#qi-hardware IRC log for Thursday, 2011-02-24

rohhm. routing to chandra seems broken01:08
wpwrakhmm, today's yield so far, 2 dead atusb (one killed by stupidity, the other by mysterious forces of evil), plus one halfway done that hasn't revealed any fatal flaw yet.01:37
kristianpaulzrafa: (no get the original thread mail), thats odd..01:56
kristianpaulzrafa: before that you got the Art program mail?01:56
wpwrakzrafa: i think there were some problems yesterday at the hoster that - i think - hosts both openmoko and qi-hardware02:11
wpwrakzrafa: also, openmoko's mail server is currently down/unreachable02:12
Action: kristianpaul needs a nagios as a service...02:13
wolfspraulwpwrak: qi was down?03:06
wolfspraulI didn't get any notification - the notification service checks every 3 minutes.03:06
wolfspraula variety of services (smtp, ssh, http, etc)03:06
wpwrakwolfspraul: i think it was a network problem. i tried to access the mailing list archive and only got extremely slow responses or timeouts.03:17
zrafaare the qi and openmoko servers related? I mean, if they live on the same datacenter or house :)03:21
wolfspraulzrafa: not sure, they are at the same hosting company at least. roh said the hoster is good (at some point), and I listened :-)03:23
wolfspraulthat's a big hosting company though, with 3 datacenters in different cities in Germany03:23
wolfspraul200 gbit/second outward connections with 10+ peering partners03:23
wpwrakzrafa: traceroute lists.qi-hardware.com -> hos-tr4.ex3k7.rz13.hetzner.de ... turandot.qi-hardware.com03:24
wpwrakzrafa: mail.openmoko.org -> hos-tr4.ex3k2.rz13.hetzner.de ... static.88-198-23-33.clients.your-server.de (and then loops)03:24
wpwrakzrafa: so they may even be in the same room :)03:25
zrafawolfspraul: I am not complaining anyway ;) (about qi services).. I just complained openmoko community. I asked them, after openmoko MLs were down for several days, about if they have ideas or suggestions to continue living (and for example, if they have some ideas about how to have another openmoko ML better), but no one of them replied my mail. I felt really bad about .. nobody replied. All the people continued writting mails to openmoko ML like if nothi03:26
zrafawolfspraul: now.. openmoko MLs are down again.. fff03:27
zrafawpwrak: same room: I see :)03:27
wolfspraulthe openmoko servers are in a difficult situation03:29
wolfspraulbasically their setup is very maintenance heavy, and nobody can invest the time needed to reduce the maintenance effort03:29
zrafawolfspraul: yes, that is why I asked people/community for ideas or suggestions.03:29
wolfspraulmeanwhile afaik even the hosting of the servers is paid from private donations, just for goodwill03:29
wolfspraulzrafa: so what do the answers tell you? :-)03:30
rohwolfspraul: we just lost chandra. got rooted, nobody noticed. just killed all the rootkits03:30
zrafawolfspraul: the "no answers" tell me that people is okey using openmoko ML if it is still there.. if it is down for ever they are okey.03:31
wolfspraulouch, I see03:31
rohzrafa: i think the intent to get them back online is there. just not the time03:31
wolfspraulhow did the rooting happen?03:32
wolfspraulany clues?03:32
zrafaroh: no no.. I am happy with all the work people have done with openmoko services. I would like to see more people of community trying to discuss some ideas about how to continue03:32
rohzrafa: i think the easiest way is to get the important services into a 'care-take-able' status. less work than setting all stuff up from zero. more than null.03:34
rohits mostly a 'bitrot' problem. we still got loads of lennys around03:35
rohwolfspraul: the exim thing.03:35
rohwolfspraul: had it before, but didnt check closely enough. the other machine was 'broken' (wget misbehaved) so we noticed and fixed it.03:35
rohthat one.. well.. the person a) cleaned up properly behind him in step1, and just got an account. then nothing for >6 weeks03:36
wolfspraulso there are actual rootkits scanning the net to go after the debian/exim-4.69 remote root vulnerability?03:36
rohthus we didnt notice it.03:36
rohwolfspraul: yes there are. beware.03:36
wolfspraulgood thing I was on 100% alert when I saw it.03:36
wolfsprauloh I reacted within minutes :-)03:36
rohwolfspraul: we got attached a day after is was in the milw0rm package03:36
roheh attacked03:37
wolfspraulfirst removed exim on some servers where it wasn't needed. switched to postfix on others. applied some workaround that was proposed, and a day later upgraded to the official Debian fix.03:37
wolfspraulI've had my share of rootings in the past, so the adrenalin kicks in right away :-)03:37
wolfspraulhey, I've even shipped viruses to customers ages ago (on floppy disks). that was painful.03:38
rohpostfix isnt any better when it comes to security. its all code in C03:38
wolfspraulbig recall, won't forget.03:38
wolfspraulyes sure, I just tried to immediately do something about the root exploit, which sounded really scary.03:38
wolfsprauland now I learn some om servers got rooted - phew03:38
wolfspraulI gotta be more careful still.03:39
rohreminds me.. need to move all my remaining crap into openvz vms03:40
rohrebooting hw remotely is still sucky. so minimize the amount of hw without vm system03:41
wolfspraulroh: so I didn't get it. some om servers are now down for good?03:44
rohdown for now. need gismo to get routing back on03:48
wpwrakhmm, how i like my sendmail ...04:01
rohwpwrak: yay. get rooted even faster with even olfer exploits04:22
wpwrakroh: seems that the last major vulnerability was in 2003 :)04:28
rohwpwrak: if you want to be sure. use qmail04:31
rohi just were to lazy to set it up.04:31
wpwraknaw, never change a running system :)04:32
rohi still use a djbdns binary from 200404:32
rohtinydns that is.04:32
wpwrakand on my port 80, i have awhttpd. unlike apache, there it's quite feasible to review every single line of the code yourself ;-)04:33
rohdunno that one. i use lighty in such cases.04:33
rohbut sometimes there is no way around apache. svn-dav.. webdav in general...04:34
rohno features is really easy04:35
wpwrakyeah, awhttp is very light on features. you can ask for a file and if it exists, you'll get it. that's about all :)04:36
