2011-06-03 00:01 whitequark: for atben, they're in tools/lib/atben.c 2011-06-03 00:01 whitequark: atusb doesn''t bit-bang but uses the SPI controller in the MCU. code in atusb/fw/spi.c 2011-06-03 00:03 whitequark: there's also unusual bitbanging in tools/atrf-xtal/atben.c (unusual in the sense that they're not used for the regular register interface) 2011-06-03 00:04 whitequark: of course, none of this has anything to do with the kernel :) for that, you need to look at the qi-kernel tree. there, i just use the regular spi-bitbang driver, slow as it is 2011-06-03 00:04 whitequark: so the answer depends a lot on what exactly you're looking for ;-) 2011-06-03 00:07 Does the libncurses-dev package provide the libraries for gcc-mips? 2011-06-03 00:10 hmm, what is "libcurses-dev package", what are "the libraries" and what is "gcc-mips" ? ;-) 2011-06-03 00:11 there are certainly libcurses-dev packages in existence that provide libraries compiled with gcc for some MIPS target 2011-06-03 00:11 wpwrak: I'll decide tomorrow probably. it's 4:10 already... 2011-06-03 00:12 whitequark: ;-)) 2011-06-03 00:19 Hmm, well I just recently got the nanonote, so I'm only familiar with the packages provided at qi-hardware.com So I refer to the libncurses-dev package provided there. As well as the gcc compiler provided there. The libraries would be the libraries required to compile a program using ncurses. 2011-06-03 00:21 there are two distributions there, openwrt and jlime. they're unfortunately not library-compatible. but as long as you pick things consistently from the same, what you wrote ought to be correct. 2011-06-03 00:22 (the distribution the ben comes with is openwrt) 2011-06-03 00:22 ah, and i don't know which version they pre-install. so if it's very old, you may have to upgrade first 2011-06-03 02:54 [commit] Xiangfu Liu: new package: gtkguitune: for tuning guitars and other instruments by using the method of Schmitt-triggering http://qi-hw.com/p/openwrt-packages/6c1f174 2011-06-03 05:26 xiangfu: note that link from qi-bot is not working -\ 2011-06-03 05:26 The build was successfull, see images here: 2011-06-03 05:26 http://fidelio.qi-hardware.com/~xiangfu/compile-log/openwrt-xburst.trunk-06012011-1849/ 2011-06-03 05:26 should be "openwrt-xburst.trunk-full_system-06012011-1849/" 2011-06-03 05:33 kyak: which link is not working? 2011-06-03 05:35 wolfspraul: the one qi-bot is announcing once the build has finished 2011-06-03 05:57 oh it does? 2011-06-03 05:59 yep.. 2011-06-03 06:09 i guessed it ;) 2011-06-03 06:11 Fusin: you might've noticed that qi-bot is referring to him as just "Bill" 2011-06-03 06:12 they are good friends 2011-06-03 06:14 first nokia, now qi-hw ... what's the world coming to ? 2011-06-03 06:15 i thought his name came from the fact, that those who use his software are passing a Gates of Bills, hence Bill Gates :P 2011-06-03 06:15 s/a/ 2011-06-03 06:17 shitty waiting-day 2011-06-03 06:17 waits technician from telco to reconnect his DSL-line... 2011-06-03 07:03 Fusin: always fun to have to wait for some to ring your doorbell :) 2011-06-03 08:04 interesting ... if i have two pointers  volatile uint32_t *foo, *bar;  and i use them in an assignment of the type  *foo = *bar = value;  then gcc seem to drop one of the stores 2011-06-03 08:15 hah, worse. it does the equivalent of 2011-06-03 08:15 *bar = value; 2011-06-03 08:15 *foo = *bar; 2011-06-03 08:16 it has a certain logic to it ... but seems wrongish nevertheless 2011-06-03 08:17 as Mr. Spock said: fascinating. 2011-06-03 08:18 ANSI C leaves this ambiguous. the value of the assignment should be "the value stored". that could be either. 2011-06-03 08:18 ah well, splitting the assignment then 2011-06-03 08:18 and Yoda says: Not the logic important is, results matters 2011-06-03 08:20 wpwrak: the compiler did exactly what you told it 2011-06-03 08:20 Fusin: atrf-gpio says "hzHhhl" 2011-06-03 08:22 larsc: depends on how you interpret what the "stored" value of an assignment expression ought to be. the value that was _written_ or the value that can be _read_ after writing. 2011-06-03 08:22 hrhr 2011-06-03 08:33 wpwrak: regarding the bitbanging, I was interested in the method used to access atben, so it probably rules out atusb stuff and in-kernel spi driver (you don't use that right?) 