#qi-hardware IRC log for Saturday, 2015-01-24

DocScrutinizer05name=basename $0;00:21
Action: DocScrutinizer05 rather should use name=`basename $0`;00:22
wpwrakand don't forget the rm -rf /$DIRECOTRY/ :)02:09
DocScrutinizer05well, that's simple. But the missing `` make a nice forkbomb from a silly convenience thing for stderror output like "foo: wrong parameter" where foo is the basename of the script02:16
DocScrutinizer05>>name=`basename $0`;<< is fine, you can use $name later on in script like "$name: wrong parameter". However >>name=basename $0;<< does what ? ;-)02:17
DocScrutinizer05it's a rarely ever used ancient way to provide parameters for a executable ($0), as environment vars02:19
wpwrakrarely used ? i use it quite often :)02:20
DocScrutinizer05it probably predates the parameter appending to the command name02:20
Action: DocScrutinizer05 also uses it often02:20
wpwrakand the "fork" bomb isn't so bad. it should build up a lot of processes, hit an error, but then clean up after itself.02:20
wpwrakerr, "fork bomb"02:21
DocScrutinizer05related: 02:25
DocScrutinizer05: ${debug='false'}   ;#valid: true | false [| <any cmd>]. 02:25
DocScrutinizer05#  You may run `debug=true <this-file>`, or`debug='echo "!!!! "' <this-file>`02:25
DocScrutinizer05(lot of processes) err, yep. nevertheless it almost got my box to aa grinding halt before it reached that error02:27
DocScrutinizer05since the error reason been "process table full" or "out of memory (incl swap)"02:28
wpwrakyeah there's a reason why i wrote "it should", for i didn't care to verify it by experiment ;-)02:29
DocScrutinizer05wtf? where went that colon?  : ${debug:='false'}  of course02:33
DocScrutinizer05and I missed to quote the third line in that block:   ${debug} && set -vx && exec 2>&402:34
DocScrutinizer05I guess the last one only makes sense when knowing the previous line too:02:39
DocScrutinizer05# get rid of shell errormsgs - yeah i know...02:39
DocScrutinizer05exec 4>&2 2>/dev/null02:39
DocScrutinizer05gosh, I wrote some funny code 8 years ago ;-)  http://paste.opensuse.org/8918244902:48
larscon a modern and good operating system it should ;)09:24
DocScrutinizer05http://paste.opensuse.org/19035620  GRRR09:31
wpwrakxiangfu: (soldering) great ! it's best to start with the MCU and the surrounding capacitors, then try to see if SWD works10:38
wpwraka bit tricky if you don't have your UBB board yet, though :)10:41
xiangfuthe cable ordered from taobao.com was arrived. :)11:23
xiangfubut we are not receive the kl26...11:23
wpwrakah, without the kl26 that step will be a bit difficult, yes :)11:32
wpwraki guess it's best to wait until you have the kl26. this is the most difficult part to solder and you don't want the board full of other components at that time.11:33
wpwrakmaybe you want to post pictures of the PCBs ? that would be the first industrially-made ones, so far we only have my ugly kitchen jobs :)11:35
xiangfuI didn't get it yet. :)11:40
xiangfuit was at the hardware people place.11:41
xiangfuthe usb connect is interesting.11:43
xiangfuwpwrak: I am thinking how to make some of those cases. 11:44
arhuacoI am had not used LiPo batteries before. I discharged them too much. A working batt like this one should charge above 8v and last more than 5 minutes with the load I was using (a motor). http://arhuaco.org/tmp/batt.png11:51
arhuacoSo, that's how 2 broken batts look like :-/11:52
xiangfuwpwrak: http://downloads.qi-hardware.com/people/xiangfu/anelok/pcb/ <--- Just asked them post some pics to me. it taked by mobile phone.11:55
xiangfuwpwrak: looks like there are something wrong with my Ubuntu + Freecad (http://downloads.qi-hardware.com/people/xiangfu/anelok/tmp/freecad-anelok-case-0.png)12:17
wpwrak(pcb pics) thanks ! the fab was a bit sloppy with the bottom silk screen, but that's not a problem for manual soldering12:26
wpwrak(freecad) oh, that looks interesting ;-)12:27
xiangfuPCB: there is no silk screen on PCB. 12:28
wpwrakmaybe graphics driver. what video card do you use ?12:28
xiangfuyes. there are two video card in my laptop. the Macbook pro retina laptop.12:28
xiangfuI think the ubuntu use only one. 12:28
wpwrak(silk) the white stuff. at the bottom, it's visibly misaligned. and it even seems to overlap a little with pads.12:29
xiangfuwpwrak: any advice on fatfs coding work?12:29
Action: xiangfu reading this http://elm-chan.org/fsw/ff/00index_e.html12:30
wpwrak(macbook) hmm, dunno about these. are you using the open source drivers or the proprietary ones ?12:30
xiangfuI am using the open source driver.12:31
xiangfu(pcb) I will post some bigger picture tomorrow.12:31
