#qi-hardware IRC log for Saturday, 2012-02-11

abushcrafterforgI was thinking over simple cuts. For say voice recs.00:24
abushcrafterforgover = or00:24
abushcrafterforgor = of00:24
abushcrafterforgIs there a nice little GUI for simple cuts? is that possible or is that too much:D?00:31
kristianpaulno gui afaik00:54
kristianpaulbut i bet some command line tools could help you 00:55
abushcrafterforgok. sox i guess00:55
kristianpaulyup :)00:55
abushcrafterforgmhwaveedit is a small gtk audio editor. does it need to suport framebuffer or does GTK do that for it?00:56
kristianpaulhmm you got me on that00:57
kristianpaulbut gtk uses framebuffer, yes00:57
kristianpauldunno how this is "GTK do that for it" works for the nanonote 00:57
abushcrafterforgsorry bad english00:58
kristianpaulhe mee to ;)00:59
abushcrafterforgGTK handles the frame-buff side of things. So mhWaveEdit does not need to know about frame-buffer.01:00
kristianpauloh, cool !01:00
abushcrafterforgno I mean does it?01:00
kristianpauldont know..01:00
kristianpaulwait some minutes, i think xiangfu knows01:00
abushcrafterforgand English is my native lang...01:01
kristianpaulergh, sorry i'm not native so i gues i misundertood you in the first place01:01
abushcrafterforgsorry to put you though the suffering.01:02
abushcrafterforgany replays I will see in the logs.02:40
rex_fernandodoes anyone use the micro-sd wifi card mentioned in the wiki?06:32
rex_fernandoI can't find it for sale anywhere; is it not offered anymore?06:32
whitequarkDocScrutinizer: very interesting06:42
whitequarkphenolphtaleine might work06:42
whitequarkand I actually thought of kirlian effect in the past, but I was uncertain if it would really work06:42
whitequark(photos) if the chinese fab can only do photos, that's not worth 100USD06:44
whitequarkbut if they do actual tracing, then it indeed is06:44
whitequarkDocScrutinizer: also, I think it is fairly easy to use matlab to do some image processing and vectorize the polygons from these photos06:45
whitequarkI've seen a script like that on StackOverflow06:45
whitequarkthe vias are indeed the problem, but how chemical methods would help?06:46
whitequarkI don't quite get it06:46
whitequarkor do you mean ignoring the structure of inner layers and just tracing the connections between the actual pads?06:49
whitequarkrex_fernando: as far as I know, there are 0 (zero) wifi sdio cards with good Linux support06:50
whitequarkthere's one or two for which you can find half-done drivers (I think eyefi kind of works), but these aren't good anyway06:51
rex_fernandook, thanks06:53
whitequarkthey're also expensive06:54
wpwrak_i think the company that made one that's relatively popular on the ben stopped that product06:54
whitequarkeyefi is $120-$15006:54
wpwrak_or maybe even went out of business completely06:54
whitequarkthere's no need in sdio wifi anymore06:54
whitequarkeven alarm clocks have integrated wifi06:55
wpwrak_yes, that's the problem. we're among the very few that don't have built-in wifi06:55
whitequark... due to the fact that it is non-free, and not cost or other concerns06:56
whitequarkwpwrak_: how wifi chipset differs from the cpu in terms of freeness?06:57
whitequarkyou don't have netlists for both06:57
whitequarkand the standard is open, isn't it?06:57
wpwrak_openness is very bad for small devices. it seems that there are now reasonably open chips for laptop-sized devices.06:59
whitequarkwhat's bad? firmwares? absence of specs?07:00
whitequarkthere's Atheros, isn't it?07:00
wpwrak_at least the earlier versions of the standard (.11b, .11g) are sufficiently open. they have been at least partially implemented with SDR.07:01
wpwrak_firmware, access to chips or modules, no specs07:01
wpwrak_yes, there's atheros ...07:02
wpwrak_... that's where a lot of my bad experience comes from :)07:02
whitequarkwhat's wrong with it?07:02
wpwrak_oh, binary-only firmware full of bugs that never gets fixed07:02