rohhm.. for that one can use fnord very well.. or lighty without modules04:38
wpwrak(fnord) hah, http://www.fefe.de/fnord/ -> "fnord 1.6 contained a buffer overrun in the CGI code."04:47
wpwrakthat's what you get for implementing such fancy nonsense ! :)04:47
larsckyak: I just had a look at the code in qt, which does the fontdir scanning07:12
larscit shouldn't be to hard to make it recursive07:13
kyaklarsc: that's good news! you could make it both recursive and using /usr/share/fonts. Maybe it will even be accepted by upstream? ;)07:56
larscthey don't accept any code except you sign some contributer agreement08:05
kyaki thought so08:06
mthyou could still report it as a wishlist bug08:09
qi-bot[commit] Xiangfu Liu: new package: Easy Tic-Tac-Toe game http://qi-hw.com/p/openwrt-packages/a0b6f7809:03
qi-bot[commit] Xiangfu Liu: new package: Easy Tic-Tac-Toe game http://qi-hw.com/p/openwrt-packages/7af155209:05
kyakjow_laptop, xMff: if it helps you in any way, here is the patch that made pango build for me: http://pastebin.mandriva.com/2199413:26
kyakltmain.sh is a piece of unknown shit to me13:26
kyakand for some reason it spreads its version of libtool everywhere else13:27
kyaki've added this info to original byg report here : https://dev.openwrt.org/ticket/881313:30
kyaknow let's see if w3m is really capable of drawing images in gtk2..13:37
xiangfukyak: do you know if there a way can switch back to kernel default font after run 'setfont' or 'setfont2' ?13:40
kyakxiangfu: i thought about it. Seems there is no way, because the kernel font is a C file, built-in kernel13:41
xiangfukyak: BTW: http://downloads.qi-hardware.com/software/images/NanoNote/Ben/2011-02-23/. it's a clean build. and it's for stable release13:41
kyakoh, ver good. i think i will reflash it, not just opkg upgrade13:42
xiangfukyak: yes. the kernel font is binary inside kernel. or maybe we can find a similar 6x11 font for 'setfont' for now the smallest for 'setfont' is 8x**13:42
xiangfukyak: I am not send email to list. maybe tomorrow. I will do a little test.13:43
kyakxiangfu: we could do it like this:13:43
kyaksetfont -o filename (to save the default kernel font)13:43
kyakthen we can always restore it by setfont filename13:44
kyakthis is theory13:44
Action: xiangfu downloads the images and SDK to notebook, the network connect slow today. 18K/s, it's always 300K/s13:44
xiangfukyak: thanks . testing now.13:44
kyakxiangfu: btw, have you tried the ter-v12n font?13:45
xiangfukyak: don't know that one.13:45
kyakit give us 53x20 term and, it looks very nice13:45
kyakit is from terminus-font13:45
kyaksetfont /usr/share/kbd/consolefonts/ter-v12n.psf13:46
xiangfukyak: you rock. setfont -o filename; setfont2 ...; setfont filename works just fine.13:46
kyaki like ter-v12n also because it has cyrillic glyphs :)13:47
Jay7kyak: seems terminus fonts are well-known only on cyrillic side :)13:47
Action: Jay7 is using ter-v16n in kexecboot UI now13:48
kyakJay7: well, it's defintiely super popular in Russia :) but it also has a very wide range of other supported alphabets13:48
xiangfudo you think the qi server slow today??13:49
jow_laptopkyak: ok13:49
Action: xiangfu think it's not my network problem.13:49
kyakJay7: how is your progress? Do you have it working well now?13:52
Jay7kyak: I hope today we will have OE recipe for kexecboot for NN13:52
xiangfuter-v12n.psf is better.13:53
kyakoh good! it should be a piece of cake then porting it to openwrt13:53
kyakxiangfu: btw.. i wanted to ask you - could you type some meaningfull text with kinyin? If it's really working fine, then we could add the screenshot to wiki13:55
Action: Jay7 hope to fit with kexecboot into Feb's news update ;)13:55
kyaki can type something. But i'm not sure, i have a good chance of typing some bad chinese words :)13:55
xiangfukyak: since I still using the old QT. not update. so I don't have that package. can you send the .ipk to me13:56
xiangfukyak: I will update all feeds.conf after this release.13:56
kyakxiangfu: http://downloads.qi-hardware.com/people/kyak/tmp/wqy-microhei_0.2.0-beta-2_xburst.ipk and http://downloads.qi-hardware.com/people/kyak/tmp/kinyin_0.43-1_xburst.ipk13:57
kyakyou'll have to install updated wqy font (it installs the necessary symlink for qt)13:57
xiangfu9.01K/s :(13:59
kyakmy downloads are momentary13:59
kyakso guess it's your network -\14:00
kyak2011-02-24 16:58:01 (592 KB/s) - «wqy-microhei_0.2.0-beta-2_xburst.ipk» saved [2099757/2099757]14:00
tuxbrain_awaywpwrak: have you read about this? http://zedstar.org/blog/2010/08/18/slowfi-protocols-on-copyleft-hardware/14:03
xiangfukyak: great. it's have two input methods of Simple Chinese.14:04
xiangfuthere are four input methods of traditional Chinese14:06
xiangfukyak: the 'save dialog' is too big.14:06
kyakyeah, it's standard dialog. Should be adapted to Ben's size14:07
xiangfualso 'open dialog' is big too.  others are great. finally I can take note in Chinese.14:08
xiangfukyak: thanks again. ;-).14:08
kyakok, so it's a worthy application. I was afraid that it would type nonsense or be non-usable :)14:08
kyakxiangfu: np! do i understand correctly that you type 2-4 times slower with these input methods?14:09
xiangfuemacs also have chinese input method. but emacs is a little scare for End Users ;-)14:09
kyakor you have more information in each symbol, so it's even faster?14:09
kyakxiangfu: i will try to adapt save/open dialogs. They work pretty slow, too - maybe replace it with something faster14:10
wolfspraulxiangfu: you need a Russian to help you add Chinese text input to your NanoNote?14:11
xiangfuhmm.. I only know the Simple Chineses. the WUBI works fine. some problem with PinYin.14:11
kyakxiangfu: the mappings are configurable, so you might want to have a look in /usr/share/kinyin14:13
wpwrakkyak: doesn't just "setfont" revert to the default font ? at lesat that's what the man page says ...14:15
xiangfukyak: wait. PinYin also works fine. it's need a little time understand the 'Kinyin'14:15
wpwrakkyak: ah no .. reading on, i see that's this is not the same14:16
kyakwpwrak: :)14:18
