#qi-hardware IRC log for Friday, 2010-12-24

qi-bot[commit] Xiangfu Liu: update debian/changelog http://qi-hw.com/p/xburst-tools/2319c7701:04
qi-bot[commit] Xiangfu Liu: update ChangeLog http://qi-hw.com/p/xburst-tools/0fc697201:04
xiangfu[xburst-tools] new version: http://projects.qi-hardware.com/index.php/p/xburst-tools/downloads/01:32
qi-bot[commit] Adam Wang: ignore $savepcb.000 and $savepcb.brd http://qi-hw.com/p/mmone-jtag-serial-cable/b6624a102:34
wolfsprauladamw_: hey, great patch! you slowly become a git expert, fantastic!02:43
adamw_he..just found those files seems temporarily generated. but I don't need them.02:45
adamw_the shop store in Kwuanhwa they said yes, the 300MHz probe is became more common.02:46
adamw_but well...since ours scope is only 100MHz, so I don't need to buy a 300MHz probe.02:47
adamw_but sure ideally an input probe with more wide range is always good for 100MHz scope. :)02:47
wolfsprauladamw_: how much was the 300mhz probe you found?02:59
adamw_ntd 1300 roughly.02:59
adamw_100MHz >> ntd 75003:00
wolfspraulhmm. the difference is just 20 USD. Should we just buy the 300mhz one?03:01
Action: xiangfu will remove the ImageBuilder and Toolchain from config.full_system.03:06
adamw_why not? if you want. :)03:06
xiangfuadamw_: ;-)03:06
wolfsprauladamw_: ok I think the difference on a 100mhz scope will be minor. So buy the one that is easier for you. Since you are back already, no need to go out to kuanhua again. buy online. but if it's easier on kuanhua, buy there. Your time is also important :-)03:08
adamw_I'll directly buy two 300MHz probes while go back home along my way. :)03:13
wolfspraulok great03:13
yizhangxiangfu: while i am in gmenu2x, press 'alt+ctrl+F2' will take me to the console. however i can only press 'alt+ctrl+F5' to get back to gmenu2x, 'exit' command will not do the same.03:28
yizhangxiangfu_:  while i am in gmenu2x, press 'alt+ctrl+F2' will take me to the console. however i can only press 'alt+ctrl+F5' to get back to gmenu2x, 'exit' command will not do the same.03:29
yizhangis this an intentional feature or is it a bug?03:29
xiangfu_yizhang: not a bug.03:29
xiangfu_yizhang: if you using 'alt+ctrl+F2' goto console. you must using 'alt+ctrl+F5' go back, there are different 'Terminal'03:31
yizhangthis is not a ash?03:31
xiangfu_in ctrl+alt+F2 is a ash :)03:36
wpwrak_(at&t pc) hehe ;-)09:50
wpwrak_kristianpaul: "no please no more discuss about changing NAND schema again ;-)" no discussion, just changes ;-)09:52
wpwrak_wolfspraul: doesn't the scope have three inputs, CH1, CH2, and trigger ? then you need three probes.09:54
wolfspraulhmm, good point don't know09:55
kristianpaulwpwrak_: :)09:55
wolfspraulwpwrak_: btw, --plot works as well (in pcbnew), now I'm doing --drc, and after that probably --pos (module positions)09:56
wolfspraulevery one is different though, thanks to the magic myriad C++ dependencies...09:56
wpwrak_wolfspraul: (not that the scope vendors would usually include a probe for trigger - but hey, accessories are a nice market, too :)09:56
wolfspraulif I get stuck somewhere, I will cleanup what I have so far and upload the patch, then you can try. I think what you need the most is --drill and --plot in pcbnew, and I have those, need your feedback.09:57
wpwrak_wolfspraul: excellent ! i admire your c++ skills already ! :)09:57
wpwrak_wolfspraul: and your persistence in the face of ugly code :) if that IT and hardware thing goes sour, you can always have a stellar career in waste management ;-)09:58
wolfspraulso I do --drc now, then --pos, then --svg, --vrml, --bom09:59
wolfspraulsince I'm at it, I may as well try to get it all out programmatically, could be helpful later09:59
wolfspraulI will also go back to eeschema and add netlist10:00
wpwrak_great. the next step would be to apply your patches to the latest svn version. and then see how they like them. with so much cool stuff, it'll be very hard to just dismiss it :)10:01
wolfsprauloh one by one10:01
wolfspraulfirst we get it applied to the old 2448, then you try to use it, feedback etc.10:02
wolfspraulyou may very well run into segfaults, or you need additional cmdline switches, etc.10:02
wolfspraullet's first make it work well10:02
wolfspraulthen, if we update our base version, of course there will be work in upticking those patches. but that's good stuff so I can slowly clean it up more.10:02
wolfspraulin parallel I can write to the kicad devel list and see whether they want to accept a few patches that lead to a cleaner separation of UI and data processing10:03
wolfspraulit's not too bad, but in a lot of places the two are still intertwined...10:03
wpwrak_yeah, once you've done it the first time, it shouldn't be too hard to update it.10:03
wolfspraulah, don't say that10:03
wolfspraulthose c++ dependency jungles are very fragile10:04
wpwrak_i think cleanup would be great. the lack of a proper separation is what prevented me from going any further with the command-line-driven file output10:04
wolfspraulfor example the last few hours I am trying to trace down a weird segfault somewhere in the drc code10:04