wpwrak_then, you can't get chips, just modules. and these are also hard to find.07:03
whitequarkthey have open-sourced the fw for at least some chips07:03
whitequark(modules) oh yes, this is bad indeed07:03
wpwrak_which don't include the really small ones, e.g., the 6k series07:03
wpwrak_as things get bigger, laptop-sized, then you have better choices07:04
whitequarkok. I understand it now07:04
whitequarkyay! my M1 has arrived to Moscow07:06
whitequarkalready passed customs07:06
whitequarkwhich, according to the history, took 1 (one) minute07:07
DocScrutinizerwhitequark: yes, I mean ignoring the structure of inner layers and just tracing the connections between the actual pads10:58
whitequarkDocScrutinizer: ok, I was not able to find the board17:30
whitequarkeven for (reasonable) sum of money17:30
whitequarkso, unfortunately that's out of question atm17:30
DocScrutinizer51a pity17:30
whitequarkrepairmen say that the phone is quite popular, and as the most failing component is obviously screen, they just replace it17:31
whitequarkthere's nothing on ebay and I'm not spending $600 just to trace the board.17:31
DocScrutinizer51that's why I say not disclosing schematics is an offense to your customers, *only* - no sane reationale17:31
whitequarkI agree.17:31
whitequarkwell. there was a precedent before17:32
DocScrutinizer51chinese copycat has no problems with missing schematics17:32
DocScrutinizer51customer has17:32
Action: DocScrutinizer51 idly tries to pair his MX5500 cordless mouse via BT to N90017:39
DocScrutinizer51whitequark: look, even at OM our NDA pope Tony was afraid to even publish schematics that had circuit details similar to those in the application notes of the chip manuf17:42
DocScrutinizer51whitequark: you can see in OM GTA02 schematics that we asked uBlox for permission, and didn't dare to publish anything regarding calypso17:44
whitequarkhm, yes, that's quite paranoid17:46
whitequarkcould he really get sued?17:46
DocScrutinizer51he for sure not, his concern was about OM Inc17:47
DocScrutinizer51and me ;-D17:47
DocScrutinizer51as finally I took the responsibility17:47
DocScrutinizer51at least my name is on first page17:48
whitequarkthat's pretty screwed up17:48
wpwrak_well, that's due diligence for you :)17:49
DocScrutinizer51yeah, idustry wide17:49
wpwrak_of course, it's understandable. if something does happen, then the poor little guy who boldly waves the thing through will be in hot water.17:50
wpwrak_also a policy of openness doesn't really help. getting it authorized by someone higher up does, though.17:51
Action: DocScrutinizer51 repacks that silly ridiculously expensive mouse17:51
DocScrutinizer51well, my auth been by wolfspraul17:52
DocScrutinizer51and finally ny z17:52
DocScrutinizer51by Sean17:52
wpwrak_ny z ?17:53
DocScrutinizer51cold here17:53
wpwrak_ah :)17:53
wpwrak_a very pleasant 26 C here :)17:53
DocScrutinizer51also my adrenaline - just had to call the paramedics for an old lady that stumbled and hurt her face17:53
wpwrak_so you've been a good citizen17:54
whitequarkwhy are you trying to pair your mouse with N900 in a cold place outside? :)17:54
DocScrutinizer51n900 is a *mobile* device17:55
DocScrutinizer51so my situation is changing frequently17:55
DocScrutinizer51actually when I tried to pair I was inside a cafe sipping some latte17:57
DocScrutinizer51now I'm outside for a cig17:57
DocScrutinizer51friggin smoker prohibition17:57
whitequarkhere in Russia we have some cafes when you're allowed to smoke. e.g. a dedicated floor17:58
whitequarkquite convenient17:58
DocScrutinizer51had to swap the  mouse, as the one bought a week ago was broken17:59
DocScrutinizer51whitequark: indeed17:59
whitequarkMX5500 is Logitech18:00
whitequarkI've been trying to make a Logitech mouse (albeit other one) nicely with bluez for a ~year18:00
whitequarkthen gave up and bought the cheapest Dell one I could find18:01
DocScrutinizer51at my company, a whole building with 7 floors they urge employees to leave building and smoke at entrance in the icy cold18:01
