#qi-hardware IRC log for Thursday, 2010-12-23

wpwrakkristianpaul: why AA55AA55 and not AAAAAAAA or 55555555 ?01:02
wpwrakthat looks a bit suspicious01:02
yzhangi try to find out how to use NanoMap, but didn't find any readme file or documentation. does anybody know where i can find some documentation for NanoMap?03:17
yzhangnever mind, i find something here03:24
yzhangxiangfu: except for the info on your page, do we have any other docs?03:27
yzhangxiangfu: for NanoMap03:27
wolfspraulyzhang: did you press 'h'?03:27
wolfspraulwhen you press 'h', you see a nice little help menu popping up03:27
wolfspraulzoom in and out is '+' and '-'03:28
yzhangxiangfu: so h is to popup the help menu, not ?03:28
wolfspraulthe problem now is that it has no index, so you cannot search for a street name or other poi03:28
yzhangwolfspraul: thanks. on xiangfu's page said i should press '?'03:28
wolfspraulalso we are not very smart yet in which tiles (parts) of the map we are packaging, and how to reload/update missing tiles03:28
yzhangit ididn't work03:28
wolfspraultry 'h' first03:29
yzhangthere are also other feathers03:29
wolfspraulxiangfu already changed this so that the help menu will show by default after launching the app03:29
yzhangnot in 2010-12-14 release yet, or yes?03:31
xiangfuyzhang: not in 2010-12-14, I did that change after 2010-12-14.03:32
yzhangby the way, i accidentally went into direction page (i think it is) by press 'e', but i can't do anything there, except 'tap' between controls. i assume all these features don't work, correct?03:33
yzhangxiangfu: i see.03:33
wolfsprauldon't know never went there. you are pioneering :-)03:36
yzhangk, i will play it around and see what we have in there and what we can actually use03:37
xiangfuyzhang: "Alt + s" will exit the 'direction'. I think you pressed 'r'03:39
yzhangxiangfu: yes, 'r'03:40
yzhangxiangfu: where do you find all these controls?03:40
yzhangxiangfu: guessed or i can find it somewhere, in sources?03:40
yzhangxiangfu: i mean like 'alt+s' is to exit the direction page03:41
xiangfuyzhang: there is a 'underline'  under [S] in "Show map"03:42
yzhangxiangfu: ah, i see03:43
yzhangxiangfu: pretty good :)03:44
adamw_wolfspraul, while i translating *.pos & *.lst to the new smt vendor which they wanted me collecting qty in *.xls file03:58
wolfspraultranslation to xls, wonderful :-) well, once we get to this we can automate this on the server too, for more 'traditional' partners so they get their beloved .xls files...03:59
adamw_means that I manually fill them into *.xls Excel file...well03:59
wolfspraulplease do this: upload the .pos and .lst files as created by KiCad, plus the .xls file you create manually, to a new location on downloads04:00
wolfspraulthen we can look at this and automate it later04:00
wolfspraulonly when you are done with the conversion...04:00
wolfspraullike in downloads.qi-hardware.com/people/adam/pos_lst_conversion/...04:00
wolfspraulso you document the work you did, and we can look at it and see how we can automate it04:00
adamw_yeah...i suddently thought one thing that I always didn't check those 'value' & 'reference' in KiCad EESchema.04:01
adamw_yeah...surely I'll upload them all once I done. :)04:01
wolfspraulincluding the pos & lst please, so we can easily see what work you did04:01
wolfspraullike side-by-side...04:02
adamw_i used boom 'cat usb_jtag.ord' to see those reference and count.04:02
adamw_sure sure pos & lst.04:02
adamw_also *.lst itself I can see this file so that I can also know real qty by manually count.04:03
adamw_so my question is : if designer forgot to fill or write wrong 0402 or 0603 or value, can boom realize it? or designer suddently changed sch, but forgot to change info in PCBnew/CVpcb?04:05
adamw_then I already ordered parts.04:06
wolfspraulI understand your question/idea, but need to look into it.04:08
wolfspraulthe bom can also be exported from pcbnew04:08
wolfspraulyou are saying we could add some checks that the schematics and the gerbers match, for example in footprint04:08
wolfspraulI need to look into the specific files and how we can compare/match them for consistency. right now I believe the link between schematics and gerbers is weak, and not much checking goes on _between_ those two files.04:09
wolfsprauladamw_: did I understand you correctly?04:11
adamw_partial correctly. Not only include footprints, also those references/value.04:17
