waiting for reflashing to finish22:06
DocScrutinizerI didn't feel like updating mtools if there's *anything* mor esane than this pile22:06
dvdkyeah, the "-i" option for the "::" pseudo-drive isn't documented well22:06
DocScrutinizerfor reference, from info:22:07
DocScrutinizer>> The syntax described herein is new for version `mtools-3.0'. The old line-oriented syntax is still supported. Each line beginning with a single letter is considered to be a drive description using the old syntax. Old style and new style drive sections may be mixed within the same configuration file, in order to make upgrading easier. Support for the old syntax will be phased out eventually, and in order to discourage its use, I 22:07
DocScrutinizerpurposefully omit its description here.22:07
DocScrutinizerthe last sentence really made my day22:07
wpwrakDocScrutinizer: (dosfsck) yeah, sorry, i got a bit carried away there. the experience of making a file system checker/repairer that - unlike DOS' CHKDSK - wasn't always in "conan" mode was too exciting :)22:09
DocScrutinizernow the best, the icing on top (quote from a script named mmc-rename.sh):22:09
DocScrutinizer# convert device name to an MS-DOS-style drive letter22:09
DocScrutinizerL=`eval grep '\"$DEV\"' /etc/mtools.conf | awk '{print $2}' | sed 's/://'`22:09
DocScrutinizerif [ "x$L" = x ]; then22:09
DocScrutinizer  echo "$0: could not determine drive letter"22:09
DocScrutinizeryou'd think usage ow awk obsoletes a pipe to sed, anyway. the eval still is arcane to me as well22:13
wpwrakthe pipe through sed is considerably shorter than doing the substitution in awk22:16
DocScrutinizershorter in what?22:18
DocScrutinizerin chars to type?22:18
DocScrutinizerhmm, as long as you know messybox by heart, you even might get away with that then, without getting fired for it22:19
DocScrutinizerin  a decent shell however it will load an additional binary22:19
wpwrakat your job, they fire people for loading binaries ? interesting :)22:20
DocScrutinizerI wish they did at Nokia22:20
wpwrakwhy bother ? the self-destruct will take care of all that anyway :)22:20
DocScrutinizernot for the OS on my daily phone22:21
DocScrutinizerwhich btw got a spare for the spare recently (the device)22:22
DocScrutinizerso I hope it will last until something decent emerges22:22
wpwrakyeah, everyine is biding their time, it seems ;)22:26
qi-bot[commit] Werner Almesberger: genex/: added page header with name, hierarchical path, and path to library (master) http://qi-hw.com/p/eda-tools/208d29b22:38
qi-bot[commit] Werner Almesberger: genex/sym2xps: only add question mark to F0, not to all fields (master) http://qi-hw.com/p/eda-tools/1e5dbb422:38
qi-bot[commit] Werner Almesberger: genex/pdf.c: rearranged header to show hierarchy first; balanced spacing (master) http://qi-hw.com/p/eda-tools/1a0631422:38
DocScrutinizerhmmm bringing USB hostmode to this N9 harmattan abomination. I must be insane23:20
whitequarkDocScrutinizer: you seem to like usb-host mode much23:23
DocScrutinizerwell, I know a bit about it meanwhile23:23
DocScrutinizer51but I hate this harmm shit Nokia invented23:47
infobothttp://www.developer.nokia.com/Community/Wiki/Harmattan:Developer_Library/Developing_for_Harmattan/Harmattan_security/Security_guide , or "The purpose of this framework is: ... to make sure that the platform meets the requirements set by third party software that requires a safe execution environment.", or http://en.wikipedia.org/wiki/Trusted_Computing#Criticism, or  http://en.qi-hardware.com/w/images/1/10/ME_382_LockedUpTechnology2.gif23:47
DocScrutinizer51got a N9 and even a N950 here - rarely ever touch any of both23:51
DocScrutinizer51it feels like touching something sticky slimy23:52
whitequarkso good my new lovely android doesn't have anything like that! :-)23:57
--- Fri Apr 13 201200:00

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