whitequarkI'm happy ever since18:01
wpwrak_DocScrutinizer: at least they don't make you go to russia for it ;-))18:01
whitequarkdo they understand you are likely to waste more time by doing so?18:02
whitequark(formally, in Russia smoking is also prohibited in e.g. my university. but you can estimate the actual force of this regulation by a simple example: right under the "no smoking" sign there's an ashtray)18:03
DocScrutinizer51they don't care how long I work, as long as my reults are ok18:03
wpwrak_read: the more you smoke, the later the evenings :)18:05
DocScrutinizer51and as an external consultant I am not allowed to cary any minute plus or minus of work time to next month, so the whole idea is moot18:05
DocScrutinizer51wpwrak_: kinda18:06
Action: whitequark has bought a butane-powered soldering iron, Dremel Versatip18:06
whitequarkI wonder how useful would it actually be18:07
DocScrutinizer51meanhile my time accounting sheet looks rather uniform, 8:30-17;00 each day, in accordance with the 'boss'18:07
DocScrutinizer51whitequark: dang, I *loved* that one18:07
DocScrutinizer51used it for almost everything, everywhere18:08
whitequarkit was one of the cheapest, and it was also Dremel, which I heard makes quite good tools18:08
DocScrutinizer51lightweight and cordless is a huuuge convenience18:08
whitequarkwhat has happened with yours?18:08
DocScrutinizer51well, it vanished with all my other former life, as a indirect result of some fire in apartment18:09
DocScrutinizer51now all tools I got are about like a set of torx, a 5EUR DMM and a weller soldering station18:10
DocScrutinizer51stopped real hw work completely18:11
DocScrutinizer51ATM I got enough of nice tools like LeCroy scopes and R&S CMU200 etc, at work18:12
DocScrutinizer51whitequark: it's just too expensive to get a new completely equipped good lab at home, for... what?18:13
DocScrutinizer51I'm recently not doing any HW development so getting tools for this was kinda nonsense18:15
DocScrutinizer51and when I had the tools I wasn't doing too much at 'home' either18:15
DocScrutinizer51in former times, when I was 'young', yeah18:15
DocScrutinizer51been fun to mod high class casette recorders and whatnot else18:16
DocScrutinizer51or build own little controllers etc18:16
lindi-we formed a hackerspace here so that we don't need to keep HW tools at home :)18:16
DocScrutinizer51nowadays you buy that shit for 1.99 at china18:17
wpwrak_... on a weekly basis18:17
whitequarkI've bought some tools in China recently. tweezers are nice, I like them. kind of cheap too18:18
Action: DocScrutinizer51 walking ->home18:19
whitequarklindi-: how long does it take for hackerspace members to get to it?18:21
whitequarkon a workday evening for example18:21
lindi-whitequark: what do you mean how long?18:21
lindi-whitequark: that of course depends on where they live but around 30 minutes by bus for me18:22
whitequarkyou're not living at the place, right? then you need to go to it from home/school/university18:22
lindi-(bus takes 12 minutes but I'm including all the walking and waiting)18:23
lindi-a new bicycle route will make this easier this summer too :)18:23
lindi-whitequark: but yes, it'd of course be better to always have it closer :)18:25
whitequarkit's just that we have one in Moscow, but it's quite far from my home, and I'm quite lazy18:25
whitequarkwhich is an unfortunate combination18:25
whitequarkalso: -30 deg. C outside18:26
DocScrutinizer>>This service is permanently experimental & now twice as unreliable<< LOL18:44
whitequarkDocScrutinizer: what's that about?18:47
whitequarkabout the dremel18:52
whitequarkyou're so right! it's a great little gadget18:53
whitequarkthe nozzle for hot air on the soldering tip doubles as heatshrink, er, shrinker18:53
whitequarkincredibly convenient18:53
DocScrutinizerwhitequark: (what) quote from that website lindi posted18:54
DocScrutinizer(weller soldering station) actually ERSA18:57