adamw_for examples, now I am counting the qty for those value are '100nF', the qty is 16pcs for 'C10, C13, C15, C18, C20, C21, C22, C23, C24, C25, C26, C28, C30, C32, C33, C35'04:18
adamw_later if designer changed C23 to be 82nF then the rules of generating '*.lst' file must EVERYTIME generate it from KiCad again.04:20
adamw_or if designer suddently changed C23 to 0603, then if I missed this info before I ordered parts..then ....bad04:21
adamw_so seems that at least: 1, suppose believing the layout footprint is fixed already and I 'believe' it was right. 2, everytime I must git clone then generate .lst04:23
adamw_well...we can discuss later..i need to finish my translation work and send to them.04:24
kyakwolfspraul: didn't know Spectec were like that..05:04
kyakwolfspraul: in wiki, it read some time ago that "In October 2009, Renesas released driver sources licensed under the GPL (version 007 for Linux 2.6.24), and a binary firmware (v0.2f)"05:05
kyakso i just didn't care (and don't care) who bought whom and whose name is what :)05:05
wolfspraulsure, all good05:06
wolfspraulthe Renesas release took months of me lobbying05:06
wolfspraulI think that whole team has been disbanded since.05:06
wolfspraulit's a long story05:06
kyakit's good you could make it05:07
wolfspraulshortly before death :-)05:07
wolfspraul(their 'business' death)05:07
kyakit's just a coincidence, right? ;)05:08
wolfspraulwell I think what they tried was extremely ambitious05:08
wolfspraulI never spoke to the founder, he already left when Renesas bought the remainders of the startup05:09
wolfspraulso I'm not exactly clear what it was that they tried and why they thought they could be successful05:09
kyakand what do you think they failed? were trying to bite too much?05:10
wolfspraulRenesas only bought it for the people, and to absorb the wifi technology into their SoC monster chips, including everything, to compete with Qualcomm etc.05:10
wolfspraulI don't know what they tried to accomplish :-)05:10
wolfspraulso I also have no opinion, neither my own nor relayed, on why they failed05:10
kyaktheir site is still running, and i had some discussions with them this summer before byuing Ben05:10
kyakperhaps these were already people from Renesas..05:11
wolfspraulwho are you talking about? renesas?05:11
wolfspraulyes for sure05:11
wolfspraulI forgot the name of the startup they bought.05:11
kyaki thought i was talking with spectec05:11
wolfspraulwell, maybe you were?05:11
wolfsprauldon't know05:11
wolfspraulspectec is a small company, 20-30 people, specializing in making external memory cards that include not memory, but other IC-based features05:11
kyakdoesn't matter know, they suggested the price for SDW-823 much higher than i got from zhangyu :)05:12
wolfspraul'low quantity' in comparison to the memory card makers05:12
wolfspraulso they find an IC they think could work in a card, then they build a card around it05:12
wolfspraulnormally in collaboration with the IC maker, the driver will come from the IC maker, and the IC maker needs to support Spectec for any more serious technical challenges05:12
kyaki think it's pretty awesome anyway05:12
wolfspraulso Spectec's part is to know how to make this work in the (small) format of an external card05:13
wolfspraulpackaging, 'mechanical', antenna? those things...05:13
qi-bot[commit] Xiangfu Liu: cleanup code style http://qi-hw.com/p/xburst-tools/24aa5a205:24
qi-bot[commit] Xiangfu Liu: xburst_stage2: don't skip bad block when nand erase http://qi-hw.com/p/xburst-tools/6b3f14b06:12
qi-bot[commit] Xiangfu Liu: xburst_stage2: don't skip bad block when nand erase http://qi-hw.com/p/xburst-tools/0e92eb806:41
kristianpaulwolfspraul: because data is a 1, 0, loop http://kristianpaul.org/~paul/tmp/acquisition.png06:46
kristianpaulwolfspraul: oh sorry, i meant wpwrak06:46
kristianpaulwpwrak: i updated pic http://kristianpaul.org/~paul/tmp/acquisition.png06:46
kristianpaulbut yes, i'm going to make a better simulation (this firstone is was too laz for me :-))06:47
wolfspraulkristianpaul: how did you create that png?06:50
wolfspraul(which software)06:50
kristianpaulwolfspraul: gtkwave06:52
kristianpaulthen screenshot from the selectec area :-)06:52
wolfspraulah OK, nice. never worked with gtkwave so couldn't tell right away :-)06:54
kristianpaulwpwrak: "AAAAAAAA" thats a *huge* number i'm just shifting 8 bits registers so no more that a "AA"06:55
