#qi-hardware IRC log for Monday, 2011-10-10

qi-bot[commit] Werner Almesberger: m1rc3/norruption: moved material of 1st series tests to 1/, 2nd to 2/ (master) http://qi-hw.com/p/wernermisc/5037afb01:50
qi-bot[commit] Werner Almesberger: m1rc3/norruption/1/: distribution data and plot script (master) http://qi-hw.com/p/wernermisc/8b4ef5001:50
qi-bot[commit] Werner Almesberger: m1/perf/sched.c: synchronize with upstream version (master) http://qi-hw.com/p/wernermisc/9e32d5601:50
qi-bot[commit] Werner Almesberger: labsw/Makefile (bom): generate .lst file from eeschema (master) http://qi-hw.com/p/wernermisc/834157401:50
qi-bot[commit] Werner Almesberger: m1rc3/norruption/1/loop8: forgot to commit this one (master) http://qi-hw.com/p/wernermisc/f3bfd1f01:50
qi-bot[commit] Werner Almesberger: m1rc3/norruption/1/peekcmp: compate standby dump obtained with "peek" (master) http://qi-hw.com/p/wernermisc/3599e3e01:50
kristianpaulwpwrak: http://elespectador.com/impreso/vivir/articulo-304360-el-genio-de-jobs-era-el-opuesto-al-de-da-vinci02:00
qi-bot[commit] Werner Almesberger: m1rc3/norruption/2/: new loop and tools, with auto-reporting and auto-recovery (master) http://qi-hw.com/p/wernermisc/d433f7302:02
qi-bot[commit] Werner Almesberger: m1rc3/norruption/2/aloop: corrections and put cycle number at beginning of file name (master) http://qi-hw.com/p/wernermisc/f1ebf7403:59
qi-bot[commit] Werner Almesberger: m1rc3/norruption/2/upset: deliberately corrupt the NOR (master) http://qi-hw.com/p/wernermisc/7cb967903:59
kyakviric: the newsbeuter thing is getting really slow when it is 15000+ articles in a cache. It also uses around 250 Mb of RAM at this point -\ Have you tested offrss in such harsh conditions?06:44
qi-botThe build was successfull, see images here: http://fidelio.qi-hardware.com/~xiangfu/compile-log/openwrt-xburst.full_system-10092011-0220/06:51
Action: xiangfu needs prepare the next nanonote release. linux 3.1 and merge all atBen stuff to next release.06:53
xiangfu1.5 month passed.06:53
wolfspraulyes, great! :-)06:54
kyakindeed!07:13
virickyak: hmm I've maybe 4000 articles, with no slowdown, and almost constant ram use07:17
virickyak: it's up to the filesystem cache07:17
kyakthis is great07:37
wolfspraulkristoffer: you've asked Andrew Holme about the license of this GPS work already, right?09:55
wolfspraulthen I don't need to send more mails...09:55
C-Keenthe nanonote is an excellent zcode player for long train travels! Thanks guys :)10:06
viricC-Keen: what is that?10:12
C-Keenviric: playing text adventure games10:13
viricah ok10:14
viricI thought text adventure games were more for devices with easy typing10:19
wolfspraulC-Keen: zcode - is that in the image, a package, or you built it yourself?10:30
wolfspraulwe ported frotz to the NanoNote once upon request by Nick Montfort, who then also made sure his Curveship software runs on the Ben10:30
C-Keenwolfspraul: it is a package I found on the wiki, the name of the zcode interpreter is zoom10:31
C-Keenwolfspraul: interesting, I will have a look at nick's port10:31
wolfspraulyes, definitely10:31
wolfsprauldo you happen to have a url to the package? if you use it we should include it by default :-)10:32
C-Keenwolfspraul: ah sorry, I was wrong. I have been using frotz actually. So all is good!10:35
C-Keenjust checked10:36
wolfspraulyou played directly with frotz?10:47
wolfspraulor you had some game that used frotz?10:47
C-Keenfrotz is an interpreter that reads zcode images of games10:48
C-KeenI have played with it directly10:48
C-Keenthe typing is still a bit awkward for me on the keyboard10:48
wolfspraulC-Keen: are there any resources/sites for zcode games?11:10
wolfspraulfree stuff that we could just include11:10
C-Keentons of it, there are even new games created every year. the "official" archive is at ifarchive.org11:17
wolfspra1lC-Keen: excellent link, thanks!12:07
wolfspra1lis all that stuff freely licensed?12:07
wolfspra1ldo you think we can or should bundle some with the Ben?12:07
wolfspra1lah well, they make their life easy :-) http://www.ifarchive.org/misc/license.html12:08
viricI'm very bad at interactive fiction12:08
wolfspra1lwhat does that mean in practice?12:08
wolfspra1lmost of the stuff is not free?12:08
wolfspra1lor most is?12:08
wolfspra1lC-Keen: can you point me to your top-10 or so (or 5 or 3) zcode games on ifarchive.org?12:10