xiangfuwolfspraul: kinyin found by kyak. I work on that a little. but kyak is always faster :-)14:19
kyakxiangfu: now we need to find a vietnamese guy to check the ti¿ng ViÇt IM :)14:20
wolfspraulI haven't sold a single NanoNote to Vietnam, I think.14:20
wolfsprauleven though I went there once and there is a scene down there - but small.14:21
tuxbrain_awayme neither (yet) :)14:21
kyakwolfspraul: it's because of the lack of ti¿ng ViÇt :)14:21
wpwraktuxbrain_away: nice ! a customer for bens and maybe also for atbens ;-)14:21
xiangfukyak: you mean tencent? QQ?14:23
kyakxiangfu: i mean, the input method14:24
wpwrakbtw, it may be worth checking at the end of this week if there was any increase in sales in german-speaking countries. the ben has been mentioned quite favourably in a sub-thread in a heise article forum. the name is neither in the article nor the title of the thread, so many people may overlook it, but it may still help to establish a lower bound for the reach this has. http://www.heise.de/newsticker/foren/S-Re-China-und-MIPS-Eine-Lov14:24
wpwrakinteresting ... the board that i claimed was destroyed by unknown forces of evil has a reset signal floating well below VDD. that would explain the symptoms. now to the cause ...14:28
wpwrak(i had another one that had reset shorted to ground - not sure how i missed that during connectivity checking - that also didn't have a clock until i disentangled the reset line. that's how it found the issue.)14:30
wpwrakkyak: seems that his boss installed new surveillance cameras :)14:31
jow_laptopkyak: I concluded that replacing just libtool in an automatic manner is nearly impossible, I recently repared the old libtool fixup (which never worked fully) and promptly broke packages with it, therfore I think the only way is a complete autoreconf with all associated issues14:35
jow_laptopkyak: do you have an overview how many qi packages use libtool fixups atm?14:35
jow_laptopkyak: once we merge the fixups to backfire I expect fallout here and there14:35
kyakjow_laptop: there are 16 packages with PKG_FIXUP:=libtool14:37
kyaki'm not sure if it was really needed in all of these.14:38
wpwrakwolfspraul: MM > 1-rc3 should get an externally accessible 8:10 card slot, so that we can leverage all the cool toys people will hopefully build ;-)14:38
kyakjow_laptop: if we have an understanding how to fix the fallout, it shouldn't be a problem (liek we had the nice explanation for iconv/gettext)14:39
wolfspraulthat's a big change and first time I hear anybody suggesting it14:39
tuxbrain_awaywolfspraul: +1 to wpwrak mm proposal (second time)14:39
wpwrakroh: any news from the mail.openmoko.org front ? will there be at least a video of the hanging of the rooters ? :)14:40
wolfspraulnot that I'm opposed to it, it just needs to be added to the discussion then we see where things are going14:40
rjeffries_MM having an RF link would provide real value add14:40
tuxbrain_awayNN remote controling MM :)14:41
kyakwpwrak: i had a nice bash alias to open a random file from /usr/include on a random line :) Mostly for fun, but also for simulation of active work :)14:41
wpwrakwolfspraul: i'm about 50% serious :) if we want to establish the 8:10 card slot as a peripheral interface, then this would make sense. similar to how sony put their memorystick into everything, only that we're not evil ;-)14:41
rjeffries_I asked lekernal a few days ago his POV is that memory card that allows removal creates many tough s/w problems14:41
wpwrakkyak: does your window manager have a "boss key" ? ;-)14:41
Jay7tuxbrain_away: next device should be named LL :)14:41
wolfspraulmy first two ideas how Ben and Milkymist One work together is line-cable to get audio from Ben to m1, and USB keyboard14:42
tuxbrain_awayLazy Lizzard?14:42
rjeffries_tuxbrain_away or name could be OO or oh-oh14:42
wpwrakrjeffries_: my mail is down at the moment. there are no news on the UBB quote yet, are there ?14:43
rjeffries_wpwrak we got quoute back last night similar to what was expected14:43
tuxbrain_awaybtw the remote controling can be achieved with atusb on MM I guess14:44
rjeffries_good point tuxbrain_away14:44
tuxbrain_awaywill atben/atusb have enough bandwith to transmit audio?14:44
wpwrakrjeffries_: great ! what's the next step ? will you try to modify it for better pricing or such ? or just let it rest for now ? if you could forward the quote to werner@almesberger.net, that would be great. (that one gets lots of spam but it usually doens't fail at the same time openmoko.org does ;-)14:46
rjeffries_wolfspraul where on qi site do you tell new Ben owners to go for info on how to use Ben?14:46
rjeffries_wpwrak will send to that email no problemo14:46
zrafatuxbrain_away: 8bit, mono, 8 muestras por segundos? :)14:46
wpwrakrjeffries_: thanks !14:47
rjeffries_I do not have a good sense of how this quote compares to what tuxbrain_away got14:47
wolfspraulrjeffries_: hard to say, there are good pages all over the wiki.14:47
zrafatuxbrain_away: if it can send 1byte of audio, 8000 times per second you have something ;)14:47
tuxbrain_awayzrafa: well is a little bit less than expected :P14:47
wolfspraulI spent quite some time to cleanup the homepage, which I think is in a somewhat presentable and stable form now.14:48
zrafatuxbrain_away: if it can send 1byte of audio, 8000 times per second you have something ;), if you can send 2bytes, then stereo14:48
wolfspraulof course it needs more thought and intuitive links and arrangements too14:48
wolfspraulbut at the second level, the product level, there is just darkness14:48
zrafatuxbrain_away: just ask werner how many bytes per second of pure data (no frames) you can get14:48
wolfspraulso there is no great 'Ben NanoNote' product/user homepage14:49
rjeffries_wolfspraul I wait until the current testing image gets polished a bit more then I can give it a fair try14:49
wolfspraulthe wiki will improve when there are more people that help improve it14:49
zrafatuxbrain_away: and we can estimate14:49
tuxbrain_awayrjeffries_: why you don't have good sense?14:49
wpwraktuxbrain_away: raw data rate is 250 kbps (without playing non-standard tricks)14:49