kristianpaulme neither, first time06:55
kristianpaulworks well so far i can tell :-)06:55
qi-bot[commit] Xiangfu Liu: update version to 201012 http://qi-hw.com/p/xburst-tools/8ad5b5808:34
qi-bot[commit] Andres Calderon: 5V Boost for the USB Host has been selected http://qi-hw.com/p/xue/360398e08:45
wpwrakkristianpaul: (AAA...) what i mean is that, if you have a pattern of 010101..., it should either be all 5 or all A, but not change between them, no ?09:04
wpwrakxiangfu: (xburst-tools fixed) great ! let's hope this solves the debian nightmares :)09:10
xiangfuwpwrak: yes. :)09:10
wpwrakxiangfu: by the way, why did this affect debian but not openwrt ?09:13
xiangfuwpwrak: because when reflash debian it's always use "nerase 16 4080 0 0".09:14
xiangfuwpwrak: but in openwrt.we use "nerase 16 1024 0 0 " which will not overflow the nand block count.09:14
xiangfuso when there is some bad block in [16 ~ 1024] it will overwrite the data partition.09:15
wpwrakaah, so they use the whole nand for rootfs andf don09:15
wpwrak't keep a "user" partition. i see09:15
xiangfuwpwrak: the yes. debian kernel use the old nand layout.09:16
xiangfuthe bug is 'nerase' always try to erase good block. skip the bad blocks.09:16
wpwrak... which is a bad idea if you have fixed partition sizes.09:17
wpwrakbtw, i like the fixed partition sizes. despite everything, they're much easier to handle than the dynamic insanity we had at openmoko.09:18
xiangfuyes. agree. it's simple and easy maintain. I have read some text form you. :)09:20
wpwrakhehe ;-)09:21
larscwpwrak: the gpio-charger driver is for chargers which indicate whether they are charging or not through an gpio.09:44
wpwraklarsc: ah, for the SE9016. i see.09:52
kristianpaulwpwrak: (patter), the shift is continuos so is normal see bits moving from LSB to MSB every clk cycle09:54
wpwrak(charger) i wonder why we need D4 there ...09:54
kristianpaulxiangfu: the last tarball (xbusrt tools) is up to date as the debian package?09:55
wpwrak(aa/55) ah, so one of 55/aa is not valid. i see.09:56
xiangfukristianpaul: I only upload a new debian package here : http://projects.qi-hardware.com/index.php/p/xburst-tools/downloads/61/09:57
xiangfukristianpaul: not upload hte source tarball yet.09:57
wpwrakbut ... isn't mcu_data what the cpu is supposed to read ? i would think that this should only change for each "word" (byte or whatever)09:57
wpwraki guess that would be mcu_clk09:57
kristianpaulwpwrak: mcu_data should be read just when mcu_clk bit 0 is set to hihg10:01
wpwrakis this buffered somewhere or does the cpu have that quick ?10:07
kristianpaulno fuffer10:08
kristianpaulI was thinking implement a FIFO10:09
kristianpaulor make mcu_data 16 bits wide and implement two mcu_data registers then10:10
wolfspraulwpwrak: --drill works :-) now onto --plot ...10:10
wolfspraulI start to get the hang of this, I think my rough plan is to first do a lot more of the various file types it can generate10:11
wolfspraulI think everyone will be slightly different in terms of how to hook into KiCad, how the classes are organized etc.10:11
wpwrakwolfspraul: wheee ! i'm very much looking forward to doing the whole cam files chain with just "make"10:11
wolfspraulthen maybe over time as we upgrade KiCad, I will move the patch forward and try to hook it into KiCad better10:11
wolfsprauland in parallel I can see on the kicad devel list whether someone is interested in a few small patches10:12
wolfspraulthere is no need to talk them into the whole idea, what would help a little already is a cleaner separation in some places between GUI and data handling.10:12
wpwrakyes, things are quite heavily interconnected10:13
wolfspraulso as long as they are open minded to this very abstract, yet valuable goal, we can make the patch easier to maintain over time10:13
wolfspraulnot that bad actually10:13
wolfspraulso let's see, I try to get things to work that's the highest priority10:13
wolfspraulso many more, --drc --svg --bom etc.10:13
wpwrakyup. worst case, they may reject it but it may shame them into doing it "right" themselves :)10:14
wolfspraulwell I have a plan, just need to find enough time10:14
wolfspraulI think the source quality is not bad10:14
wpwrakhow do you like the indentation style ?10:14
wolfspraulfirst I make the -- stuff work, then I try to find the cleanest way to insert the patch.10:14