wpwrakviric: try this:  apt-get install tads3; wget http://mirror.ifarchive.org/if-archive/games/tads/Rematch.gam; t32run Rematch.gam12:12
viricwpwrak: debian not spoken12:13
wpwrakviric: yum install ? ;)12:13
viricnix :)12:13
wpwraksend your kid to school ! :)12:13
virichehe12:14
viricI've played some 'instead' games12:15
virichttp://code.google.com/p/instead/12:15
wpwrak"rematch" is quite unique. very out of the box.12:16
wpwrakyou'll notice that by the time you die (in the game) ;-)12:17
C-Keenwolfspra1l: sure! http://ifarchive.org/if-archive/games/zcode/LostPig.zblorb http://ifarchive.org/if-archive/games/zcode/anchor.z8 http://ifarchive.org/if-archive/games/zcode/photopia.z5 http://ifarchive.org/if-archive/games/zcode/photopia.z5 [more]12:23
wolfspra1lgreat!12:24
wolfspra1lthank you!12:24
viricwpwrak: ah ok I may try :)12:24
wpwrakviric: actually, the qtads engine may be better than tads3, because it has a command line history. you'll need that :)12:25
wpwrakviric: here's a bit more background. and a link to the solution: http://www.wurb.com/if/game/111412:26
C-Keenhttp://ifarchive.org/if-archive/games/zcode/vgame.z8 http://ifarchive.org/if-archive/games/zcode/awaken.z5 http://ifarchive.org/if-archive/games/zcode/adverbum.z5 http://ifarchive.org/if-archive/games/zcode/Savoir-Faire.zblorb and lots of others12:26
viricah, of course I *never* looked at a solution12:26
C-Keenwolfspra1l: ^12:27
wolfspra1lthat's a great start12:27
C-Keenwolfspra1l: see also http://www.intfiction.org/forum/viewtopic.php?f=23&t=3272&hilit=top+10 for ideas, but this list also contains non zcode games or nonfree games12:27
wolfspra1lthose are all freely licensed?12:27
C-Keenwolfspra1l: yes the ones I have listed are freely licensed AFAIK12:28
C-Keenall of these have been written from the late nineties and have been either submitted to a competition or are declared free by the authors12:28
C-Keenwolfspra1l: although the following ifarchive disclaimer applies: The archive makes no blanket claim about the copyright status of the12:29
C-Keenarchive files. Unless a file, its included license, or its description12:29
C-Keenexplicitly states that it is in the public domain, then the file's12:29
wolfspra1lsure sure12:29
C-Keencopyright rests with its author and the file is only available on the12:29
wolfspra1lI saw that12:29
C-Keenarchive for personal use.12:29
zrafahi14:21
zrafawpwrak: you there? .. some pm says about mmc cmd  :  Each command token is preceded by a start bit (0)14:24
zrafastart bit is 0.14:24
zrafaand succeeded by an end bi14:24
zrafat 1.14:24
zrafa(sorry the bad paste)14:25
zrafawell, if the clock is enought slow like you suggested14:25
zrafaI should check the CMD for any bit 1 right? (in order to know when to stop the controller)14:26
zrafawpwrak: and no just waiting for the final bit 1. I mean, we just want to have cmd = 1, so I guess that 14:27
zrafawpwrak: any bit = 1 in CMD bit is okey right? (I ask because I understood firstly to wait for the final bit 1, but I think now that you meant any cmd bit = 1)14:28
zrafasome pm says = 4740 pm says :)14:32
kristianpaulhave a trivia http://puu.sh/6MX315:07
wpwrakzrafa: back :)15:07
wpwraklemme find a good SD/MMC reference ...15:08
wpwrakhere we go: https://www.sdcard.org/downloads/pls/simplified_specs/Part_1_Physical_Layer_Simplified_Specification_Ver_3.01_Final_100518.pdf15:14
wpwrakthe SD command structure is on page 5815:14
wpwrakyou can set the command index and the argument to anything you want. so you could just make them all 0 or all 1 and then try to catch the right clock level. either by polling the i/o or by predicting the time from the start of the command to the middle of the corresponding clock phase.15:16
wpwrakwell, you probably don't even need to worry about command and value. there are both fixed 1 and 0 bits in the structure, so you can just use one of these15:18
zrafawpwrak: hey15:18
wpwrakMMC has a bit more low-level details. lemme see if there's a good description of the clock15:20
zrafawpwrak: first bit is 0 and last bit is 1 in a cmd command, but if I catch any bit 1 in CMD in the middle of a transmision that would be oeky for our plans right?15:20
zrafano more low level details please ;-)15:23
wpwrakour good friends in russia have a registration-free copy: http://yourcmc.ru/wiki/images/5/55/EMMC_JESD84-A441.pdf15:24
wpwrakpage 19415:24
