2011-01-17 01:05 [commit] Xiangfu Liu: INSTALL_FGORTRAN needs those two options http://qi-hw.com/p/openwrt-xburst/9ddfc59 2011-01-17 01:32 [commit] Xiangfu Liu: pmars: fix compile error, remove static CC ... in Makefile http://qi-hw.com/p/openwrt-packages/3259de0 2011-01-17 01:32 [commit] Xiangfu Liu: mandoc, remove BROKEN, compile just fine http://qi-hw.com/p/openwrt-packages/deb6c2e 2011-01-17 01:32 [commit] Xiangfu Liu: robots: remove BROKEN, install to /usr/bin/ http://qi-hw.com/p/openwrt-packages/57b658a 2011-01-17 01:32 [commit] Xiangfu Liu: bwbasic, fix compile error, remove BROKEN http://qi-hw.com/p/openwrt-packages/1a1d25d 2011-01-17 01:41 back ... 2011-01-17 01:41 kristianpaul: cutting wood ... well, for simple things, i just write a script that generates a set of coordinates in gnuplot format 2011-01-17 01:41 kristianpaul: then i pass that through my coordinate transform scripts to mill 2011-01-17 01:42 kristianpaul: that's some ten or so uses so far. before, i also generated RML-1 toolpaths directly, but that's not really convenient. 2011-01-17 01:43 kristianpaul: once, i used HeeksCAD with scripting (for the counterweight) 2011-01-17 01:45 kristianpaul: for PCBs, I can now generate toolpaths for drilling and milling directly from kiCad's gerbers. 2011-01-17 01:46 kristianpaul: so the process is not just to decide how to fit the pieces on the raw board, to enter the respective transforms in my mkmk script, to check the coordinates with gnuplot (make plot), and then go to the mill 2011-01-17 07:37 kyak: he, I ran into the libiconv problem too now. linking libSDL.so, I get undefined reference to libiconv_open/close 2011-01-17 07:38 so I need to change the dependency of libSDL from iconv to iconv-full? 2011-01-17 07:39 actually the earlier error is libiconv.so.2 not found... 2011-01-17 08:11 wolfspraul: considering http://downloads.qi-hardware.com/people/werner/tmp/c2-use.jpg 2011-01-17 08:11 and http://downloads.qi-hardware.com/people/werner/tmp/c2-pogo.jpg 2011-01-17 08:11 how long would adam scream if i sent him this as the means of flashing the microcontroller in atusb ? :) 2011-01-17 08:12 (the setup picture doesn't include the finger pressing the pogo pins firmly onto the pads) 2011-01-17 08:13 xiangfu: hey there! I ran into a bug compiling SDL_image 2011-01-17 08:13 it complains about libiconv_open/close 2011-01-17 08:13 I meet that too. 2011-01-17 08:13 how do I make SDL_image use libiconv-full instead of libiconv? 2011-01-17 08:14 wpwrak: nice picture! the Qi cloth is useful for something :-) 2011-01-17 08:15 wolfspraul: oh, i love it. makes the perfect background :) 2011-01-17 08:15 I don't understand the -pogo pictuer 2011-01-17 08:15 picture 2011-01-17 08:15 perhaps it's a bit too twisted ;-)) 2011-01-17 08:16 this is what i use to flash the microcontroller in atusb (the chip comes empty and can't talk over usb until some firmware is loaded) 2011-01-17 08:17 the ben runs the flash software. to make the connection, i hold that thingy onto the pads. on the pogo pic, you see the spring-loaded pogo pins that make the contact 2011-01-17 08:18 I understood what it was meant for, but wasn't aware there are such contacts on atusb? 2011-01-17 08:18 the pitch looks pretty big 2011-01-17 08:18 so this is for atusb, how about at810? 2011-01-17 08:19 the pitch is 100 mil. standard header. 2011-01-17 08:19 heh, took me a while to connect this to 8:10 ;-) 2011-01-17 08:19 that one has no microcontroller and thus needs no programming 2011-01-17 08:19 the picture in c2-use, is it using the c2-pogo? 2011-01-17 08:19 yes 2011-01-17 08:19 oh. hard to see. 2011-01-17 08:20 c2-pogo is the left side of the cable in c2-use 2011-01-17 08:20 (upside down, to show the pins) 2011-01-17 08:20 (hard to see) yes, it's all yellow-white on yellow-white 2011-01-17 08:20 ah yes 2011-01-17 08:20 finally 2011-01-17 08:21 after looking at the two pictures more closely, I get it 2011-01-17 08:21 wolfspraul: if you using the last openwrt-xburst. it's already using the full libiconv. 2011-01-17 08:21 c2-use looked like the connection was flat 2011-01-17 08:21 soldered or so 2011-01-17 08:21 I didn't see those press-down pins, the angle is not so good for that 2011-01-17 08:21 xiangfu: ok great, I update. 