rjeffries_I am stuck so far down in the weeds now with the olkder image that it is best to let Ben sit there and get a really good charge. ;)14:49
rjeffries_tuxbrain_away I do not understand your question sir14:50
zrafawpwrak: raw data = useful data for applications? or if we use some protocol or control that data would be part of those 250kbps?14:50
tuxbrain_awayrjeffries_says "I do not have a good sense of how this quote compares to what tuxbrain_away got"14:50
rjeffries_tuxbarin_away I mean I cAN NOT TELL HOW THE STWO QUOTES COMAPRE e.g. at different volume levels14:51
wpwrakzrafa: no, that's the raw bit rate inside a frame. you need to subtract framing overhead, pauses between frames, and any higher-level protocol overhead.14:51
rjeffries_this PCB house suggested ummersion gold treatment of entire surface at no extra cost vs gold on14:52
rjeffries_the UBB fingers only14:52
rjeffries_wpwrak you have mail14:52
tuxbrain_awayhow many do you inquiry rjeffries_?14:53
rjeffries_tuxbrain_away would you like to see the quoute? it is not secret.14:53
zrafatuxbrain_away: then you will have, at least, 8bit stereo or 16bit mono14:53
wpwrakJay7: NN = Nano-Note, MM = Minor-Miracle, LL = Long-Leap ? ;-)14:53
kyakjow_laptop: i will try to see how many packages _really_ need this fixup. I just checked, and not all of them who have it in Makefile really need it.14:54
rjeffries_wpwrak then we get to the long-awaited OMG model14:54
rjeffries_where wolfspraul has to add factories all working 7x365 no holidays14:55
tuxbrain_awayI prefer the WTF version14:55
Jay7good name for book reader14:55
wpwrakheh, ubuntu already switched from openoffice to libreoffice. that was quick :)14:55
tuxbrain_awayrjeffries_: yes I would like to see that quoting :)14:56
tuxbrain_awayJay7: +1 to the RTFM epaper NN version :)14:56
wpwrakrjeffries_: got the mail, thanks ! the per unit price is very high. almost ten times what tuxbrain got, at 500 units. setup cost is lower, though.14:57
Jay7btw, why not14:57
Jay7epaper + touchscreen :)14:57
tuxbrain_awayof cours with 8:10 technology :)14:57
rjeffries_tuxbrain_away you have mail14:58
Jay7sure :)14:58
rjeffries_wpwrak I thought the setup charge was way way low14:58
Jay7book reading shouldn't eat much CPU, so we can use the same14:59
rjeffries_I want to know if they would llaser cut or mill (curioisty)14:59
Jay7just add RAM and usb-host14:59
wpwraktuxbrain_away: (wtf version) longing to a second serving of the openmoko experience ? ;-)15:00
tuxbrain_awaywpwrak: haahahaahaahahaahah15:00
rjeffries_tuxbrain_away when  you sell a Ben what instructions do you include with the product15:01
tuxbrain_awaythe already send by Wolfgang. Apart of the Nanowar edition I don't do any other modification on NN15:02
wpwraktuxbrain_away: btw, if you try to draw a straight line through the two price points you got (total prices, with setup cost), you get a rather interesting equation :-)15:02
rjeffries_wpwrak I have only seen the USA quute so please share your thoughts ;)15:02
rjeffries_it was pretty cheap at 10K qty. ;)15:03
rjeffries_wpwrak I wonder if there would be economies if one panelized more than one design on same panel]15:04
rjeffries_e.g. UBB atben atusb (what else is there, so far?)15:04
wpwrakrjeffries_: one of the fundamental problems of a wiki like the qi-hw wiki is that there's no editorial oversight. so things tend to get chaotic. a typical condition is that there are a small number of pages regularly maintained by some individuals, many dead pages, only limited connectivity among all this, and no up to date index.15:04
rjeffries_I also want to knowm what size panel they used, if it only fits qty 100 UBB that is rather small15:05
wpwrakrjeffries_: (10k) hmm ? i only see it go to 2k15:05
rjeffries_wpwrak agree n sterngths and weaknesses of wili format15:05
rjeffries_wpwrak I need another cup of coffee you are correct15:05
wpwraktuxbrain_away: maybe you want to send ron your quote ? i think he'll be on the next plane to spain ;-)15:06
rjeffries_wpwrak please NO I do not read Spanish;)15:06
rjeffries_and the wholemEuro currency is so scary and Un-American. //xsmile//15:07
wpwrakrjeffries_: (straight line) the inclination is something like 16 euro-cents per unit ;-) not sure if this straight line is an appropriate approximation for the true unit cost, though, but it's interesting nevertheless15:07
wpwrakrjeffries_: (spanish) i thought you lived in northern mexico ? (-:C15:08
tuxbrain_awayrjeffries_: use the google Luck!15:08
tuxbrain_awayrjeffries_: yo have mail15:08
rjeffries_wpwrak I will soon be a mexican subject. the drug loards will occupy California15:08
tuxbrain_awaybtw you mail doen't include attachment _P15:09
rjeffries_tuxbarin really? I thought when I forwraded the attachments went along15:09
wpwrakrjeffries_: (unit cost) that's also why i find it rather amusing how people are trying to break down the packages tuxbrain sells. i don't think they realize that this is one of those cases where the slogan "the more you buy the more you save" is almost true also in total numbers :)15:10
rjeffries_tuxbrain_away I resent15:12
wpwrakrjeffries_: (spain) let's just say that, for the 500 unit cost in your quote, you could afford a biz-class flight to barcelona, spend a week there (it's a nice city), get the boards made locally, and still have some change left ;-)15:12
rjeffries_wpwrak but will David's boards have immersion gold surface? ;)15:13
rjeffries_David I think your quoute says single sided am I wrong?15:13
rjeffries_I am llearning as I work through this process15:14
wpwrakrjeffries_: no, they don't have ENIG. i don't know if that would have made them more expensive.15:14
rjeffries_David how will your pcb fab finish the board edges did they agree to laser cut?15:14
rjeffries_I think the gold layer is not expensive15:14
tuxbrain_awaydouble sided15:16
wpwrakrjeffries_: (gold) when i get a rough estimate at the online quoting of 4pcb.com, tin and ENIG were the same price. your specification has a thick ENIG layer, which may drive up cost.15:17
wpwrakrjeffries_: (that is, unless the fab just ignored it :)15:18
tuxbrain_awaywpwrak: can you resend me the rjeffries_quotation?15:21
tuxbrain_awaywpwrak:  not needed it has arrived :)15:23