wolfspraulif I run through the UI, it works10:04
wolfspraulbut when I try to run the same (well, probably not exactly the same) logic bypassing the gui stuff, there is a segfault somewhere10:04
wolfspraulbut I try to make it work fast10:05
wolfspraulespecially since --drill and --plot are ready for testing. so if I get stuck too long, I'll upload that first.10:05
wpwrak_ah, one important thing: you need to trigger the zone filling before outputting the layers (drc will do it)10:05
wolfspraulI thought --drc is fast, that was wrong...10:05
wolfspraulzone filling?10:05
wpwrak_ground areas and such10:05
wolfspraulno idea :-) let's see when you have it running whether it's usable or not. need a bit more time...10:06
wpwrak_of course, doing the zone filling there will also avoid the nasty problem of not having the zones filled (they make things very slow, so for working on the layout, it's convenient to unfill the zones), and then plotting - because kicad doesn't automatically fill the zones in this case. if you don't notice in time, you make a board without ground/etc. areas ...10:08
wolfspraulwell one by one. I first try to introduce some scriptability into a lot of the GUI driven codepaths.10:10
wolfspraulthen we can look at what it actually writes, and finetune more parameters.10:10
wpwrak_very nice :)10:10
wolfspraulit's not too bad, really.10:11
wolfspraulI like their work.10:11
wpwrak_in 2011, kicad will be twice as good as it was in 2010 ;-)10:11
wolfspraulthey do try to separate stuff.10:11
wolfspraulof course there are still a lot of pointers here there, back forth, etc. and 'friend' and what not, but definitely it's a good work.10:11
wpwrak_(separate stuff) good. maybe they also cleaned things up a bit more since i last looked at the code10:12
wolfspraulhe, my --drc just finally wrote a complete report without segfault.10:12
wolfspraulbut a lot of flickering since I left the entire GUI codepath in. Now I only need to see when the segfault reappears as I remove the GUI stuff...10:12
wpwrak_what bothered me most were some very long and deeply indented functions - and c++'s inability to easily break them into more manageable pieces10:13
kristianpaulwpwrak_: (probe for trigger) i dint knew that, there is something distinctive on it?10:17
wpwrak_kristianpaul: it's the same kind of input. you basically have 2.5 channels. so if you have a trigger signal and want to watch two other signals, you can do that. you don't need this a lot, but when you need it, it's nice to have.10:19
kristianpaulJay7: fb menu, well there is gmenu2x, do you know it?10:41
kristianpaulbasically an app lanucher, with wallpaper and battery indicator is needed10:42
kristianpaulJay7: http://en.qi-hardware.com/wiki/Applications some pics of gmenu2x and other apps10:45
Action: FrankBlues waves13:59
Jay7kristianpaul: well, I see now14:27
Jay7it's a bit other than I've imagined14:28
kristianpaulJay7: what you imagined at first place?16:09
Jay7kristianpaul: just something like we have already in kexecboot16:21
Jay7kristianpaul: gmenu2x is more like desktop manager16:21
kristianpaulTwo things:16:23
kristianpaul1). kexecboot will be really usefull for the nanonote, as there is a jlime OS that can be shipped on usd, so dual boot is a TODO16:24
kristianpaul2). (gmenu2x is more like desktop manager) sure... thats why simpler alternatives are wellcome :-)16:24
kristianpaulwich simpler i meant, not to be mean like a desktop manager at all, just a simple application launcher16:25
Jay7btw, I've thought already about reporting battery level in kexecboot ;)16:27
Jay7well.. christmas/new year holidays almost started.. so I hope I'll do something soon ;)16:29
kristianpaulOrnotermes: hey :)16:39
kristianpaulSomething new to tell with the flash memory bus hacking? :-)16:39
Ornotermessadly, no16:41
Ornotermesit's a bit in the future16:41
Ornotermesbut i have a stm32 devboard with a mcu that should be able to have a 25MHz clock16:42
kristianpaulJay7: http://ur1.ca/2o0g016:42
kristianpaulstm32?.. :/16:42
kristianpaulwhy not a M1 ? :-)16:42
Ornotermesany way, merry christmass to you all (we celibrate it the 24th is sweden)16:43
Jay7we celebrate it the 7th Jan in russia :)16:44
Jay7so.. new year is first holiday here..16:44
kristianpaulThanks Ornotermes16:44
Jay7anyway, merry christmas to all who celebrating it now :)16:45
kristianpaulnot me ;-)16:46
Jay7.. and to all who ignoring it too ;)16:46
Ornotermessweden is quite odd when it commes ti christmas16:49
kristianpaulsys0m 0.01s19:49
kristianpaullol https://secure.wikimedia.org/wikipedia/en/wiki/Uboot21:37
kristianpaulsure this uboot is bettet that nanonote's one21:49
kristianpaul:s ./dumpif: double free or corruption (out): 0x0041100822:05
kristianpaulhmm is gone after replace calloc with malloc...22:35
qi-bot[commit] Cristian Paul PeƱaranda Rojas: Copy data to memory if sync register is set to high (work in progress) http://qi-hw.com/p/ben-gps-sdr/168516622:44
--- Sat Dec 25 201000:00

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