2011-01-17 08:21 ah yes, you're right. looks misleading :) 2011-01-17 08:21 wolfspraul: I guess you already have that. 2011-01-17 08:21 how about htop and fetchmail, should I disable them to save trouble? 2011-01-17 08:22 xiangfu: I checked out just yesterday, and run into the sdl_image/iconv bug 2011-01-17 08:22 wolfspraul: this commit from kyak, make our using the openwrt-package's libiconv. : 46b04421f92cfd3bae5964858771a3086ad1c8e1 2011-01-17 08:23 that didn't help me, I definitely have that 2011-01-17 08:23 my problem is in the configuration (!) of SDL_image 2011-01-17 08:24 both libiconv.so.2 and libiconv-full/lib/libiconv.so.2 exist (in staging_dir/...) 2011-01-17 08:24 but I think the configuration step in SDL_image will not find the one in the libiconv-full subdirectory 2011-01-17 08:25 wpwrak: sure it's perfect. Adam still has 1 or 2 8:10 breakout cables 2011-01-17 08:26 maybe he can just reuse those for now 2011-01-17 08:29 wolfspraul: after that commit OpenWrt will using the openwrt-packages's libiconv,(not upstream stab version)Â Â which is same with libiconv-full 2011-01-17 08:29 xiangfu: http://pastebin.com/Up7QyJ2P 2011-01-17 08:30 wolfspraul: i would just send you a pair of those cables. i have several. easiest piece of qi-hardware ever ;-) 2011-01-17 08:34 this is a bit clearer: http://downloads.qi-hardware.com/people/werner/tmp/c2-closer.jpg 2011-01-17 08:34 xiangfu: you are right, the two libiconv.so.2 look the same, strange... 2011-01-17 08:35 wolfspraul: maybe we need add -liconv. to CONFIGURE_VARS 2011-01-17 08:35 ha. yes. only a 'little bit' though 2011-01-17 08:37 wolfspraul: try add this to your local feeds/qipakcage/libsdl-image/Makefile : http://pastebin.com/p4bAgFKr 2011-01-17 08:37 i don't think adam will need a 30 minutes training video ;-) 2011-01-17 08:37 wolfspraul: we meet different errors. 2011-01-17 08:46 wolfspraul: (at810) may be a tad confusing. maybe just "atben" ? 2011-01-17 08:48 atben is nice, sure 2011-01-17 08:50 xiangfu: yes I think it worked! 2011-01-17 08:50 thanks a lot 2011-01-17 08:51 wolfspraul: ok. I will commit that change 2011-01-17 08:51 the build is still running, but I think it's past SDL_image already... 2011-01-17 08:54 [commit] Xiangfu Liu: libsdl-image, add -liconv to CONFIGURE_VARS http://qi-hw.com/p/openwrt-packages/3d0c2c5 2011-01-17 08:56 wolfspraul: btw, is "USB" actually safe to use ? 2011-01-17 08:57 xiangfu: I got an error in emacs now, from strip... http://pastebin.com/G8pyn13N 2011-01-17 08:59 [usb] don't know, for now I assume yes it is safe 2011-01-17 09:00 usb seems to be focused around their logos 2011-01-17 09:00 there is no yearly fee, only a one-time payment for the vendor id, which we can use for all Qi projects 2011-01-17 09:00 (logo) excellent 2011-01-17 09:01 but they do seem to work on some 'added value' logos, so you can only put this or that logo on your box if you meet this or that requirement. 2011-01-17 09:01 yes, certification, i think 2011-01-17 09:01 (for any logo) 2011-01-17 09:02 and then there's that gazillion of usb variants, 2.0, 3.0, wireless, ... 2011-01-17 09:02 in contrast the SD people have this concept of a 'host license' that anybody who manufactures an SD host device has to pay 2011-01-17 09:02 which is not even cheap, I think 1000 or 2000 usd / yr 2011-01-17 09:02 yr / manufacturer 2011-01-17 09:03 per year per manufacturer 2011-01-17 09:03 well, anything less and the paperwork overhead would be more expensive :) 2011-01-17 09:03 true 2011-01-17 09:04 the USB people are probably also smaller (guessing) 2011-01-17 09:04 maybe only a few office admin/secretary types 2011-01-17 09:04 they make 2000 USD for every vendor id 2011-01-17 09:04 maybe a little more for certification programs? don't know 2011-01-17 09:04 I think SD is bigger. 2011-01-17 09:04 (the organization I mean) 2011-01-17 09:05 not the usefulness :-) 2011-01-17 09:05 the fee could be symbolic, without relation to the real cost of operation 2011-01-17 09:05 well, 2k/vendor ought to feed the registrar :) 2011-01-17 09:08 ah, the next ben-wpan push will derail schhist (atusd -> atben renaming. directory and files inside). it should be able to figure out where things are/were when you give it the new names, though 2011-01-17 09:11 wtf 2011-01-17 09:12 welcome thy friends ? 2011-01-17 09:12 "Kernel panic - not syncing: VFS: Unable to mount root fs on unknown-block(0,0)" 2011-01-17 09:13 ubifs corruption ? 