wpwrakthe original is here: http://www.jedec.org/sites/default/files/docs/JESD84-A441.pdf15:25
wpwrakyou shouldn't need anything but the first two bits. see page 8 of the SD physical layer spec. start bit is 0, transmitter (direction) bit is 115:28
kristianpaulwolfspraul: i wrote him yes, dunno if kristoffer did it too 15:29
wpwrakso if you set the clock to frequency f, you'll have CMD = 0, CLK = 0 around t = X (but you don't know for how long), CMD = 0, CLK = 1 around t = X+1/4f, CMD = 0, CLK = 0 around t = X+3/4f, CMD = 1, CLK = 1 around t = X+5/4f, and CMD = 1, CLK = 0 around t = X+7/4f15:32
kristianpaulwolfspraul: his words "I had not thought about licensing."15:32
wpwrak"i was too busy killing AGPS" ;-)15:33
kristianpaul;)15:33
kristianpaulbut i asked again, about what he think about other people re-using its work, and i introduce him this irc and qi-hw15:33
kristianpaulno reply yet, that mail was from yday15:36
wpwrakzrafa: e.g., if you set the MMC bus clock to 250 kHz, you should see the following pattern: (CMD, CLK) = (0, 1), (0, 0), (1, 1), (1, 0)15:37
wpwrakzrafa: each of these four values should be stable for about 2 us15:38
wpwrakzrafa: now, if you want to be sure you hit the right combination, you would wait 1 us until after the value has been set. that way, you can err by about (-1 us, +1 us) and still get the desired setting15:39
wpwrakso this would be X+0*2+1 us, X+1*2+1 us, X+2*2+1 us, X+3*2+1 us15:40
wpwrakX is the time between telling the MMC controller to send the command and the beginning of the transmission15:40
zrafakristianpaul: (now I need ten days more to understand that werner just wrote  ;-) ) 15:42
wpwrakMSC_CMDAT.INIT controls whether 80 clocks are added before the command or not. you don't want that ;-)15:42
wpwrakbah, it's easy :)15:42
zrafawpwrak: saved. I just read the page 8 and I see those first two bits. Then15:43
kristianpaulzrafa: ;)15:44
DocScrutinizerhaha, low level fun? :-)15:44
wpwrakzrafa: have you been able to requisition a scope ? seeing what actually happens will help a lot :)15:45
zrafawpwrak: sorry, I do not understand yet why CLK changes, because I have not read well about clk. Let me a while (I di not found that at 4740 pm)15:45
wpwrakthe bus clock CLK changes to indicate then the data on the bus is valid. see page 194 of http://yourcmc.ru/wiki/images/5/55/EMMC_JESD84-A441.pdf15:49
wpwraki.e., data inputs and outputs must be stable when CLK raises from 0 to 115:51
wpwrakah, about the values ... what i wrote is probably wrong. more likely it's (0, 1), (1, 0), (1, 1) (x, 0)15:52
wpwrak(i.e., the data is stable on the rising edge of CLK but changes on the falling edge)15:52
wpwrak(x would be the first bit of the command)15:54
wpwrakso you would have (1, 0) at t = X+1 us and (0, 1) at t = X+3 us. couldn't be any easier :)15:54
zrafawpwrak: no scope yet.. Maybe this week15:57
kristofferwhat16:05
kristofferpeople keep paging me and then run off16:05
kristofferkristianpaul, whats the licensing you are talking about? Feel like Ive missed something16:06
kristianpaulkristoffer: about this project http://www.holmea.demon.co.uk/GPS/Main.htm16:07
kristofferkristianpaul, sweet :) It appeals to my nerdy side. Looks like a cheap design also16:10
wpwrakseems that there are at least two FPGA boards he didn't show :)16:10
wpwrak(but then, with, say, M1, we have a big fat FPGA that should be more than sufficient for the rest of the processing)16:11
kristofferIt would be worth alot to have that as an open design so can see what you are all excited about.16:11
wpwrakit would be even better to get him interested in joining kristianpaul's little project. he visibly knows his signal processing incredibly well.16:14
wpwrakif he "just" opens the design, that may still be too difficult for mere mortals to understand ...16:15
kristianpaulwpwrak: yes i already asked that too :)16:15
wpwraki find it particularly interesting that he just eliminated AGPS with a little FFT. some companies are building whole subcultures around the "A" in AGPS ...16:15
kristianpauland talk about milkymist, etc, same history :)16:15
kristianpaulwpwrak: how you get that relation between fft and agps?16:16
wpwraklook in section "Search": "A "warm start" using almanac data to predict positions and velocities still requires a code search."16:20
wpwrakand then "This GPS receiver uses an FFT-based algorithm that tests all code phases in parallel. From cold, it takes 2.5 seconds on a 1.7 GHz Pentium to measure signal strength, Doppler shift and code phase of every visible satellite."16:20