2011-06-03 08:34 wpwrak: particularly, I was looking for a 'good' way to remap mcu registers to userspace 2011-06-03 08:35 larsc: but yes, in the ANSI C standard, it reads more like the second interpretation. and also in the narrative of K&R 2nd ed. 2011-06-03 08:36 whitequark: for atben, i have all sorts of access methods :) 2011-06-03 08:36 whitequark: perhaps the bit of code i use for ben-blinkenlights can help you, though: http://projects.qi-hardware.com/index.php/p/ben-blinkenlights/source/tree/master/bbl/bbl.c 2011-06-03 08:37 wpwrak: the code in tools/lib/atben.c is simple enough 2011-06-03 08:37 just an mmap 2011-06-03 08:39 whitequark: (atben) my tools use bit-banging from user space, but usually encapsulate it in a library. in some cases, they bypass the library, though. for "real" use, you'd use the kernel driver, which is currently based on the bit-banging spi driver. (but i have to try and see if i can't make it a lot faster with an optimized driver. speed may matter here.) 2011-06-03 08:39 yeah, if you look past all the power cycling stuff, atben.c isn't too bad 2011-06-03 08:41 wpwrak: talking about power cycling, how did you determined the right intervals (the ones you provide as a reference, i.e. 2ms and not 10 for poweron)? 2011-06-03 08:41 just used a scope and measured them directly? 2011-06-03 08:43 i don't remember. maybe i just tried some values until it worked ;) but a scope may have been involved, too 2011-06-03 08:44 could you point me to atben schematics? 2011-06-03 08:45 one click from here: http://projects.qi-hardware.com/schhist/ 2011-06-03 08:46 thanks 2011-06-03 08:49 wpwrak: looks like it is not very complicated. at least not significantly more than cc2500-based circuit 2011-06-03 08:50 there is one problem for me through. atrfs are not imported to russia 2011-06-03 08:51 strange. the ti stuff has US export restrictions (or at least had it last time i checked), while the atmel stuff doesn't. i thought those restrictions were there mainly to keep you evil commies at a distance ? ;-) 2011-06-03 08:52 the problem is not with export restrictions at all 2011-06-03 08:52 there's just no people working with them here, and so no one buys them 2011-06-03 08:52 the same applies to any other non-very-common chip 2011-06-03 08:55 any chance that digikey.ru could be of use ? they probably know how to get things through customs efficiently (if possible at all) 2011-06-03 08:57 the difference between knowing how to do international shipments and only thinking one knows it can sometimes be rather significant. such as 3x the shipping cost and 5x the customs processing cost 2011-06-03 08:57 (from examples i have here) 2011-06-03 08:58 from the banner on front page: "THE COST OF DELIVERY TO RUSSIA IS $120" 2011-06-03 08:59 ah, pricy ... 2011-06-03 09:01 huh, chipfind.ru promises delivery of anything from Farnell, from one part 2011-06-03 09:02 the minimal cost is around $20, and there's no delivery fee 2011-06-03 09:02 neat 2011-06-03 09:02 they deliver it for ~18 days, through 2011-06-03 09:03 still much better than nothing :) 2011-06-03 09:03 i.e., 7 days at customs :) 2011-06-03 09:03 there's still same problem with RF parts 2011-06-03 09:04 *passive RF parts 2011-06-03 09:04 you mean the balun ? 2011-06-03 09:04 yeah 2011-06-03 09:05 hm, there's no problem. farnell has them too, for like $0.5 2011-06-03 09:05 there are two: the one from johanson and one from wuerth. the latter is more expensive. 2011-06-03 09:06 WURTH 748421245 2011-06-03 09:06 JOHANSON 2450FB15L0001 2011-06-03 09:07 digikey has Wurth ones for $4 2011-06-03 09:08 they should have both 2011-06-03 09:09 USD 2.58 @1 2011-06-03 09:09 the johanson is USD 1.27 @1 2011-06-03 09:10 I've looked up costs on that russian proxy. also, johanson is ~$0.5 somehow 2011-06-03 09:10 if you're determined, you can also make a discrete balun 2011-06-03 09:11 0.50 for MOQ 1 ? or 1000 ? :) 2011-06-03 09:30 for 1 2011-06-03 09:30 for 1000 it's even cheaper 2011-06-03 09:33 you have some funny pricing in russia :) 2011-06-03 09:34 I guess it has something to do with nuclear physics 2011-06-03 09:34 you know, the particles with negative mass 2011-06-03 09:36 somewhere customs will take away 5x the real cost, and that'll get compensated due to law of conservation of money 2011-06-03 09:39 that sounds like the most plausible explanation indeed :) 2011-06-03 09:40 so I guess that my workflow should be like this: 2011-06-03 09:42 order ben, atben & atusb [wait a month], then play with RF stuff a bit to understand how it works, then order atrfs and related parts [wait a month while I can learn to make two-side pcbs], and then try to assemble something working 2011-06-03 09:43 (and then accidentally fry everything and start over.) 2011-06-03 09:45 sounds like a plan, yes :) 2011-06-03 09:46 you could of course even order the parts before the boards arrive. pipelining :) 2011-06-03 09:47 sure 2011-06-03 10:10 [commit] David Kühling: liballegro-digmid: make percussion instruments work. http://qi-hw.com/p/openwrt-packages/7ce9a12 2011-06-03 10:10 [commit] David Kühling: Merge branch 'trunk' of projects.qi-hardware.com:openwrt-packages into trunk http://qi-hw.com/p/openwrt-packages/7dc3d44 2011-06-03 10:12 wpwrak, hi i remembered that there's a tool you did to have a graphic history(pdf) about KiCad. 