2011-01-17 09:13 so what now? 2011-01-17 09:13 reflash? 2011-01-17 09:13 can running jlime from sd have caused this? 2011-01-17 09:14 because I remember they had incompatible partition tables 2011-01-17 09:14 bartbes: the *previous* messages should be more informative than that last 2011-01-17 09:15 but the OpenWrt splash is hiding those 2011-01-17 09:15 ( can running jlime from sd have caused this? ) i doubt it 2011-01-17 09:15 i think i did get ubifs corruption after running jlime from usd. not sure about cause and effect, though. the partition layout shouldn't have been a problem in this case 2011-01-17 09:16 even if it tried to mount an ubifs volume at the wrong place? 2011-01-17 09:16 wpwrak: you have a pic of atusb without pogo? 2011-01-17 09:17 wpwrak: (bifs corruption after running jlime from usd) wow really? i never had such us proble here 2011-01-17 09:17 wolfspraul: it's try to using host strip which is wrong. 2011-01-17 09:17 kristianpaul: here's an old one: http://downloads.qi-hardware.com/people/werner/tmp/nemesis.jpg 2011-01-17 09:18 ah no pongo female? there is a famale conector inst? 2011-01-17 09:18 kristianpaul: i don't know if it really was jlime doing something bad. could also be the usbboot nand erase bug or such 2011-01-17 09:18 s/pongo/pogo 2011-01-17 09:19 kristianpaul: pogo connect to pads on the pcb 2011-01-17 09:19 kristianpaul: if you look at the layout, that's all there it 2011-01-17 09:20 kristianpaul: the header i soldered in http://downloads.qi-hardware.com/people/werner/tmp/nemesis.jpg was just a quick hack because i hadn't made the pogo connector back then 2011-01-17 09:20 let me se 2011-01-17 09:21 ah, a *hack*, that answer my question :-) 2011-01-17 09:24 well, the pogos are admittedly no less of a hack ;-) what's nicer about them is that you don't have to cut off the connector when you're done 2011-01-17 09:25 (cutting the connector usually damages the pads, so you better make sure you don't need them again) 2011-01-17 09:27 yeah i tough data (damages the pads) 2011-01-17 09:27 s/data/that 2011-01-17 09:29 so ehm 2011-01-17 09:29 how will I restore my ben? 2011-01-17 09:29 reflash? 2011-01-17 09:30 yeap 2011-01-17 09:30 well you can try usbboot for just uimage 2011-01-17 09:31 no, for the ubifs uimage. the kernel is okay 2011-01-17 09:31 (well, should be.) 2011-01-17 09:31 ah ... if the u-boot/kernel combo changed, maybe that changed the partition layout too 2011-01-17 09:32 in this case, reverting to the u-boot/kernel (not sure which one of the two defines the layout) of openwrt may restore the ubifs 2011-01-17 09:32 I never changed those 2011-01-17 09:33 then it seems that you need to replace the ubifs 2011-01-17 09:39 wolfspraul: it looks like yo uneed to build from scratch 2011-01-17 09:39 i.e. make clean; rm -rf build_dir staging_dir feeds tmp; make package/symlinks 2011-01-17 09:40 libSDL should build just fine 2011-01-17 09:40 it may be using some lefovers from your previous builds.. 2011-01-17 09:41 kyak: no I think we tracked it down already, added -liconv 2011-01-17 09:42 not libsdl, but SDL_image in qi's openwrt-packages 2011-01-17 09:42 this is strange, because i didn't have such problem with SDL_image 2011-01-17 09:42 it was not a problem between iconv and iconf-full, it was just a missing -liconv for the linker 2011-01-17 09:43 in any case, good that it worked.. -liconv won't do harm :) 2011-01-17 09:44 xiangfu: btw, i mentioned that you link manually against -lncurses a lot. Since we have ncursesw in openwrt-packages, you could link against that 2011-01-17 09:44 adding libncursesw to depenedcies, of course 2011-01-17 09:45 i mentioned/i noticed 2011-01-17 09:45 kyak: I totally forget that. sorry. 2011-01-17 09:45 wolfspraul: actually, i had such problem with SDL, before i rebuilt from scratch :) 2011-01-17 09:46 kyak: ok. I will try using the ncursesw, will try on some packages that I just committedÂ Â todayÂ Â :) 2011-01-17 09:46 xiangfu: no problem.. just occured to me 2011-01-17 09:48 wolfspraul: I think I found the emacs compile error problem. 2011-01-17 09:48 xiangfu: you could try entering wide-char characters to your ncursesw packages :) 2011-01-17 09:49 wolfspraul: http://pastebin.com/RAabMhiN , patch is like this. now I am testing. 2011-01-17 09:49 kyak: yes. 2011-01-17 09:50 time to go home now. see you later! 