wolfsprauland then I try to submit patches to them that would make a cleaner insertion possible.10:14
wpwrakalso, some functions are horribly long. but not all of them, of course10:14
wolfspraulah I don't care10:14
wolfspraulmental flexibility...10:15
wpwrakheh :)10:15
wolfspraulof course the C++ stuff is horrible. Well, I used 'features' too, like 'friend'10:15
wpwrak /* false */ friend10:15
wolfspraulanyway no need to waste time bashing on this lost cause10:15
wolfspraulno with 'friend' you can override the protected/private stuff10:16
wpwrakyes, that's why it's not very friendly :)10:16
wolfsprauloh of course, you can also override it in a .h file, even for binary libs. because in the myriad of C++ name mangling, protected/private is the one thing that is not reflected in the mangled names. go figure.10:16
wpwraks/friend/infiltrator/ :)10:16
wolfsprauloh it's super friendly, why not? we love C++, don't we?10:16
wolfspraulanyway, I'll get it to work first. --drill does already. Yes, we will get your entire cam workflow make-ready. that's the goal.10:17
wpwrakwonderful. thanks !10:18
wolfspraulthank me when it's working.10:21
wolfspraulthis myriad of inter-twined classes, argh.10:21
wolfspraulhwo can anybody take this serious?10:21
kristianpaulwpwrak: before try real sige data, i'm wroking on a simple counter in roder to simulate data and then verify if cpu is quick enought to cath it all10:30
wpwrakkristianpaul: great. let's see how bad it gets :)10:35
kristianpaulNow that sees apps like stardict,gmu and nanomap are wroking well, there is a easy way to quit then using same keys?10:36
kristianpaulmay be with F keys.. wait for a keystrok when receive it killall app :p10:37
kristianpaulstil kindof messy remenber how to quit every app10:37
kristianpaulas i saw yzhang was askinf some hours ago10:37
wejpno need to kill gmu, you can fully customize its button mappings, so just map exit to the key you want10:39
kristianpaulwpwrak: 8bits at least reduce troughput to the half (~1Mps ), and if still bad, i'll move to 16bits then10:39
kristianpaulwejp: i still wonder why i should press F1 then Alt+ s to quit gmu :S10:40
kristianpaulbut yes patch every apps seems more reasonable and clean :-)10:40
wejpyou don't have to press F110:40
wejpF1 is for the help screen10:40
kristianpauldid you tried last image release?10:41
wejpno, not yet10:41
kristianpaulyou should :-)10:41
wejpi will10:41
B_LizzardDid anyone see this, then?10:52
B_LizzardSorry if this is bad sport10:53
wpwrakB_Lizzard: looking cool ! btw, a style remark: it's written in the first person, but there's no indication who the author is11:05
B_LizzardAh, it's specified in the front page11:06
B_LizzardSorry bout that11:06
B_LizzardI am the author11:06
B_LizzardI know the writing is bad, but I got bored after a while for a full writeup11:07
wpwrakah, you fixed it already :) thanks !11:07
B_LizzardYeap, a bad mistake on my part11:08
B_LizzardThanks for the help, wpwrak :)11:08
B_LizzardNewer versions come with a J2ME environment which runs kinda well.11:09
B_LizzardOpera Mini supposedly works but I'm having trouble connecting to the net11:10
wpwrakthat's a sort of important feature :)11:11
qi-bot[commit] Andres Calderon: fped footprints added http://qi-hw.com/p/xue/e82363012:47
kristianpaulB_Lizzard: still no ethernet on that preview?14:35
kristianpaul(that stop me from use it...)14:35
B_Lizzardlarsc says it's not stable14:44
B_LizzardI can't remember where the patch is14:44
zrafakristianpaul: ping lars, no B_Lizzard15:45
andres-calderonHi wpwrak  and wolfspraul  I got a rookie problem with KiCad: I got a lot of error because of the netlist.16:10
qi-bot[commit] Andres Calderon: minor fix http://qi-hw.com/p/xue/786284f16:13
qi-bot[commit] Andres Calderon: cleanup http://qi-hw.com/p/xue/6a075dd16:13
wolfspraulI'm more rookie than you are!16:13
wolfspraulso let's send some prayers somewhere that wpwrak_ hears our suffering... :-)16:13
wpwrak_errors where ?16:14
wolfspraulif not, we shall forever burn in KiCad hell...16:14
wpwrak_i hear but i'm in and out a lot today. need to prepare stuff for tomorrow.16:15
qi-bot[commit] Werner Almesberger: f10/: seasonal fun project - distribute current to ten ignitors (2 x 22 R) http://qi-hw.com/p/wernermisc/3427d6916:17
wpwrak_among them, this little thing. no ben, not even active components, but fun anyway :)16:17