wpwrakso, as i understand it, no almanac needed16:21
kristianpaulah, i see you point, so you think GPS receiver are lasy now becaue relay on AGPS for fasttest fixes when they could just do more things in parallel?16:25
wpwraki'm not quite sure what the story is. maybe he discovered something new. maybe i'm just not up to date. maybe his approach has drawbacks. maybe i completely misunderstand.16:31
wpwrakbut i'm curious what the truth is :)16:31
kristianpaulfor me is not new :-)16:42
kristianpauli mean i had read like 3 books all pointed almost same technquies for signal tracking16:44
wpwrakah, you knew about this ? okay, maybe i just missed that then. i just remember that AGPS was all the rage some five years ago16:44
wpwrakheh, okay :)16:44
kristianpaulactually fft is ideal for software based processing16:49
kristianpauli dont expect tosee that on hw, i think PLL is better aprouch in that side16:49
kristianpaulor something like namuru like a correlator and a baseband processor16:49
kristianpaul(baseband processor) did not exits at least in open/free way yet :)16:50
kristianpaulwpwrak: but yes i already asked him to about why did not add processing stuff on the fpga17:39
virichm memstat needs rework on 3.x17:47
virickyak: offrss may take ~1MB of RAM for the amount of posts you said17:50
virichm maybe I'm exagerating even.17:59
viric500KB? :)17:59
kyakviric: i already shut down newsbeuter for good :) it is the first step towards offrss18:02
virichaha18:03
virickyak: all the ram offrss takes is to load a file with 15000 lines.18:11
viric(for 15000 articles)18:11
kyakviric: actually, my requirements are quite bigger.. I have around 3500 new items every day18:16
kyakviric: is there "mark all as read"?18:16
kyakcan't seem to find it..18:17
kyakheh, the pdf looks really cute :) it's adapted for the small screen18:18
kyakhm, having spaces in feed's name is not allowed?18:21
kyakah, it's a pity -\18:22
wpwrakin printable ASCII we trust :)18:25
kyakthe PDF has no pictures, so you are probably right :)18:32
virickyak: what would that mean? mark all as read? do you plan to lie? ;)18:53
virickyak: why would you want to get 3500 posts every day, and mark all as read, storing them forever?18:53
virickyak: maybe you should visit a specialist :)18:54
virickyak: I've not programmed anything to allow removing data, because I've not needed it. But it looks like you will need it18:54
wpwrakviric: i think with 3500 posts a day, the only way to survive is by just marking them read - without actually reading them ;-)18:58
viricwpwrak: looks like so :)18:58
viricmaybe it's me who makes a weird use of the reader, having hundreds of articles "still pending read"18:59
virickyak: put offrss under stress :)18:59
kyakviric: it's just that we have different uses for rss.. i use it primarily to track torrents. and i'm really good at going through the list and spotting what i need, and then marking the rest as read :)19:03
kyakespecially when i have a list of regexpes to higlight it for me :)19:04
wpwrak3500 torrents per day. man, i want your bandwidth !19:06
kyakthen, even on a news site rss you wouldn't click 'mark read' on each article, but rather read the articles that caught your eye and then mark the rest as read19:06
kyakwpwrak: it's the hard disk space that's important :)19:08
kristianpaul(no mark as read feature) nice trick :-)19:09
kyakbut i really dont't download much. I just dont' miss what i'm looking for..19:09
virickyak: I think that you need something made more specifically for the task19:19
virickyak: I thought of having some kind of mark, other than 'read'... but as long as I'll not need it, it's fine for me. :)19:20
zrafawpwrak: rodolfo found between all his old stuff almos forgotten a scope :) He will bring it to uni tomorrow19:45
wpwrakzrafa: whee ! did he tell you which brand/model it is ?20:02
zrafawpwrak: nope, and i guess that it could be an old one :)21:40
wpwrakperhaps the one, his family, when then came to argentina before the war, remember their grandparents considering a precious family heirloom. you should feel honored that he would let you touch it :)21:46
zrafawpwrak: ;-))22:21
kristianpaullol22:26
kristianpaul:)22:26
--- Tue Oct 11 201100:00

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