2011-01-17 09:50 kyak: see you 2011-01-17 09:57 xiangfu: you removed the MFLAGS= section? 2011-01-17 09:58 wolfspraul: no. the terminal don't display that. 2011-01-17 09:58 wolfspraul: I just copy form the terminal output. not a real patch. this is just fyi. 2011-01-17 09:59 ah ok 2011-01-17 10:15 xiangfu: I just added -liconv to libsdl-gfx/Makefile as well 2011-01-17 10:20 wolfspraul: ok. 2011-01-17 10:20 wolfspraul: I think the problem about emacs, is the strip is not correct 2011-01-17 10:21 wolfspraul: I need find a good way to pass correct STRIP to it.Â Â will work on that after I wake up. now need goto sleep. :) 2011-01-17 10:21 cu 2011-01-17 10:22 [commit] Werner Almesberger: introduced 8_10-card.fpd as politically correct alias of usd-card.fpd http://qi-hw.com/p/kicad-libs/85fb77c 2011-01-17 10:27 n8 2011-01-17 10:28 [commit] Werner Almesberger: atusd.brd: rearranged component references and values for printing/display http://qi-hw.com/p/ben-wpan/6ac9168 2011-01-17 10:28 [commit] Werner Almesberger: great atusd -> atben renaming: changed the directory name http://qi-hw.com/p/ben-wpan/3392647 2011-01-17 10:28 [commit] Werner Almesberger: great atusd -> atben renaming: renamed design files and some references http://qi-hw.com/p/ben-wpan/319bd83 2011-01-17 10:28 [commit] Werner Almesberger: removed atben/tools/ (functionality has moved to atrf-id long ago) http://qi-hw.com/p/ben-wpan/0dc344a 2011-01-17 10:28 [commit] Werner Almesberger: great atusd -> atben renaming: updated references in atben/cam/ http://qi-hw.com/p/ben-wpan/fb07d57 2011-01-17 10:28 [commit] Werner Almesberger: great atusd -> atben renaming: changed "uSD" in project title to "8:10 card" http://qi-hw.com/p/ben-wpan/f7a8bbd 2011-01-17 10:28 [commit] Werner Almesberger: great atusd -> atben renaming: renamed tools/atusd.c and Makefile reference http://qi-hw.com/p/ben-wpan/32de4b6 2011-01-17 10:28 [commit] Werner Almesberger: great atusd -> atben renaming: changed remaining references in tools/ http://qi-hw.com/p/ben-wpan/1cd079e 2011-01-17 10:28 [commit] Werner Almesberger: great atusd -> atben renaming: use 8_10-card.fpd instead of usd-card.fpd http://qi-hw.com/p/ben-wpan/5281f2c 2011-01-17 10:28 [commit] Werner Almesberger: great atusd -> atben renaming: update references in bom/ as well http://qi-hw.com/p/ben-wpan/98b4387 2011-01-17 10:47 [commit] Werner Almesberger: use 8_10-card.fpd instead of usd-card.fpd http://qi-hw.com/p/ben-blinkenlights/9037090 2011-01-17 10:47 [commit] Werner Almesberger: README: changed "uSD" to "8:10 card" http://qi-hw.com/p/ben-blinkenlights/d783951 2011-01-17 10:53 wolfspraul: does adam have access to a spectrum analyzer in general / for small-run testing ? 2011-01-17 10:59 don't know we have to find out 2011-01-17 10:59 what do you need it for? for design verification, or for production verification (each board)? 2011-01-17 10:59 what exactly are you looking for? 2011-01-17 11:01 mainly for production verification. to see if a board has an abnormal spectrum (e.g., very noisy) 2011-01-17 11:01 (or very weak) 2011-01-17 11:01 with a spectrum analyzer, you see such problems immediately. without, it's a bit harder 2011-01-17 11:02 (well, spectrum analyzer or equivalent. but i guess adam didn't get an usrp for christmas ;-) 2011-01-17 11:03 can you use an already tested board in order to measure spectrum from other one? 2011-01-17 11:04 kristianpaul: no, it doesn't have a high enough resolution. not in the frequency domain and also not quite on the signal strength. 2011-01-17 11:05 well, maybe you can cheat with the crystal trim. lemme check ... 2011-01-17 11:05 so, no hope for a self test in the future? 2011-01-17 11:07 self tests never do more than 'fail / could be ok' seperation 2011-01-17 11:08 at least a self test between atusd atusd, 2011-01-17 11:09 measure some spected variables, rates, .. 2011-01-17 11:09 trim range is about 150 ppm. that would be 0.37 MHz. i can step the constant wave +/- 0.5 MHz around the center frequency. and i can step the center frequency by increments of 5 MHz. 2011-01-17 11:10 so, no, no way to cover the spectrum continuously 2011-01-17 11:10 :-( 2011-01-17 11:10 .s 2011-01-17 11:10 wpwrak: you dont have anything calibrated right? 