2011-06-03 10:16 [commit] David Kühling: liballegro: make allegro-setup operate on correct config file http://qi-hw.com/p/openwrt-packages/58cdfd4 2011-06-03 10:18 wpwrak, is there info linked to how can i get that graphic history? tks. 2011-06-03 10:30 aw_: the easiest approach is to add your project to the list of things that get automatically processed 2011-06-03 10:30 aw_: i.e., the ones you can see here: http://projects.qi-hardware.com/schhist/ 2011-06-03 10:31 aw_: all this is being run by fidelio:/home/schhist/schhist_update_all 2011-06-03 10:33 aw_: if you only want to run it locally, then you'd use schhist2web from the eda-tools project, directory schhist/ 2011-06-03 10:34 aw_: first argument is the path to the target project's top-level directory, then the relative path from there to the top-level schematics, and finally the name of the output directory 2011-06-03 10:35 aw_: e.g., if your project is called "foo" and your schematics begin at foo/bar.sch, you could simply cd to foo, then run ../eda-tools/schhist/schhist2web . bar.sch whatever 2011-06-03 10:35 (where "whatever" is the name of the output directory) 2011-06-03 10:36 you can then fire up a browser on that foo/whatever/index.html 2011-06-03 10:43 wpwrak, okay, thanks a lot. seems that i need to run it locally. since http://projects.qi-hardware.com/index.php/p/m1/source/tree/master/jtag-serial already moved to https://github.com/milkymist/extras-m1/tree/master/jtag-serial 2011-06-03 10:45 wpwrak, do you think that PCBnew can still have possibility this graphic history on routings? 2011-06-03 10:45 ah, guthub makes things more complicated, indeed :-( 2011-06-03 10:45 sure. just need to find the time :) 2011-06-03 10:47 got it. as long time, the KiCad is good for projects, it would be great if PCBnew tool can be fit schhist tool though. :) 2011-06-03 10:49 yeah, agreed :) 2011-06-03 10:49 wpwrak, e.g. i sometimes needs to compare both (before/after) KiCad routings, so i did as this way: http://en.qi-hardware.com/wiki/JTAG_Serial_Cable_run2_for_Milkymist_One#Change_Notes 2011-06-03 10:50 wpwrak, but it's okay though on picture or graphic view. :-) well...thanks for hints on schlist. :) 2011-06-03 10:52 [commit] Maarten ter Huurne: Surface: Use the destructor as intended. http://qi-hw.com/p/gmenu2x/0a9229f 2011-06-03 10:52 [commit] Maarten ter Huurne: Surface: Load image with factory method instead of constructor. http://qi-hw.com/p/gmenu2x/52f4686 2011-06-03 10:52 [commit] Maarten ter Huurne: Surface: Minor cleanups. http://qi-hw.com/p/gmenu2x/25e8e62 2011-06-03 10:52 [commit] Maarten ter Huurne: Surface: Made "raw" field private. http://qi-hw.com/p/gmenu2x/7861e07 2011-06-03 10:52 [commit] Maarten ter Huurne: Surface: Make pixel format conversion explicit. http://qi-hw.com/p/gmenu2x/8a81837 2011-06-03 10:52 [commit] Maarten ter Huurne: PNG: Only add alpha channel on surface if loaded image has an alpha channel. http://qi-hw.com/p/gmenu2x/36260e9 2011-06-03 10:59 aw_: np. let's hope it works :) 2011-06-03 11:54 so, finaly I'm back with WiFi & DSL :) 2011-06-03 13:09 kyak: qi-bot told me you wa looking for me? 2011-06-03 13:11 jow_laptop: he, right :) What a smart thing! 2011-06-03 13:11 jow_laptop: may i ask you to have a look at bugs 9410 and and 9478? 2011-06-03 13:11 they have proposed patches, should be working fine... 2011-06-03 13:15 done 2011-06-03 13:17 thanks a lot :) 2011-06-03 13:17 np 2011-06-03 13:18 [commit] kyak: moc: pushed upstream http://qi-hw.com/p/openwrt-packages/2447082 2011-06-03 18:45 russian glibc translation is hilarious 2011-06-03 18:45 they've translated errno=0 (Success) as "Victory" 2011-06-03 18:46 like "Cannot exetute binary file: Victory" 2011-06-03 18:46 *execute 2011-06-03 18:46 well, the binary file may escaping execution quite a victory ... 2011-06-03 18:46 s/may/may consider/ 2011-06-03 18:49 heh. that's a nice cross-lingual joke 2011-06-03 19:09 whitequark: "button accordion" :) 2011-06-03 19:11 kyak: such a great potential for localization abuse 2011-06-03 21:10 The build was successfull, see images here: http://fidelio.qi-hardware.com/~xiangfu/compile-log/openwrt-xburst.full_system-06032011-0001/