#qi-hardware IRC log for Tuesday, 2013-03-05

qi-bot[commit] Werner Almesberger: ircstat/ML: update for 2/2013 (master) http://qi-hw.com/p/wernermisc/9d784c216:44
kyakwow, they managed to fuck up wireless really bad in 3.817:06
wpwrakoh ?17:10
kyakstarting from "capabilities/regulatory prevented using AP HT/VHT configuration, downgraded" and ending with "fail to flush all tx fifo queues"17:10
kyakall kinds or weird messages, random deauth from AP, reboot helps and things like this17:11
kyaki'm wondering, do they have quality control at all?17:12
kyaklike, some sort of regression testing at least?...17:12
kyaki still remember how my trivial patch didn't pass their weird "patch submission system?:17:13
kyakand still we see that other patches easily get in, and cause big problems17:14
kyaklike, see, this one was supposed to fix some bug: https://patchwork.kernel.org/patch/2123681/17:14
kyakit didn't take long before another fix for this fix had to be issued: http://comments.gmane.org/gmane.linux.kernel.wireless.general/10426817:15
kyak"When the if statement is rewritten as 'if (A || B)', it turns out that A is 17:17
kyaktrue, and B is false."17:17
kyakyeah, you have to teach kernel developers these things now :)17:17
wpwrakhmm :)17:36
larscI'm innocent, I don't do wireless17:59
larscspeaking of bugs, wpwrak, can I blame all the vfat corrution I get on you?18:00
wpwrakonly fat corruption :)18:10
kyak--)18:19
kyakwpwrak: btw, are there any plans do upgrade atusb do the latest kernel?18:20
kyakfor what i've tried, it compiles and loads on 3.7 and 3.8, but doesn't work18:20
kyakand also kernel panics sometimes :)18:20
kyaki've packaged the atusb module and lowpan-tools for arch linux, and it was not so hard (i had to push some config options in default kernel config for arch, which were not there)18:21
kyakbut the problem is that at the moment it's all for nothing -\18:22
wpwrakyeah, i have that in my queue. actually to rewrite the driver and do more processing on the device itself (to avoid those expensive usb round-trips)18:25
kyakwpwrak: sounds great!18:28
kyakwait a minute.. usb?18:28
kyakah.. sorry :)18:28
wpwrakdid you mean atben ?18:53
whitequarkwpwrak: (cpld or small fpga between cpu and 8:10 card(s)) PSoC?20:40
larscstill trying to pitch that one?20:47
whitequarklarsc: hm? I don't even own a board20:48
larscjust kiding, but you seem to bring it up every now and then20:48
larscwhich I dont mind20:49
whitequarkwell I did try to bring it up once20:49
whitequarkthere was a neat kickstarter20:49
whitequarkthen I noticed a windows-only IDE or so and became disgusted20:50
larscunrelated. btw. https://www.coursera.org/course/vlsicad just started20:51
wpwrakwhitequark: naw, too little cpld in these. and you don't need the core.20:51
larscmight be interesting for some people in here20:52
wpwraklarsc: i actually like the concept of psoc. and they got more open-friendly with time. but there are issues, too ...20:52
whitequarkwell didn't they always have docs on bitstream?20:53
whitequarkbut then xilinx also does that20:53
wpwrakpsoc don't have a bitstream. they have (had ?) docs with all the registers. (that do the equivalent of the bitstream)20:57
wpwrakand the issues i mean aren't so much about openness but more about tininess and the things that come with it20:58
rohhm... ive seen those in the ebike controllers we use20:58
wpwrakalas, they're a rather non-ideal candidate for joining forces among the downtrodden, because they also firmly believe in windows20:59
rohyeah. there they are abused as a 3phase bldc brushless controller ,) thats about what one wants to do with em20:59
larschaha, "solving a 40 variables boolean equation with karnaugh maps takes 1,099,511,627,776 squares... there has to be a better way'21:06
wpwrakit helps a lot if you already know the solution :)21:09
larscwhich makes you wonder why the hell they let you solve hundreds of kmaps in university by hand, yet they never have more than a handful of variables, which doesn't really have any practical applications21:09
wpwrakwell .. perhaps to understand the structure ?21:09
wpwrakbut yes, "hundred" seems a little excessive. i don't think i ever encountered more than perhaps "tens"21:09
larscI might be exaggerating21:11
larscbut i met them in both CS and EE21:11
wpwrakthe two tend to overlap :)21:23
larscwpwrak: it's a bit as if they thought you addition that only works for numbers between 1 and 9921:26
--- Wed Mar 6 201300:00

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