2011-01-17 11:11 meaning 'more than usrp' 2011-01-17 11:11 roh: nada 2011-01-17 11:11 kristianpaul: test atrf-atrf is my plan B 2011-01-17 11:12 kristianpaul: it's good to have anyway. but being able to see the spectrum would help to pinpoint problems. after a while, you'd know where the problems usually are and a simple pass/fail will be sufficient. 2011-01-17 11:13 maybe we should make the usrp our official test tool? 2011-01-17 11:13 what is the cheapest/easiest way to test, still with reasonable quality and speed? 2011-01-17 11:13 usrp is certainly not the worst possible choice 2011-01-17 11:14 wpwrak: yeah i agree (help to pinpoint initials problems) 2011-01-17 11:14 what minimum usrp equipment would we need? 2011-01-17 11:14 the cheapest are probably these guys: http://www.aaronia.de/produkte/spectrum-analyzer/ 2011-01-17 11:14 minimum in terms of cost 2011-01-17 11:14 wolfspraul: about 100 km from where you are (if you're still near saarbruecken) :) 2011-01-17 11:16 minimum usrp: USRP1 (USD 700) plus RFX2400 transceiver (USD 275) 2011-01-17 11:17 plus a few bits and pieces you get at the electronic shop. (cables, antenna, gender changers, maybe an attenuator or two. probably less than USD 100 in total.) 2011-01-17 11:18 I'm still near Saarbruecken, yes, but leaving tomorrow morning, so too late to quickly go there now... 2011-01-17 11:18 which one of the aaronia.de devices would suffice for our needs? 2011-01-17 11:19 wpwrak: if the aaronia is better than what you have... the cccb has a device from them 2011-01-17 11:19 i can just borrow it 2011-01-17 11:20 the 2025E V3 should be sufficient. the 4040 V3 is a bit nicer, though. 2011-01-17 11:20 roh: aaronia should be better, yes. calibrated, for one thing :) 2011-01-17 11:20 roh: they probably also have a wider range 2011-01-17 11:21 oh, the usrp1 has a fairly narrow frequency range. usrp2 is much better, but also at twice the price 2011-01-17 11:22 (narrow range) in terms of what you can see at a time without tuning 2011-01-17 11:22 if you include tuning, then they're the same 2011-01-17 11:24 xiangfu: you can have a look at gcc-mips/Makefile for some STRIP usage example 2011-01-17 11:24 Einsteiger 2011-01-17 11:24 .. 2011-01-17 11:25 roh: not sure how good the readout is on the aaronia. 2011-01-17 11:25 :-) 2011-01-17 11:25 kristianpaul: http://spectran.comÂ Â Â Â :-) 2011-01-17 11:26 what's bad about the aaronia things is that they come in this huge aluminium box. so shipping ought to be very expensive. 2011-01-17 11:28 you should use RF screening in your place 2011-01-17 11:29 kristianpaul: you mean to make sure nobody mistakes it for an apartment where people live ? ;-) 2011-01-17 11:29 :-) 2011-01-17 11:32 (shipping of the alu box) to argentina, it would be EUR 280. (i once investigated the possibility of getting one of their devices, but eventually gave up. the peripheral costs (expensive shipping, special customs treatment, etc.) just add up to basically doubling the price) 2011-01-17 11:33 (doubling) and then add taxes. so final cost in .ar = 3 x .de price 2011-01-17 11:45 wpwrak: did you have to pay customs fees for the 3 nanos we sent recently? 2011-01-17 11:45 wolfspraul: i have USRP2 + XCVR2450. the XCVR2450 is a bit more expensive than the RFX2400 and also has the 5 GHz band in addition to 2.4 GHz. (the RFX2400 does 2.3-2.9 GHz) the XCVR2450 also has twice the transmit power of the RFX2400. 2011-01-17 11:46 wolfspraul: what the XCVR2450 doesn't have (and which i didn't find out until later) is simultaneous TX and RX. that would have been useful for my antenna testing. 2011-01-17 11:46 why would the spectrum analyzer create 'special customs treatment' like you wrote? do you know specifics? 2011-01-17 11:46 (3 nanos) yes, of course. 2011-01-17 11:47 how much did fedex charge you in total? 2011-01-17 11:48 (customs treatment) there is a USD 1000 limit for the value of private imports that arrive via courier or regular mail. anything above it gets the "commercial" treatment. 2011-01-17 11:48 there's one exception: EMS has a limit of USD 3000. 2011-01-17 11:49 that's why i used EMS when i bought the USRP. went smoothly, no problem. (non-EMS couriers get better taxes, though. about 30% on total value while EMS and regular mail get 50% on the value of the goods.) 2011-01-17 11:50 alas, there is no EMS in germany and aaronia aren't present in any other country. 