wpwrakgreat :)15:23
tuxbrain_awayrjeffries_: yes you can rebuild the fucking concorde and take an UBB bag with that prices...15:24
wpwraktuxbrain_away: the 100 units price isn't too bad. so this may be a pcb house that specialized on low volumes. e.g., for prototype runs15:27
tuxbrain_away100 US units is still about <4 times spanish 100 units15:28
wpwraktuxbrain_away: no ... 150%, i think15:29
wpwraktuxbrain_away: you have to include setup cost and all this15:29
tuxbrain_awayyes I'm just counting the per unit costs15:29
tuxbrain_awayI only have to pay the setup cost once,15:29
tuxbrain_awaynext bathc is setup cost free15:30
tuxbrain_awayand shipping included in price15:30
wpwraktuxbrain_away: tax too, i guess ?15:31
wpwraktuxbrain_away: in ron's case, there would be CA sales tax. shipping should be negligible15:31
wpwrakrjeffries_: so the next step would be to find out what makes it so expensive. one thing you could try is to get a proper quote from 4pcb.com. that would give you a price point for comparison.15:32
tuxbrain_awayyep tax included, I must leave now, just not that is confirmed than will be ready on 3th March and I have asked to do a photo/video report to manufacturer :) will see what they answer :)15:33
wpwrakrjeffries_: (find out) i.e., ask the fab if there's anything they could suggest to lower the price15:33
wpwraktuxbrain_away: kewl one more week then :)15:33
dvdkwpwrak: the world is small, you know15:38
dvdkwpwrak: guess who's the author of the heise forum comment you mentioned :)15:38
dvdknow does that qualify as astroturfing?15:39
wpwrakdvdk: excellent undercover work ;-)))15:51
dvdkwpwrak: we are everywhere. we do not forgive. we do not forget. expect us.15:56
Action: dvdk couldn't resist)15:57
dvdkbtw i've noticed UBB isn't looking too profitable currently?15:59
wpwrakdvdk: you mean in terms of volume, margin, or both ?16:00
dvdkin terms of no break-even16:00
wpwrakdvdk: hmm, if tuxbrain_away can sell a few more, he should break even on expenses. so as long as his work is for free, that'll work :)16:01
dvdkwpwrak: "can sell a few more" might be the problem :)  in the open-source  world everybody's work is for free, isn't it :)16:02
dvdkwpwrak: of the 50 UBBs I preordered it looks like i'll only be able to "resell" 10 pcs.  So what do I do with 40 UBBs?  Buy another 39 Nanonote's? :)16:03
wpwrakdvdk: i think having around ten units per developer isn't excessive. if you think of them as something "cheap", like some 100 mil connector, then can see how you may use them up relatively quickly16:08
wpwrakdvdk: with 40 units, you're a bit on the high side, but if you make lots of experiments, you'll use them up, too :)16:09
wpwrakdvdk: i'll probably get 50, too. home-brewing them is getting boring :)16:10
dvdkwpwrak :)16:11
dvdkwpwrak: any tips about which microcontrollers could be programmed directly from the UBB's pins?  asynchronous serial programming won't be possible, ruling out most Atmega chips, won't it?16:12
dvdkwpwrak: have been thinking about hooking up a parallax propeller to the ubb16:12
wpwrakdvdk: i've programmed silabs (C2 protocol) and AVR (serial memory programming) over UBB or equivalent16:13
dvdkwpwrak: so AVR can be programmed using SPI?16:13
wpwrakdvdk: AVR is very synchronous :) yes, they use SPI16:14
wpwrakdvdk: silabs have a protocol with reset acting as clock plus a bi-directional data line. reset/clock timing has some restrictions (maximum low time is limited to a few us). also this works fine.16:15
dvdkwpwrak: maximum low time can be easily guaranteed using some R/C network :)16:15
wpwrakdvdk: eek !16:15
dvdkwpwrak: propeller has similar restrictions: pulse-width programming protocol:long low=0, short low=1 or something16:16
dvdkwpwrak: so here i also thougt about some R/Cs (using one pin vs. two pins=double current=half pulse width)16:17
kristianpaulsteve|m: :-)16:18
wpwrakdvdk: i think anything that doesn't need timing with better than 1-2 us accuracy will be a problem16:18
Action: dvdk does not understand16:18
Action: kristianpaul curios about the heise forum comment16:18
wpwrakdvdk: what are the prop's pulse widths ?16:18
dvdkwpwrak: negotiated (detected) initially.  1/19000 us should be fine16:19
dvdk1/19000 s16:19
dvdkwpwrak: note that i don't want to go into kernel space.  use gforth if possible.16:19
wpwrakdvdk: 50 us is slooooow ;-)16:20
dvdkkristianpaul (heise): note link is broken, need to be spliced from 2 comments16:20
wpwrakdvdk: i don't know about forth, but with C you'll be fine16:20
dvdkwpwrak: you mean probability of IRQ hitting is low enough?16:20
dvdkwpwrak: dosen't that depend on amount of data to transmit?16:20
dvdkwpwrak: last time i bit-banged the internal buzzer i could well "hear" the interrupts.16:21
wpwrakdvdk: yeah. and if you really have to, you could just disable interrupts in the critical partt16:21
dvdkwpwrak: huh? disable irqs from user-space?  not possible!?16:22
wpwrakdvdk: with 50 us, the interrupt may actually have enough time to just run to completion ;-)16:22
dvdkwpwrak: optimist!16:22
dvdkwpwrak: i mean if i can disable irqs, i could also bit-bang a uart16:22
wpwrakdvdk: (not possible) it's all a question of access permissions ;-) not that i've tried it on the ben. i've bit-banged a lot of stuff into the silabs chips with a 5 us deadline with very few problems16:23
wpwrak(uart) this is a bit trickier, because you need to know when the frame starts16:24
wpwrakof course, if you just let the uart driver take over the system completely, then you can be as precise as you want :)16:25
wolfspraulsteve|m: another hackaday article :-)16:25
dvdkwpwrak: ok, thanks for the info.  if bit-banging works that well, i'll safe some time designing RC networks :)16:26
wolfspraulglad to see they took yours16:26
kristianpaulwpwrak: you use sendmail as MTA?16:45
Action: kristianpaul usex exim and try hard to keep it update after recently sec bug16:45
kristianpaulbtw nanomap developer around?16:51
wolfspraulI think his nick is nielsle, but he's here rarely (if that's even the right nick)16:52
wolfspraulhow does this bot work again?16:52
wolfsprauloh, was here just 3 hours ago - but only for 9 minutes. well, now we know :-)16:53