wpwrak(fatfs) looks nice. what's important is that we need fat32 support, for those large usd cards12:32
wpwrakPerit FAT looks even nicer, but may be a bit too limited12:33
wpwrakah yes, too many limitations. pity.12:34
xiangfuwhich one?12:35
wpwrakPetit FAT. FatFs seems to have everything we need.12:36
xiangfuI even didn't notice there is 'Petit FatFs' at the first paragraph12:37
wpwrak(video) freecad on ubuntu 14.04 works fine here on nvidia and ati, both with the open source driver12:37
xiangfuit have a nice start doc: http://elm-chan.org/fsw/ff/en/appnote.html12:37
wpwraki built freecad from git, but that probably doesn't make a difference. i had used the ubuntu version before and it also worked fine12:38
wpwrak(start) nice ! and yes, 2 TB should be sufficient for now :)12:38
wpwrakseems to be nicely optimized12:40
wpwrakand the license is very friendly. excellent.12:41
xiangfuand looks it still active.12:41
wpwrakreleased as zip. very old-skool ;-)12:57
wpwrakfor making a case, you basically have three choices:13:20
wpwrak1) freecad -> cnc. the problem: this path doesn't officially exist yet and what i've seen of the development work in the are doesn't look as if it could be sufficient for this kind of case.13:21
wpwrak2) FreeCAD -> STL -> HeeksCAD/HeeksCAM -> CNC. this should work but it's a bit messy13:22
wpwrak3) FreeCAD -> STL -> slicer.py + cameo -> g-code13:22
wpwrak3) is my process. cameo outputs "gnuplot" output which i then convert to RML-1, the strange language my mill speaks13:23
wpwrakalmost everything else uses g-code, but i don't have a converter for that. it shouldn't be overly difficult to write one if you have a mill to test it on.13:23
wpwrakthis is my gnuplot to rml converter: http://projects.qi-hardware.com/index.php/p/cae-tools/source/tree/master/gp2rml/gp2rml.c13:24
wpwrakmy freecad-to-rml process seems to work pretty well now. there is one remaining problem: sometimes cameo fails to do the "area fill". this must be some problem with rounding errors since it depends on the position of the piece. so far, i've been able to always avoid the failure (an assertion violation) by moving the addition of an xy offset (i.e., the placement of the piece) either before or after the other operations in cam.sh13:28
xiangfuhmm... sounds not easy for me. :)13:30
Action: xiangfu write down to my memo. 13:30
wpwrakit's a bit complicated, yes. i screw up often enough, too :) 13:30
wpwrakbut CNC-machining is fun :)13:31
wpwrakarchang mentioned that you were thinking of getting a cnc mill. whitequark may have some suggestions for this13:34
wpwraki guess you may also want to they a more-or-less dedicated pc for the mill13:34
whitequarkyes with an LPT port13:34
whitequarkalternatively people have been using cubieboard to drive LPT port controllers13:35
wpwrakman they're cheap !13:35
whitequarkby this point they're quite hard to find actually13:35
whitequarkanyway, that machine will require some "improvement"13:35
whitequarkthankfully i documented it all :]13:35
wpwrak(hard to find) excessive demand, or did they stop making it ?13:36
whitequarkthey stopped making it a decade ago13:36
whitequarkand by now people have mostly thrown this hw out, here13:37
Action: xiangfu write down to my memo. 13:37
whitequarkISA multicards, if you remember what they are, go for four digits in dollars13:37
whitequark(to some poor souls who have SCADA on 1995 tech)13:38
whitequarkthis reminds me of some nuclear power plant company which doesn't plan to phase out pdp-11 until at least 204013:38
wpwrakprices here .. for a 3040, about USD 3000-4000 depending on how you count. okay, that one has a 4th axis, but still13:38
whitequarkwpwrak: holy shit, that's ridiculous13:39
whitequark3040 is at most $600, without delivery13:39
wpwrak(stopped a decade ago) wow !13:39
wpwrakwhitequark: you're in a free market, not behind some iron curtain :)13:39
whitequark... I'm really not13:40
whitequarkI suppose it's not as bad as what you have, not nearly13:40
wpwrakthis is what you get here for ~USD 400-500: http://articulo.mercadolibre.com.ar/MLA-537517817-router-cnc-_JM13:40
wpwrakgood strong MDF13:40
whitequarkMDF... ugh13:40
whitequarkand crappy spindle13:41
whitequarkmy mill is fairly bad but now I understand why people like to buy these13:41
wpwrakah, and that's without motors and electronics ;-)13:41
wpwrak(3020) someone should start to make them again :_13:42
whitequarknonono, you don't get it13:42
whitequark3020s are actively manufactured13:42
whitequarkthe motherboards with LPT ports are not13:43
wpwrakah, and there are also some used MDX-15. more expensive than mine was new. the miracle of a restricted market :)13:43
wpwrakaah, i see13:43