2011-01-17 11:51 thus, i would have to involve a commercial importer to handle these things for me (or take my chances at bribery) 2011-01-17 11:52 the commerical importer would of course want a cut. and then the customs broker charges for their part, too. 2011-01-17 11:53 that suckx 2011-01-17 11:53 would it be easier to buy some used lab equipment from the us? 2011-01-17 11:54 for the bens, fedex charged ARS 534. that's about USD 130. 2011-01-17 11:54 via fedex, charges are in the 30-40% range. 2011-01-17 11:55 roh: used equipment is very bad. apparently, you're not allowed to import that at all. 2011-01-17 11:55 wpwrak: ah.. paperwork... then its refurbished 2011-01-17 11:56 roh: sure, if it looks new and the documentation doesn't say anything suspicious, then it ought to be fine 2011-01-17 12:09 wpwrak: customs in argentina look like a lot of fun... 2011-01-17 12:10 don't those people have anything better to do? 2011-01-17 12:11 lekernel: oh yes, they do. and it gets particularly bad if the thingy you're getting connects to mains. unless it happens to have the not very common certification for argentina, you can't import it without getting it certified first. they also refuse to just remove the power supply from the shipment and let the rest pass, although i'm not sure whether this is just them being bastards or some restriction with a legal basis. 2011-01-17 12:12 lekernel: oh, i think they're doing quite well. whenever they can create a problem, this usually means money for them. so they have every incentive for being attentive. 2011-01-17 14:08 http://gallery.slashhome.se/main.php?g2_itemId=3603 2011-01-17 14:10 i call it pyClock for now, it uses pygame to draw a binary clock 2011-01-17 14:14 Ornotermes: nice! I don't get how the picture of the chain and the software connect together? 2011-01-17 14:17 wolfspraul: its the default image for links in Gallery2 2011-01-17 14:17 ah !, you get me on that too :-) 2011-01-17 14:18 http://visual6502.org/ 2011-01-17 14:18 try the javascript simulation :) 2011-01-17 14:20 wow 2011-01-17 14:20 Ornotermes: ah OK, I got it now :-) I was distracted by the photo, didn't understand that this was the source tarball :-) 2011-01-17 14:21 of course this is very nice work, we'll try to package it and include in future images... 2011-01-17 14:21 is it ok that I upload the screenshot into the wiki? http://gallery.slashhome.se/main.php?g2_itemId=3604 2011-01-17 14:22 ofc :) 2011-01-17 14:22 ok, just asking because you have those nice license tags, but this one is not tagged... 2011-01-17 14:22 well great, this is a very nice clock! 2011-01-17 14:22 not yet :P 2011-01-17 14:23 all photos including RAW images is included for those interested, as CC SA 2011-01-17 14:25 6502 <-- And you are suposed to track visually all changes that happen between register? 2011-01-17 14:32 commodore times.. , i wish at least have had a Z80.. but no, anyway here i'm i got a nanonote :p 2011-01-17 14:33 kristianpaul: WYSIWYG at its best :) 2011-01-17 14:45 kristianpaul: ha - I just experience what you said the other day about the icons looking too good! 2011-01-17 14:46 built the latest image from source, just reflashed my nano. 2011-01-17 14:46 a lot of icons look really juicy now, but when you click on them you land in this text desert :-) 2011-01-17 14:47 wolfspraul: ;-) 2011-01-17 14:50 I think is opositive feeling compareted when you're testing the last bleeding edge Jlime version 2011-01-17 15:01 kristianpaul: you think the jlime icons are too austere ? 2011-01-17 15:03 wpwrak: no 2011-01-17 15:05 i mean last jlime (fro oe upstream) have some nice looking icons and theme, something similir about what is hapening with gmenu2x right now, but of course, with diferent result after you launch the app :-) 2011-01-17 15:06 wpwrak: jlime muffinman is okay for me 2011-01-17 15:06 aah. yes, the applications are nice. 2011-01-17 15:06 thats the point ! 