#qi-hardware IRC log for Thursday, 2013-07-18

rohhttp://scasagrande.blogspot.ca/2012/04/gpibusb-for-sale.html this one... i just never used pics the last 10 years...00:00
rohthus no flasher around.. of course i didnt buy kits from him.. someone ordered the parts and the pcbs00:01
Luke-Jrhttp://www.diygadget.com/tiao-usb-multi-protocol-adapter-jtag-spi-i2c-serial.html looks nice..?00:21
rohshould be ok00:37
rohseems they added a level shifter for jtag and one for rs232 levels00:37
rohsimilar to the moko board, but less specific. quite nice design. and cheap. still not fully open but who cares for such details which are obvious ;)00:41
wpwrakLuke-Jr: maybe you need to get a better soldering iron. oh, and when working with smt and such, always apply plenty of flux. that's the magic ingredient that makes things work properly.02:13
wpwrakLuke-Jr: but i can assure you that modern electronic can be soldered by hand :) for example, i soldered all these critters: http://downloads.qi-hardware.com/people/werner/wpan/tmp/boards-so-far.jpg02:17
wpwrakso wouldn't go much smaller than that, though. the contact spacing of the chips is 0.5 mm, the resistors and such have 0402 packages.02:18
Luke-Jrwpwrak: maybe. the one I use is from the 1970s02:43
wpwrakyou may want to get something a little more modern :) doesn't have to be too fancy. but temperature-controlled is nice. also allows you to adjust the heat. e.g., ~270 C for regular work, more if you have large metal pieces involved03:02
Luke-Jrwpwrak: but I wouldn't know what to adjust it to :/03:04
rozzin1Have any of you ever tried using a NanoNote w/ the µSD Wi-Fi adapter as an acess-point?04:46
sanderr_rozzin1: No, but I think I should. :)08:48
kyakguys, what's the difference between 'man 2 recv' and 'man 3 recv'?09:07
kyaki understand that 2 is Linux programmer's manual, and 3 is POSIX's09:07
kyakbut why would i refer to POSIX programmer's manual anyway?09:07
kyakit clearly states that "implementation of this interface may differ from Linux or may not be implemented in Linux."09:08
kyakso what's the point of 'man 3 recv'?09:08
whitequarkbecause you may be trying to write portable code09:09
kyakbut how it is portable, if implementation may differ from Linux or not implemented at all?09:43
whitequarkwell, portable code adheres to POSIX09:44
whitequarkso if you only follow the POSIX manual and do not use anything that is not included there, your code will be portable09:44
whitequarkin theory at least :)09:44
kyakit doesn't say that POSIX is a subset of Linux09:45
kyakin fact, it says the opposite09:45
whitequarkConforming To09:46
whitequark4.4BSD (these function calls first appeared in 4.2BSD), POSIX.1-2001.09:46
whitequarkso Linux is a superset of POSIX09:46
whitequark(for this call that is)09:46
whitequarkoh, I see what you mean09:47
whitequarkthe POSIX manual prologue is the same for all POSIX manual pages09:47
kyakah, ok.. 09:48
whitequarkand Linux indeed is not completely compatible with POSIX09:48
kyakshould one follow POSIX manuals, and cross-reference with Linux manuals from time to time, or vice versa? :)09:49
whitequarkit depends on what you want to achieve :)09:50
whitequarkthere is quite a bit of Linux-only software which doesn't even try to be portable09:50
whitequark*cough* systemd09:50
kyakok, i see your point09:50
whitequarkadditionally there are POSIX extensions which are implemented in one way or another by most *nix systems09:50
viricposix doesn't only define portability in terms of kernel syscalls09:51
viricthe same way, not all software running in a GNU/Linux has to keep posix compatibility to the kernel09:51
kyakyeah, definitely not.. That's why Linux programmer's manual is there09:53
kyakso one could create posix-incompatible software :)09:53
whitequarkviric: I'm not sure if POSIX even refers to syscalls09:54
whitequarkit's defined in terms of behavior of library routines09:54
whitequarkand `read' isn't exactly close to `sysread'09:54