DocScrutinizerWeller too expensive for the quality they offer18:58
Action: whitequark has just converted his kitchen table to work table19:14
whitequarkI still have to wait for Monday for M119:18
whitequarkand $RANDOM days to NN. tuxbrain doesn't seem very alive.19:18
whitequark*to be19:19
kyakno offense, but this is exactly what is thought when ordering atben/atusb from him19:26
kyakthinking about 10 bens/month he mentioned earlier is ML, makes me wonder... did he get a real job? :)19:27
larscwhatever a 'real' job is19:37
kyaksomething that is worth more than 80 bucks/months19:38
larscwolfgang recently mentioned in here that he has a new fulltime job and also a small child19:49
rjeffriesThis is a rather cool project. Just sayin' http://www.maxjusticz.com/introducing-the-chomp/22:03
Aylahi dvdk 22:10
dvdkhi ayla22:11
AylaI heard you did an optimized build of mplayer22:11
Aylait does use the IPU? or the MXU instruction set maybe?22:11
dvdksort of.  what are you looking for22:12
dvdkyes only the IPU22:12
dvdkfor yuv->rgb conversion plus scaling.  mxu instruction set, not not sof far.22:12
dvdk(not really neccessary to do more tuning for the small 320x240 display resolution)22:12
dvdkand the mxu stuff is ugly.  no gnu as support etc. :/22:12
Aylawell, it'd be a matter of adding the instruction set to buildroot22:13
Aylaanyway. If I'm right, your mplayer uses the IPU through /dev/mem,22:13
dvdkayla: no, it's not supported by the binutls neither upstream nor downstream anywhere.  there's only a awk script to patch assembler opcodes into hexcodes.22:13
Aylawould you be ready to work on a kernel driver for it, that uses v4l2 for instance, or help me doing it?22:14
dvdkayla: what's the need to put that into the kernel?  Xorg drivers aren't (mostly) in the kernel, too.22:15
dvdkthe driver is a library, look here:22:15
Aylathe JZ kernel has /dev/mem disabled22:16
dvdkayla: if you want to patch the kernel, than enabling dev/mem is easier than making a new driver :)22:16
AylaI don't want to enable /dev/mem22:16
dvdkbut you should :)22:16
Aylawe did disable it on purpose, we won't re-enable it again22:16
Ayla(to set the context, I do work with mth on OpenDingux, a linux distribution for the Dingoo A320 which uses the jz47xx kernel)22:17
dvdkfigured that.22:17
dvdkwell, then as a compromise, use Linux' user-space driver API and export memory access to the IPU regs only, plus a buffer via a device file.22:18
dvdkadapting the driver above would be trivial then.22:18
Aylathat's a possibility, yes22:19
viricdvdk: that driver still crashes quite a lot here22:19
viricmaybe not a lot22:19
viricI don't now if it's a deal with /dev/mem and in kernel it would work22:19
dvdki think there's an example somewhere in the mplayer sources.  for a minimalistic driver using linux kernel support.  'sh_veu_vid.c'.  my driver is based on that.22:19
dvdkviric: try to upgrade to latest jz47xx version.  i had some bugs.22:20
Aylathe other interest is that if we use a v4l2 driver, mplayer would just work without any change22:20
dvdkayla: isn't v4l2 about input, not output?  22:20
Aylait's both22:21
viricah both?22:21
viricI was about to say the same about input22:21
viricwhat example is there of v4l2 as output?22:22
AylaI believe it's not really used a lot22:22
dvdkwell, feel free to rip out any code from my sources you need.  but as it works for NN i'm not currently motivated to rewrite anything.  no time currently for such "non-productive" cosmetic work.22:22
viricdvdk: I was usin the latest tarball, some weeks ago22:22
dvdkalso i hate in-kernel drivers.  kernel is a moving target, not (internal) hard APIs.  that's bad.  22:22
Aylathat's fine, I was asking you just in case you were super-motivated to work on it22:23
viricdvdk: latest tarball on new year.22:23
dvdkviric: mayb mplayer crashes on you, not the driver?22:23
Aylabut I can do it, no problem22:23
viricdvdk: I don't know. I can't recover that fb on crash22:23