2011-01-17 15:39 [commit] Werner Almesberger: introduced 8_10-card.lib as politically correct alternative to usd-card.lib http://qi-hw.com/p/kicad-libs/b205efc 2011-01-17 15:45 [commit] Werner Almesberger: use 8_10-card.fpd instead of usd-card.fpd http://qi-hw.com/p/f32xbase/32946a5 2011-01-17 15:45 [commit] Werner Almesberger: use 8_10-card.lib instead of usd-card.lib http://qi-hw.com/p/f32xbase/b6cfb82 2011-01-17 15:45 [commit] Werner Almesberger: c2ben.sch: finally, rename the title as well http://qi-hw.com/p/f32xbase/d915463 2011-01-17 15:45 [commit] Werner Almesberger: c2-ben.c: title comment still claimed the driver was for GTA01/02 http://qi-hw.com/p/f32xbase/a3a31f4 2011-01-17 15:49 [commit] Werner Almesberger: now that all users of usd-card have been changed to 8_10-card, remove it http://qi-hw.com/p/kicad-libs/20a669c 2011-01-17 15:50 and thus the great purge ends 2011-01-17 15:51 well, almost. it still lingers at a few not so important places 2011-01-17 17:06 http://lekernel.net/presentations/Milkymist_FOSDEM2011/milkymist_fosdem.pdf 2011-01-17 17:13 gap slide +1 2011-01-17 17:14 lekernel: where did you find the B/W picture on page 3? 2011-01-17 17:15 oh are VJ too ! 2011-01-17 17:15 kristianpaul: that explains it :) 2011-01-17 17:16 ? 2011-01-17 17:16 explains what? 2011-01-17 17:16 lekernel: is the kludgy software what you found or what you wrote ? 2011-01-17 17:16 "PC towers feel heavy when carried around at 5am" 2011-01-17 17:16 no latops in that time? :-) 2011-01-17 17:17 s/ltops/laptops 2011-01-17 17:17 kristianpaul: maybe that osborne thing :) 2011-01-17 17:18 Ornotermes: do you really want to know? ;) 2011-01-17 17:18 wpwrak: wtf? 2011-01-17 17:18 lekernel: page 2, 2nd bullet. what does it mean ? 2011-01-17 17:18 ha, quoting OS X, apple will follow you close now 2011-01-17 17:19 :-) 2011-01-17 17:19 wpwrak: software that I wrote, yes 2011-01-17 17:19 nevre released isnt? 2011-01-17 17:19 or rather hacked together 2011-01-17 17:19 lekernel: is there any particular reason why i wouldn't? :P 2011-01-17 17:20 lekernel: maybe you should clarify this in the slide. e.g., that you're not lambasting other people's work 2011-01-17 17:20 wwo slide 13, why my flicknoise dint look like that?? 2011-01-17 17:20 "Boots and gets ready in seconds" 2011-01-17 17:20 yeap ! 2011-01-17 17:21 "Extra features" 2011-01-17 17:21 Where is that part? 2011-01-17 17:21 kristianpaul: is your m1 booting again? 2011-01-17 17:21 I'll like watch it at leat :-) 2011-01-17 17:21 wolfspraul: since yday 2011-01-17 17:22 flashed from scratch :-) 2011-01-17 17:22 motion vectors <- kinetic ? ;-) 2011-01-17 17:23 kristianpaul: you should read a milkdrop doc, this will tell you about those 2011-01-17 17:23 i'll explain them briefly, but not on the slide 2011-01-17 17:23 ah ok... 2011-01-17 17:24 btw my bios still saying PCB revision 1 2011-01-17 17:24 lol slide 24 ;-) 2011-01-17 17:24 +1 what next of course :-D 2011-01-17 17:25 rc1 was revision 0 2011-01-17 17:25 lekernel: slide 6, "Trend for ``open hardware''" also is a little ambiguous, because it states a process without defining a point of reference. e.g., it could be "this should happen", "this is beginning to happen", or "this is well under way" 2011-01-17 17:26 "X11 is a pain on embedded systems" 2011-01-17 17:26 try get a quote froma Jlime user :-) 2011-01-17 17:27 wpwrak: that's a complex question that wouldn't fit in a single line of a slide 2011-01-17 17:28 lekernel: slide 7, looks a little odd. isn't this less a gap than more a hole. e.g., below a certain level, there simply isn't/wasn't anything non-proprietary 2011-01-17 17:29 diagrams like this are _always_ wrong, and I think it's still worthy to place things like Kicad, PCB designs and Qcad on it 2011-01-17 17:30 lekernel: (slide 6) maybe something like "Recently, trend towards" or such ? or maybe "also" instead of "trend", because "trend" suggests a broad movement, which we probably don't have yet 2011-01-17 17:32 lekernel: (always wrong) true ;-) perhaps a more "correct" layering would be sw on top, then traditional, then one block with hdl, semi impl, and semi manuf, horizontally divided. since they're a linear process, you could show this with angled ("arrowish") boxes, e.g., Digital design > semi impl > semi manuf 2011-01-17 17:32 depends. arduino folks claim they're doing "open hardware", and "arduino" has surpassed "microcontroller" in terms of google search volumes 2011-01-17 17:32 check on google trends if you don't believe me :) 2011-01-17 17:33 can you put that on slides? 