viricin fact glibc has code to emulate kernel-side things, in case you build a glibc for older kernels not having that functionality.10:00
rozzin1kyak, whitequark: man sections 2/3 are not `Linux vs. POSIXl', it's `kernel vs. library'.13:09
whitequark>This manual page is part of the POSIX Programmer's Manual.13:09
whitequarkThe prototypes given above follow glibc2.13:10
whitequarkpretty sure it's both library13:10
rozzin1There's plenty in section 3 that's either Linux-specific or GNU-specific.13:10
whitequarkoh I see13:10
larscwhitequark: man man13:10
whitequarkit's confusing13:10
rozzin1If you specifically want the *POSIX* manual for library calls, you need to look in section 3p.13:11
rozzin1Some of the pages in 3 are `inherited' from 3p where there's no reason to have a GNU- or Linux-specific page.13:11
whitequarkI see13:12
whitequarkthanks for explanation13:12
whitequarklarsc: that also doesn't mention 3p :p13:12
rozzin1The 2-vs-3 `kernel vs. library' distinction is much clearer in some man pages than others, e.g.:13:13
wpwrakthe 2/3 distinction isn't all that useful, except maybe if you're developing a libc. and then i wouldn't be so sure whether 2/3 follows the actual implementation or just history13:15
wpwrakfor compatibility, you really need to look at what the man page says. there are also many extensions that go beyond posix. and of course, no all man page tell the whole truth :)13:16
rozzin1Hrm. I'm actually having a hard time finding a good 2/3 example.13:16
whitequarkrozzin1: :D13:16
rozzin1I had one just the other day, though--I swear :p13:17
wpwrakso if you really want to be sure, you start from the official posix documentation13:17
rozzin1It's easier to find 3/3p examples, like "man 3 basename" vs. "man 3p basename".13:17
whitequarkyou know how I feel about this?  (o°¡°o5 ;;13:17
rozzin1I work with a bunch of people who build on Mac OS.13:18
wpwrakthat looked a little odd. where those three faces ?13:18
larscI don't even have 3p13:18
larscwpwrak: one face and a table13:18
larscdon't you do unicode?13:18
whitequarkwpwrak: "table flip"13:18
whitequarkthe o5 is his left hand and a "swoosh", ;; is an upside-down table13:19
whitequarkby itself, flipping table is an iconic symbol in Japanese media, much like facepalm is for Western one13:20
whitequark(why do I even know all that stuff)13:20
wpwraki see the table. but what's supposed to be the face doesn't look very face-ish. maybe i don't have the right cyrillic fonts installed :)13:21
wpwrakor japanese fonts :)13:21
rozzin1sanderr_: If you can try it, I'd be really interested it knowing if NanoNote works as an access-point. I'm looking for an `extensible Wi-Fi AP' platform, right now.13:22
wpwrakso the japanese flip tables ? interesting. and i thought certainly M$ guys throwing chairs were bad.13:22
whitequarkno chairs :D13:22
wpwrakhmm, that looks actually what i got. i guess i need face recognition 2.0 :)13:24
whitequarko are his raised hands, ° are eyes, and ¡ is the mouth13:24
whitequarkit's really a completely different style of writing emoticons13:24
larscthere is also the reverse one: ,13:26
wpwrakcheek-hand-eye-mouth-cheek-hand-arm. a bit surrealist13:26
wpwrakah, the Î( º _ ºÎ) could be hair-cheek-eye-mouth-eye-hair-cheek. that works.13:27
whitequarkwpwrak: now that you mention it...13:27
whitequark(and yes it's not hairs, it's arms)13:27
wpwrakstrange anatomy :)13:28
larscwell it's contrast to the minimalistic ':)'13:28
whitequarkwpwrak: i think it's an attempt to convey perspective13:28
whitequarkas in, he stands 45° wrt/ you13:29
whitequarkthink isometric13:29
larscthe '5' are movement stripes btw13:29
larscto show that the table is moving fast13:29
wpwrakah. i begin to see it. still kinda strange that the arms would be attached to the head, but yes, there is a "piggy throwing table" pattern tehre13:31
whitequarktry this fine illustration: http://i2.kym-cdn.com/photos/images/original/000/167/919/38f2f82655cf6ebd5842972e1318b03e.jpg13:31
wpwrakthe moral: don't cheat on your ninja-demon girlfriend13:35