kristianpaulgreat :-)16:54
wpwrakkristianpaul: (sendmail) yup. it has served me well for the last ~20 years :)17:09
wpwrak(circuit destroyed by evil forces) found it. it was parasitic resistance of flux, causing the reset line driven by a weak pull-up to raise very slowly and/or to an insufficient level.17:12
larsceverything would be so much easier if real world circuits were ideal as well...17:14
wpwraklarsc: hmm, sometimes things like parasitic capacitances and switches with non-zero Ron help to avoid inconvenient infinite derivatives of steps in functions :)17:16
larscstill, it would make things easier17:21
kristianpaul(if real world circuits were ideal as well...) What is the sense of life after that then? :-)17:24
rjeffrieswhat fun would it be if things Just Worked??17:24
larsci only had two two EE courses, once about d.c. circuits and maxwell equations and the other about a.c. circuits, so basically know nothing beyond ideal circuits17:24
kristianpaulwpwrak: is glib so helish as some people (as always) comment?17:27
larsc(and now i'm done with university)17:27
larsckristianpaul: glib2 or glibc?17:30
kristianpaulah well, i dint knew that..17:30
kristianpaullarsc: http://ur1.ca/3buba ? (done) ;-)17:31
larscactually now it's time for a break17:32
wpwrakhmm, trace-to-trace resistance through flux: ~20-30 kOhm. GRRR.17:37
wpwraklarsc: (done with university) oh, when did you finish ?17:38
larscwpwrak: soon ;)17:39
larscone more exam and a thesis to write17:39
wpwraklarsc: (soon) ah, good. so you can delay "real life" and "real work" for a bit longer, and be productive here instead ;-)17:40
rjeffrieslarsc what is your academic specialty17:40
Action: kristianpaul need to finish-resume university17:40
larscrjeffries: computer science17:40
Action: rjeffries nods17:41
wpwrakkristianpaul: glib is evil. glibc is fine as long as it's eglibc.17:41
larscwpwrak: yeah ;)17:41
kristianpaulwpwrak: I like eglibc :-)17:41
larscwpwrak: but only 3 more months or so and i guess it's time to start looking for the "real" work17:42
wpwraklarsc: what is your thesis about ?17:43
larsci haven't started yet, but wanted to write about sw/hw codesign for linux17:44
wpwraklarsc: (real work) a PhD may be a convenient solution for further deferring "real life" :)17:44
kristianpaulhmm, interesting17:45
larsc"Dr. Lars" ;)17:45
wpwraklarsc: ah, gcc output in ELF and verilog ? :)17:45
kristianpaulwpwrak: ;)17:45
larscnah, more like a howto not waste transistors and still get a decent performance17:46
larscmy experience is that you haven't have lots of functionality in a chip that is not used by the OS at all17:47
larsceither because it is not needed or it is faster/easier to do it in sw17:47
kristianpaultransmeta? ;-)17:47
wpwraklarsc: ah, LISC - Limited Instruction Set Computer :)17:48
larscon the other hand there is stuff thats much easier to implement in hw, like atomicity17:48
larscwpwrak: i was thinking more about peripherals17:49
kristianpaullarsc: What you think about that propaganda about elimintate interrupts using dedicated hardware modules?17:49
larsckristianpaul: like a coprocessor?17:50
kristianpaullarsc: yeah,let said many coprocessors..17:51
wpwraklike a propeller ? :)17:51
larscwell, you have to still communicate between the different modules somehow i guess17:51
larscand if you don't want to use polling you need interrupts anyway, i guess17:52
rjeffrieswpwrak propeller is interesting. wonder when they will finally release the ver 2 of that co[17:53
wpwraki suppose "coprocessors" could be useful for aggregation and general deceleration. let them share memory and have some hw-assisted wakeup/task queue mechanism.17:53
wpwrakrjeffries: i'm not sure they'll ever make it beyond "interesting". in my opinion, they're way too niche-happy. e.g., that whole thing about their own programming language and the self-contained development environment. pure insanity.17:54
wpwrakrjeffries: for me, that's basically a case of a great engineer who has lost it :-(17:55
wpwrakrjeffries: do they have a decent C compiler yet ? (i haven't followed the prop for some years)17:56
larsceven current desktop processors have 8 cores as well these days17:56
wpwraklarsc: you could make very small purpose-built coprocessors. hardwire the microcode and "ROM" the application.17:58
larscI think TI is actually building something similar.18:00
rjeffriesprop is coming along. the next version should be interesting if they get it done.18:03
rjeffriesagree they have one brilliant genius level guy and he does any damn thing he want to18:04
rjeffrieslarsc however at a VERY different price point;) and prop is very i/o friendly18:04
wpwrakrjeffries: let's hope the next prop moves more quickly than psoc3/5. those could also be also quite interesting chips, but they're taking forever for materialize beyond samples18:06
wpwraks/ for / to /18:06
larscwpwrak: i had the idea of a SoC with serveral tiny coprocessors with a reduced instruction set specialized for memory transfer and io bitbanging some time ago. but i guess it's an to obvious idea these days18:08
wpwraklarsc: (davinci) looks nice18:08
larscso the idea was depending on the firmware you load, you either get an i2c or an spi or whatever bus18:09
wpwraklarsc: i hope you got it patented back then :)18:09
larsci guess it is alreay patented by some big company18:09
wpwrakone ought to create a new branch of engineering, retro-engineering. search for patents granted 18-19 years ago, then update the idea to modern technology and you'll have it in a usable state by the time the patents expire18:11
dvdkwpwrak: i thought that is how qi-hardware works.  or does anybody expect a usable product within 20 years :)18:12
larscand back then was maybe a year ago, when i was annoyed with the nanonote that we had so many gpio pins available, but only could use protocols through sw gpio bitbanging18:12
wpwrakdvdk: that's the other approach: have an idea, make a proof of concept implementation, then sit on it for 20 years until all potential patents have expired :)18:13
dvdkwpwrak: about the timeframe of MM development? :)18:14
kristianpauldvdk: :D (MM)18:14