2011-01-17 17:33 no, because I don't think Arduino is really open hardware 2011-01-17 17:34 lekernel: googlefight.com disagrees :) 911k vs. 528k. but it's still impressive. okay, so trend is good :) 2011-01-17 17:34 googlefight returns the number of pages 2011-01-17 17:34 not how many times people search for stuff 2011-01-17 17:36 as I said, it's a complex question 2011-01-17 17:36 so more are searching but they're finding less :) anyway, your point stands 2011-01-17 17:36 yup, it's exactly that :) 2011-01-17 17:37 i think the key is "opening". even with a proprietary chip, if you open the schematics, layout, and such, you're making one step towards more openness. at least today. at some point in the future, this may not be enough anymore. 2011-01-17 17:38 oh, well 2011-01-17 17:38 slide 9 is confusing. "what makes it SPECIFIC ?" ... "contains a GENERAL-PURPOSE". sounds like a contradiction :) 2011-01-17 17:39 Atmel and TI usually have rather open schematics and layouts for their reference designs, don't they... 2011-01-17 17:40 lekernel: )open ref designs) yup, and it's a good thing. there are also those guys who don't do that. NDAs for every scrap of information. 2011-01-17 17:42 lekernel: did you intentionally not capitalize the "it" on slide 12, 2nd bullet, 1st sub-item ? ("it will even disappear") 2011-01-17 17:42 hum, no 2011-01-17 17:44 lekernel:also slide 12, "If you get it right" is nicely ambiguous. one could read it as "if you don't get it right (the design/chip/whatever), people have to mess with the complexity they shouldn't have to mess with" ;-) 2011-01-17 17:45 lekernel: also on slide 12, in the last item "It is actually easy." seems redundant. 2011-01-17 17:45 lekernel: slide 13, maybe mention which GPL version ? (if that's not messy) 2011-01-17 17:48 lekernel: slide 16, might be useful to explain what "patch" means in this context (if that's straightforward, and you're not showing it in some other way at the same time) 2011-01-17 17:51 lekernel: slides 15 through 18 could perhaps benefit from still image / stylized examples, also such that you don't need to explain visual concepts with words, which is tricky in the best of cases 2011-01-17 17:52 lekernel: slide 20, s/not a FPGA/not an FPGA/ 2011-01-17 17:53 lekernel: since you mention CC-BY-SA on slide 23, how about putting it somewhere on slide 21 as well ? 2011-01-17 17:54 ah, I see you're picking up the old and rare "a FPGA" vs. "an FPGA" troll :) 2011-01-17 17:54 nice title on slide 24 ;-) 2011-01-17 17:55 well it's fosdem embedded devroom. filled with nerds. 2011-01-17 17:55 lekernel: well, it depends on how you pronounce it. eff-pee-gee-aye or fp'ga. the latter sounds like klingon :) 2011-01-17 17:56 lekernel: if you say "eff-...", it's "an". klingons use "a" 2011-01-17 17:57 a friggin' field programmable gate array!11 ;) 2011-01-17 17:58 slide 25. the first bullet seems to want an adjective that says what sort of barrier it is. or maybe you just mean "SDRAM controllers are hard to design" ? 2011-01-17 17:58 well, getting the SDRAM to work puts an end to many FPGA projects 2011-01-17 17:58 steve|m: alright. then the speaker stops and has a glass of water ;-) 2011-01-17 17:58 lekernel: suicide ? death by old age ? ;-) 2011-01-17 17:59 both industrial and hobby... 2011-01-17 18:02 perhaps what also makes the SDRAM barrier sound wrongish is that it's not actually the SDRAM but the interface. (e.g., one could interpret the sentence that SDRAM is too expensive, hard to source, etc.) 2011-01-17 18:02 I just call it a peripheral now (the fpga), with currently bad free software support 2011-01-17 18:02 like a graphics card or modem in the 90's 2011-01-17 18:03 and the other problem with the sdram statement is that it sounds boastful 2011-01-17 18:03 never a good idea that you claim you can do something that others can't 2011-01-17 18:03 that's actually a very good way to get them try harder to proove you wrong... 2011-01-17 18:03 ;-) 2011-01-17 18:05 maybe a neutral characterization of the main difficulty/class of difficulties could help to illustrate the issue. i think the rest of the slide is okay, not too boastful 2011-01-17 18:06 slide 27, floating point pipeline_s_ ... maybe make "N [parallel] pipelines" a separate bullet ? 2011-01-17 18:09 lots of bad vibes in slide 29. crappy, bugs, dirty, ... you could defang it with a quote of one of the authors himself would characterize it with such words 2011-01-17 18:10 slide 27, btw, BusyBox is camel-cased 2011-01-17 18:12 slide 31 also has "a FPGA" 2011-01-17 18:12