andres-calderonComponent "C66": Mismatch! module is [0402] and netlist said []16:17
andres-calderonComponent "C63": Mismatch! module is [0402] and netlist said []16:17
wpwrak_aaah ! ;-)16:18
andres-calderonreading the current netlist16:18
wpwrak_let's see ..16:19
wpwrak_when you invoke cvpcb, you see lots of components that don't have a footprint16:20
wpwrak_you should assign one16:20
andres-calderonok, That's what I always have.16:21
wpwrak_in fact, the underlying problem seems to be elsewhere ... wait a minute ...16:24
andres-calderonBut the .net file.  is no longer in the repository .. I thought it was deleted because it is a machine generated  file from the schematic.16:24
wpwrak_.net should be machine-generated, yes. having it just makes the problem16:27
andres-calderonso, i can  fix the .net file and then delete the .gitignore entry.16:28
andres-calderonThat's the solution? I have added the footprints in the schematic. But they have been ignored.16:28
wpwrak_the problem is that cvpcb apparently doesn't use the footprints in the schematic if there's a .cmp file16:29
wpwrak_unfortunately, the .cmp file is very incomplete16:29
wpwrak_so you have two choices: 1) add associations for all undefined footprints in cvpcb, or16:29
wpwrak_2) delete xue-rnc.cmp and run cvpcb again. it will then associate all the footprints it finds in the schematics. a few will still be undefined, though16:30
wpwrak_e.g., C7, C8, C9, F1, J1, RP1, RP2, etc.16:31
wpwrak_(they probably don't have a footprint in the schematics)16:31
wpwrak_yep. that's their problem16:31
wpwrak_anyway, gotta run. need more explosives ;-)16:32
andres-calderon:)  I'll try option 216:33
wpwrak_good choice ;-)16:35
wolfsprauldoes this mean we should not commit the .cmp file into git and recreate it every time locally?17:00
wpwrak_it may contain original data too. depends on your workflow.17:29
wolfspraulyes I know this file was always categorized as 'unclear' in terms of original or not...17:30
wolfspraulso maybe if we declare a certain workflow to be the preferred one (?) we can be more sure? guess we find out later17:32
wpwrak_that would help. but then you'd have to regenerate .cmp locally, or get a complaint from pcbnew on each invocation.17:34
wpwrak_it's basically that, whatever you do, it's not quite right17:35
qi-bot[commit] Andres Calderon: xue-rnc.cmp file has been restored http://qi-hw.com/p/xue/695a7ab17:42
kristianpaulzrafa: hello (late)19:45
kristianpaulzrafa: lars or larsc ?19:47
kristianpaulxiangfu: morning :-)19:47
xiangfukristianpaul: hi19:48
qi-bot[commit] Andres Calderon: new PSU draft design http://qi-hw.com/p/xue/851da6321:50
xiangfuhttp://www.debwrt.net/trac/wiki :)  Debian + OpenWrt = debwrt21:57
kristianpaulxiangfu: opkg is .deb compatible i remenber21:59
kristianpaulxiangfu: http://www.debwrt.net/trac/wiki/MinimumRequirements <-- too high (e device should have at least 350M(or more) of storage available for the Debian root filesystem)22:01
kristianpaulI wonder who really want debian in its router..22:02
kristianpaulbut looks intersting anyway :-)22:02
xiangfukristianpaul: now the last openwrt image  goto 400M :)22:03
kristianpaulwow, that was quick (goto 400MB)22:04
kristianpaulbecause of emacs? :-)22:05
kristianpaulmay be content?22:05
kristianpaulno please no more discuss about changing NAND schema again ;-)22:06
xiangfua lot of package. gcc-mips, emacs ...22:08
xiangfuwe need remove some useless program. I found the "sshfs" in openwrt never work.22:08
kristianpaulwe need polls :-)22:08
kristianpaulxiangfu: do you really think is convinient have all those command line apps in gmenu2, why not just a shell icon, i think people will want this commands to be run will enjoy get a shell first22:10
xiangfukristianpaul: hmm.. I am thinking list the terminal apps in gmenu2x which make people know there is such app in nanonote.22:11
xiangfukristianpaul: I don't have a good idea how to make people know there are a lot of apps in nanonote.22:12
kristianpaulhmm thats true too22:12
kristianpaulI'm just thinking in end-user people, just want this device work out the box22:13
kristianpaulhow they wil face command line apps, at first..22:13
kristianpaulmay be a manual could help22:13
kristianpaulanyway just a *tought*22:14
kristianpauli'm off now, bye22:14
zrafa10Mhz, 512Kb. Unix..22:54
--- Fri Dec 24 201000:00

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