wpwraklarsc: (many gpio) you mean, after removing the keyboard ? :)18:14
larscmany > 418:16
kristianpaulnibble bus :-)18:17
kristianpaulbtw soerkis next product will have a FPGA on board.. interesting18:18
kristianpaullarsc: so i can have dma on the nanonote with the mmc linux driver?18:21
larscyou can18:21
larscif you write support for it18:21
larsci once had a patch for it but lost it :/18:22
wpwraklarsc: the sw engineer's equivalent of http://www.flickr.com/photos/buro9/299000164/18:36
samsonitehey ho18:45
samsonitegerman or english speaking room ?18:46
samsoniteive got a problem with my matrox triplehead 2 go, does anybody knows matrox?18:47
xMffwrong channel18:47
kyakwe should put the "we are not general hardware help channel" in the topic18:48
samsonitecould you please give me an advice for the right channel ?18:48
kristianpaulsamsonite:  /j #hardware18:48
samsonitei dont know why, but i cant join to #hardware18:48
kristianpaulyou need ask in #freenode18:49
samsoniteok thanks18:49
kristianpaulBut i bet your nick is not registerd or not identified18:49
wpwrak-> NickServ18:50
samsoniteit was registred since 10 minutes ago, i dont know why but all my nicks became unregistered after 2 days :-/18:50
kristianpaulsamsonite: ask in #freenode then :-)18:51
samsoniteok thank you18:52
wpwrakwe should rename the channel to cWktaGFyZHdhcmUK18:52
wpwrakaka  echo qi-hardware| base6418:52
wpwrakwell, maybe remove the GF, ZH, and UK, lest people think it's about GirlFriends, the city of Zurich, or the United Kingdom. so, cWktaydhcm18:54
wpwraks|city|city/canton|   # just in case :)18:56
xMffjust #qi is not allowed?19:02
kyakRegistered : Jul 22 14:53:37  2004 (6 years, 31 weeks, 1 day, 04:07:37 ago)19:03
kyakoccupied :)19:03
kyakLast used  : May 23 20:49:01  2007 (3 years, 39 weeks, 4 days, 22:12:13 ago)19:04
kyaki wonder why it hasn't dropped. might be some op's channel19:04
Action: tuxbrain_away now will think on that cat any time I see the lasrc nick19:21
wpwraklars the lolcat ;-)19:23
tuxbrain_awaywhen I was young and mindless and waste my life in non productive things.... ehem ....well lets say just when I was young .... we use to play kick-pong with that kind of non-related to channel19:29
tuxbrain_awaykick-pong: if at first kick the objective have the rejoin option on ops on channel give op to every known nick and fight to kick the target first under the warcry  Kickpong!!!!19:29
tuxbrain_awaybe a foreigner was not the only reason to become a target we were not xenofobs  when saw some silent for a while known nick and we were bored... and it has rejoin on.. kick-pong!!!19:33
wpwrakfrom project "per aspera ad astra": five atusb boards now accept a program download. that means that the transceiver-clock combo provides a clock to the MCU and that the MCU talks to avrdude. RF and USB are still unpopulated.19:34
tuxbrain_awaywpwrak: and that is good or is bad? :P19:35
wpwraktuxbrain_away: that is good :) means that i may have 4-5 working atusb boards by the end of today19:36
tuxbrain_awaythen is very good!19:37
wpwrakthen i can do a little more thorough RF testing. won't be able to ship things out before the weekend, though :-(19:37
wpwraktuxbrain_away: btw, if you still have the pro forma invoice i send with the counterweights, can you please mail it to me ? i lost the original in the disk disaster a few months ago :-(19:39
wpwraktuxbrain_away: ... mail to werner@almesberger.net   mail.openmoko.org is still down :-(19:40
tuxbrain_awayok I will ask to the paper keeper (aka  Victor)19:40
wpwraktuxbrain_away: ah, you don't keep your mails ? shame on you ! deleting mails is like burning books ! :)19:40
Jay7kyak: about #qi - may be just ask on #stuff?19:50
Jay7#staff :)19:50
Jay7nice mistype19:50
tuxbrain_awayah was on email! I though it was attached to the shippment. searching19:51
tuxbrain_awaywpwrak: as our friend ron says ... you have mail19:54
wpwraktuxbrain_away: thanks a lot !20:16
larscwpwrak: i have a bricked gta01, i can revive it via jtag. but u-boot then does not have any dfu partitions. i should though, i guess?20:38
wpwraklarsc: hmm, did you do the full devirgination, including the u-boot environment ?20:47
larsci only reflashed uboot20:48
wpwrakthen you have to enter u-boot and do the environment creation ritual there.20:51
wpwrakdynpart or whatever it was20:51
wpwrakyeah, dynpart, dynenv set, and finally saveenv should do the trick20:53
wpwrakerr, dynenv set u-boot_env20:53
larsci see20:54
wpwrakwhat a bloody mess that was ...20:54
larscunfortunately the console is read-only20:55
rohi think one can config that (uboot stdin/out)20:56
wpwraklarsc: if you can bring up the boot menu, you should be able to set it to usb. otherwise, you'd have to connect the debug board.20:58
larscthere is no boot menu20:59
tuxbrainwe can consider +24h compiling an stress test? :P21:41
kristianpaultuxbrain: sure !21:41
kristianpaultuxbrain: linux on nanonote? :D21:41
larscfirefox ;)21:42
tuxbrainnop avr toolchain (binutils, gcc, libc)21:42
kristianpaulagain why>21:44
tuxbrainbinutils already build, gcc on the go (since 7:00AM now 22:45PM) next libc21:44
wpwraktuxbrain: how good are you with harmonized codes ?21:45
kristianpaulbut why? is just a bootraping  test21:45
wpwrakkristianpaul: he doesn't want to spend 5 minutes on his pc compiling all this ;-)21:45
rohwhat storage do you use? nfs?21:47
rohvia usb?21:47
tuxbrainkristianpaul: basically to be able to build Arduino squetches directly on NN, yes I know I can crosscompile it but well ... due kernel has been already compiled ... I want to have the next hit21:47
tuxbraindirectly on NAND21:47
tuxbrainFilesystem           1K-blocks      Used Available Use% Mounted on21:49
tuxbrainubi0:rootfs            1905152    365024   1540128  19% /21:49
tuxbrainI have enough space :P21:49
rohpoor nand chip21:52
tuxbrainunfurtunately I can't say it was continue compiling time due I have to fix some errors time to time (3 stops on binutils, 3 stops on gcc until now) but well since last fix at (10:20AM) there where no more errors21:52
wpwrakroh: why poor ? soon it will be released from all worldly misery ;-)21:53
wpwraktuxbrain: how good are you with schedule B harmonized codes ?21:53