viric"Would A Kickstarter Open-Source GPU Work?" Phoronix13:52
whitequarkBetteridge's law of headlines says that the answer is13:54
whitequarkalso: https://twitter.com/YourTitleSucks13:55
wpwrak"Would a Kick-Open Sourcestarter GPU work " ?13:56
larscsome company way or may not opensource their verilog gpu13:59
larscif they get enough money14:00
wpwraksome open gpu had been tried several years ago. not sure what happened with it.14:07
viricyes, I remember some.14:09
virichttp://www.phoronix.com/scan.php?page=news_item&px=MTQxMjU that's the post14:09
biotthere's also this one, a thesis project:14:27
larscwith about 1 commit14:29
biotnah, they worked on it for quite some time, I followed it on G+14:29
biotit was only after they turned it in that it was submitted to github, opencores etc14:29
larscit still looks rather dead14:32
biotyeah, but then so is that code base the company is pitching to phoronix, no doubt14:32
larscyep, probably just want to squeeze out a last few bucks14:33
biototherwise they wouldn't be pitching it -- they're just seeing if there's any money at all to be made from it still14:33
biotand quite possibly ORGFXSoC is less advanced than that one14:33
biotbut it does bear asking, why does anyone need to pay via kickstarter to get a basic GPU code base?14:34
viricthey will have to define well what they are going to deliver.14:40
rohbiot: well.. for starters to pay for the year or so of development which needs to happen first14:40
viricor they won't get much money I guess14:40
rohand even then a fpga based card will never be worth anything.. atleast outside of development.14:41
rohmuch too slow and power hungry while being expensive. bad combination14:41
viricmaybe they don't plan to deliver fpgas14:41
rohviric: heh.. i dont believe that yet.14:42
rohthe only ones i know which moved trhough fpga to real asics fast were the bitcoin miners. anybody else... nope.14:42
wpwrakso .. make a gpu that's also good at mining bitcoins :)14:45
rohi dont make gpus at all. shitty business. annoying customers. no value on quality or stability... booooring.14:46
rohjust gamerkids, nobody serious.14:46
viricis anyone delivering asic miners?14:50
viricI thought there have not been delivers yet14:50
rohthey do. but only if you ordered and payed last lear14:50
roheh year.14:50
rohbasically all they make is sold in advance14:50
viricbut did they *deliver*?14:51
rohlimited amounts, but yes.14:51
biotyou would think if they managed to get an ASIC manufactured, it would be in large amounts14:52
rohbiot: yes and no. for the first few batches you do small amounts to limit the impact of failiure.14:53
rohmaking chips is nothing risk free14:53
rohif you are sure your design is ok, you go into volume14:53
biotah right. and are they planning to make more?14:53
rohsure. but the fabs have lead times of month to years14:54
rohyou are confused by marketing cycles.14:54
rohreal hw development takes years14:54
kyakthanks guys for 2 vs 3 vs 3p explanation!14:56
sanderr_rozzin1: Might take some time until I'm able to try it, I'm on the road at the moment.19:07
sanderrrozzin1: But I surely will if I'm at home again.19:07
qi-bot[commit] Paul Cercueil: Remove all links of an OPK when moved out of the monitored folder (packages) http://qi-hw.com/p/gmenu2x/dfad7c321:17
qi-bot[commit] Paul Cercueil: Fix the function that sets the icon of a Link (packages) http://qi-hw.com/p/gmenu2x/504641a21:17
qi-bot[commit] Paul Cercueil: Change the "GMenu2X starting" message (packages) http://qi-hw.com/p/gmenu2x/32d100e21:17
qi-bot[commit] Paul Cercueil: Fix translations in user directory not correctly loaded (packages) http://qi-hw.com/p/gmenu2x/9029dec21:17
qi-bot[commit] Paul Cercueil: Order the links in alphabetical order (packages) http://qi-hw.com/p/gmenu2x/bcb3e9821:40
qi-bot[commit] Paul Cercueil: Allow an OPK to use an icon provided by the theme (packages) http://qi-hw.com/p/gmenu2x/91f381f21:53
--- Fri Jul 19 201300:00

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