rjeffriesany thoughts on the project to creata what would appear to be a reasonably open laptop? (missing bit is BIOS I guess)16:55
whitequarkchromebook has FOSS BIOS16:55
whitequarkexcept graphics bits, to some degree16:56
rjeffriesthis is one I saw recently: https://www.crowdsupply.com/purism/librem-laptop16:57
rjeffriesit seems (to me) to be in the spirit of #qi-hardware16:58
whitequark"While the BIOS is not yet free, the Librem 15 will be the first laptop ever manufactured to ship a modern Intel CPU fused to run unsigned BIOS code"16:59
whitequarkthat's idiotic16:59
whitequarkyay BIOS-level rootkits16:59
whitequarkwhat you do is provide a way to change secure boot trust root by, say, flipping a physical switch17:01
whitequarkwhat you do NOT do is remove trust root entirely and claim this somehow "respects user privacy"17:02
rjeffriesso you're saying (I think) no  (modern) Intel CPU can be in a free and open hardware system.17:02
whitequarkchromebook gives you such a system17:02
whitequarkno need for some kickstarter fraud17:03
whitequarkand chromebook does it *right*17:04
whitequarkBoth things will be nessary, if you want to flash your own version of Coreboot onto your Chromebook (in-system). It is nessary to close a circuit on the motherboard via either a switch, jumper or srew to disable the write-protection of the spi chip.17:04
rjeffriesare you saying this purism project is fraudulent? Really?17:05
rjeffriesbtw I am writing this on an early Samsung C303 Chromebook. Not a bad little machine.17:06
whitequarkmost crowdsourced projects overpromise and under/never deliver17:06
whitequarkand an entire laptop is a very complex project17:06
whitequarkwhat's worse is their claims17:06
rjeffriesthis lappie looks doable to me. not impossible. it is expensive, yes17:07
wpwrakif all else fails, there's still novena ;-)17:14
arossdotmerjeffries: i'd say your best bet is http://rhombus-tech.net19:33
arossdotmewith librem your putting a lot of funds in to a one off laptop that will quickly become an antique or you could use less funds to invest into a new future of computer design19:35
arossdotmeor you could use the same funds you would have spent on a librem to help the project move faster :D19:35
arossdotmei guess within a few years if not already embedded stuff will have the same specs as the librem but without the fan :)19:36
arossdotmeit's a lot lot easier to chun out a new computer card instead of a hole new laptop and faster/newer/diff cpu/ram/gpu and better for the environment19:38
arossdotmelaptop desisners can focus of making a great laptop in stead of a good but faster laptop for the next product cycle 19:39
DocScrutinizer05wpwrak: what do you say about my bash USB DMM readout script? Contrary to what you encountered, it's 100% reliable too20:38
DocScrutinizer05and actually I take quite some pride in the timestamp/scheduling function20:40
DocScrutinizer05100% frequency stable, skew ~20ms avg, max jitter ~ -10/+100ms20:42
DocScrutinizer05the max sample rate of 1Hz is only by the integer arithmetic of bash $(( )), I could tune that to maybe 20Hz if I'd use milliseconds instead of seconds for interval20:44
DocScrutinizer05there are two cases that are not yet handled correctly: a tiny flaw and "unbearable" 5 lines of error codes when the /dev/ttyUSB* is not readable by user due to file permissions, and no handling of disconnecting or switching off the DMM while command takes a series of samples - mostly due to my lack of ideas how it actually should react on such event. Abort? Error to stderr? Error to stdout? No sample taken at all? Retry a N times, 20:53
DocScrutinizer05then throw error?20:53
nicksydneywpwrak: http://benjaminkerensa.com/2015/01/23/get-a-free-u2f-yubikey-to-test-on-firefox-nightly23:14
wpwrakDocScrutinizer05: yeah, i would have expected you to run into painful problems by now :)23:52
wpwrakat least things like the meter sometimes not responding or such, which then would then require some tricky timeout and recovery mechanism.23:53
DocScrutinizer05which is already implemented in that script23:53
wpwrakluckily, it doesn't even support fancier things where such problems are almost unavoidable, e.g., meter-triggered data transfer (with issues of how to cancel things, etc.)23:54
DocScrutinizer05ooh, it does: *0 .. *723:57
DocScrutinizer05when turning the turn dial23:57
DocScrutinizer05and yeah, this *might* spoil one response reading, but then any response would be bogus anyway when fetching data while actually the range/type of test changed23:58
DocScrutinizer05wouldn't know why and what to cancel though23:59
--- Sun Jan 25 201500:00

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