tuxbrainwhat the hell is that?21:53
rohwpwrak: hrhr. matyr!21:54
wpwrakroh: btw, how's the derootification going ?21:54
rohwpwrak: nothing new21:54
tuxbrainmm I see some kind of international trade customs code to clasify products ....21:57
wpwraktuxbrain: okay, i think you've already answered my question ;-) i'll bother wolfgang then21:58
tuxbrainwell this page explains it very well :) http://www.aw-wrdsmth.com/FAQ/US_Exporting_of_Goods-Services.html22:01
wpwraktuxbrain: what i'm looking for are the codes for atben, atusb, and atusb-pgm :)22:03
tuxbrainha, fist you have to find a good description :) then http://uscensus.prod.3ceonline.com/#/p=022:04
wpwraktuxbrain: this search engine will drive you crazy :)22:05
wpwrakwell, i'll have to ask fedex if i actually need that code or if it might possibly actually complicate export. as usual, i'm mainly concerned about argentine customs ...22:06
tuxbraingood luck dude, I'm sure customs are the creation than evil is most proud of22:08
kristianpaulwpwrak: are you afraid of get cauth on by your local goverment? ;-)22:09
kristianpaulhe, "That technology belong to us, as you devloped here" :p22:09
tuxbrainI see him siting on his trone, looking directly in the eyes of God , saying: Fix it if you have guts!22:10
wpwrakkristianpaul: i'm specifically afraid of this: http://www.afip.gov.ar/aduana/envios.asp#finnocom22:10
kristianpaul"Esta clase de envíos puede efectuarse sólo una vez por mes y por persona" WHATTTT !!!??22:11
kristianpaulI should consult my country export laws as well some day..22:12
wpwrakkristianpaul: nice, isn't it ? :) now i have to find out whether this actually applies as states there. but i'm not overly optimistic that it doesn't. second, i need to find out which work-around is actually feasible.22:13
tuxbrainI have imported samples from US , there was a temporaly loan so barelly no taxes to do some test and to bring to fosdem , but once it arrive at customs, the amount of burocracy needed was so much (both in time and amount) than I finally decide to pay it as definitive and comercial import and pay the taxes22:15
wpwrakthere's also the question whether customs will just accept it as non-commercial or whether they'll consider it a commercial export if it looks too professional. hence the question whether i should include the harmonized code or not.22:15
wpwraktuxbrain: temporary imports are evil. never do that.22:16
tuxbrainnow I know...22:16
wpwraktuxbrain: even in countries with a more enlightened bureaucracy - if this exists - like switzerland, such things are hellish22:16
kristianpaulwpwrak: (commercial) well it is, you can lie or suguest it as research?22:18
wpwrakkristianpaul: is it commercial ? these units are not being sold. i'm not receiving any remuneration for them or their construction. so where's the commerce in this ? :)22:19
rohgnnn i need tax number foo also.22:20
rohfor the cases. still no clue about which category is correct22:20
wpwrakkristianpaul: "research" would probably mean that i would have to be an officially accredited research institute, plus handle a bunch of extra forms22:20
wpwrakroh: i was thinking of just sending them all to you and you could then act as a hub, seeing how well you're already in the spirit of mastering international shipping formalities ;-)22:21
tuxbraincomercial customs must disappear... I understand that a country want some control on what enters and what exits for ilegal trafic (money, stolen materials, etc) and asure quality on food and medical parcels but not for tax... is a total no sense.22:21
kristianpaulwpwrak: (forms) indeed22:21
kristianpaulwpwrak: Is not fedex good for those things, i mean export facilites?22:22
kristianpaulguess they are but finally have to pay then22:22
wpwrakkristianpaul: sure, fedex are the best. but there are also limits to what they can do.22:22
wpwrakkristianpaul: the issue is not paying. even if there was an export tax, it would be trivial. the issue is that a commercial export would be a completely different process that a regular individual cannot perform. if it's the same as for commercial import, you need to register your company as a commercial importer/exporter, which i think also requires some certification. and then you still can't do the transaction yourself but need to e22:25
wpwrakmploy a customs broker for this. so to send +1 parcel, there would be costs probably in the hundreds if not thousands of dollars, plus a significant waste of time (probably weeks)22:25
wpwrakkristianpaul: also all this extra stuff would probably create additional reporting obligations22:26
kristianpaulwpwrak: wow, you have lots if rules there ;-)22:26
kristianpaulwpwrak: Here there is something called  RUT, is easy to get i have one22:27
kristianpaulAnd allow me to sell and buy stuff, and no pay taxes22:27
kristianpaulAlso to get paid by companies etc...22:27
wpwrakis RUT an ID number ? or a tax ?22:28
kristianpaulRegistro Unico Tributario22:28
kristianpaulOf course there is a limit, but i dont export millions per year :-)22:28
wpwrakoh, we have that too. but that's just for the basics. i can do domestic trade to my heart's content with this. but anything that crosses the border if a different story.22:29
kristianpauli see22:29
wpwrak(to my heart's content) well, there are a number of tax categories as well. there's one for really simple businesses, one for "professionals", and then things for real companies.22:30
kristianpaulwpwrak: You pay taxes per year ?22:30
kristianpaulI mean "Declaracion de impuestos"22:30
wpwraksome per year, some per month22:30
kristianpauloh hell22:31
wpwrakfor one of the monthly ones, the most efficient approach is to simply not pay it and wait until they complain. that way, the very tedious procedure is only necessary once or twice a year :)22:32
wpwrakthe other monthly one is easier, i can handle it online22:32
Action: kristianpaul free of taxes22:43
wpwrakkristianpaul: you don't pay tax ? yeah, that's an option, too :)22:44
kristianpaulwpwrak: just when buying articles and paying services, but not per year or month22:45
--- Fri Feb 25 201100:00

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