dvdkayla: :)22:23
dvdkviric: launch mplayer from a remote ssh.22:23
AylaI can do it => I can *try* :)22:23
viricdvdk: that does not fix anything22:23
dvdkviric: but you can see log output even if fb doesn't recover.22:24
viricdvdk: I mean, it shows on screen the video, but I can't recover the fb for text22:24
viricah yes22:24
Aylaviric, gmenu2x has some code to unlock that situation22:24
dvdkvirc: also 'strace' may help to locate the problem.22:24
viricI never run gmenu2x22:24
dvdkmaybe some mmap() crashes etc.22:24
viricit works for around 5 seconds22:25
viricthen, crash22:25
viricbut well, I agree I should write a better report thant hat ;)22:25
dvdkayla: ah, the correct term about coding in-kernel drivers is "technological debt".  since you'll have to adapt to kernel changes for all the time in the future.  think like "interest rate".  i hate debts :)22:25
viricdvdk: and if you don't do tat?22:26
dvdkso that's why i love user-space driver.22:26
dvdkwith dev/mem22:26
viricAyla: why you disabled /dev/mem?22:26
Aylaviric: because the old dingux programs were doing nasty things22:28
Aylalike messing with the clocks directly22:28
Aylaand the user space should not mess with the kernel space anyway22:29
viricah, for a videogame device, I think all should be allowed :)22:29
viricit's not a multiuser mainframe :)22:30
mthI met Hans de Goede at T-DOSE and explained the IPU and he said it would fit into v4l222:33
mthI don't remember exactly though how it would fit in22:33
mththere is some kind of API where you can pass buffers to be either converted to displayed22:34
mthfrom userspace to kernel and back22:34
mthworst abuse of /dev/mem we encountered was a program ported from GP2X that would access the GP2X hardware via /dev/mem22:35
mthnote that GP2X has an ARM SoC22:35
mthand the port still had that code active22:36
mthbut mostly the problem is user space code accessing the hardware directly while that same hardware is handled by kernel drivers as well22:36
mthif that happens, there is no way to debug it22:36
mthanother issue with user space drivers is that they don't react properly to systemwide events like suspend22:38
viricthere could be /dev/mem in the kernel, and lack the device file22:39
viricjust as a 'warning'22:39
viricor maybe it can be set as a module22:39
viricah yes, the suspend thing...22:39
mththe device files are auto-created in a ramdisk, but probably it can be configured to not create them22:39
mthstill, if someone has a genuine need to use /dev/mem, it's not that hard to build a kernel that has it22:40
viricbut still, it's the situation of "a quickly written driver" vs "a well maintained driver", regardless of user or kernel mode22:40
mthI made it an option; all the code is still there, it's just disabled in our config22:40
viricI think that there should be a place for "quickly written driver"22:41
viricbut users should know what they run, yes.22:41
viricthen there are the people like dvdk, that make a driver of value using /dev/mem, and then the NN people get forced to provide it ;)22:43
mththat's why we're looking to move his driver into the kernel, or at least away from /dev/mem22:43
viricI understand.22:44
viricmove from "quickly written driver" to "well maintained driver"22:44
viricas for me, I lack the skills to do that with short dedication22:45
mthwe have our share of hacky in-kernel drivers too though, we're still several steps away from a clean solution22:45
viricmth: otherwise all would be at linus' linux22:45
mthskills can be developed; I did very little kernel programming before this and I still have a lot to learn22:45
viricyes, me too...22:46
viricI recently did some hacks on the fuloong to fix its boot22:46
viricbut nothing beyond that22:46
viric(something in the NN too)22:47
viricand I read carefully the driver from dvdk :)22:47
viricbut I'm busy with other things right now22:47
--- Sun Feb 12 201200:00

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