#qi-hardware IRC log for Tuesday, 2012-05-22

qi-bot[commit] Werner Almesberger: b2/: move all error handling from subex to subst (master) http://qi-hw.com/p/eda-tools/3ad6c2c00:25
qi-bot[commit] Werner Almesberger: b2/: add relop to struct param, in preparation of merge with struct var (master) http://qi-hw.com/p/eda-tools/fdac54200:25
qi-bot[commit] Werner Almesberger: b2/: rename "name" field in struct value to more generic "s", for better sharing (master) http://qi-hw.com/p/eda-tools/061e8f500:25
qi-bot[commit] Werner Almesberger: b2/: use (general) "struct param" to replace (specialized) "struct var" (master) http://qi-hw.com/p/eda-tools/c1b6b2a00:25
qi-bot[commit] Werner Almesberger: b2/: move basic handling of variables from subex to param (master) http://qi-hw.com/p/eda-tools/4a3e51700:25
qi-bot[commit] Werner Almesberger: b2/db.c: add (somewhat crude) parametric part selection (master) http://qi-hw.com/p/eda-tools/bcf469600:25
qi-bot[commit] Werner Almesberger: b2/boom.c: add parametric part selection (with option -Q) (master) http://qi-hw.com/p/eda-tools/256a09600:25
qi-bot[commit] Werner Almesberger: b2/: correct some whitespace anomalies (master) http://qi-hw.com/p/eda-tools/2c5850e02:30
qi-bot[commit] Werner Almesberger: b2/eval.c (decode_mult): also support "G" prefix (master) http://qi-hw.com/p/eda-tools/2014e6002:30
qi-bot[commit] Werner Almesberger: b2/: add "ignore" keyword for substitution rules (master) http://qi-hw.com/p/eda-tools/d2171eb02:30
qi-bot[commit] Adam Wang: dip.fpd, dip4of6.fpd: copied both from /labsw/modules (master) http://qi-hw.com/p/kicad-libs/9c6b38d03:37
jyfl987xiangfu: can i access the on-chip bootrom from nanonote?05:05
xiangfujyfl987, no.05:07
xiangfujyfl987, hardcode.05:07
jyfl987xiangfu: which project do you now working for?05:08
xiangfuqi-hardware project. :-)05:08
xiangfuben nanonote and Milkymist One. 05:08
jyfl987xiangfu: is it possible to let nanonote access network via mobile? gprs/3g05:09
xiangfujyfl987, here is one faster solution: you can try to buy a OpenWrt support router that have usb host.05:10
xiangfuthen you connect nanonote and usb-3g dongle to that router.05:10
jyfl987nope i mean mobile phone05:11
xiangfujyfl987, then your mobile phone need support USB-HOST and usb-either network.05:15
xiangfuusb-eithernet 05:15
xiangfuor ben nanonote ---USB--> OpenWrt router(703n mr11u) --WiFi---> Your mobile phone --> 3G --> Internet05:16
jyfl987not a good idea, why not connect NN to your mobile phone directly?05:17
xiangfujyfl987, as I said your mobile phone need support USB-HOST.05:17
jyfl987i dont know how to check it, but i think modem mobile phone might support that 05:20
cladamwwpwrak, see http://dpaste.com/751062/09:44
cladamwwpwrak, in GUI, how can I select 'marker' frame to add vector ? I found that in my *.fpd, those __0: vec @(-x/2-r-clear, -y/2+r)__1: vec .(0mm, r) is not within "marker' macro. And surely in GUI screen will get blue background. Of course I can 'manually' copy vectors into frame macro, but I'd like to do this in GUI. What else i can do ? or just use "Swap var&code" ?09:50
wpwrakcladamw: you just click on the name ("marker") in the list of frames. then what you do on the right side happens in the marker frame10:10
wpwrakcladamw: note: if you later click on an item that's in a different frame, fped may change the current frame to the one containing that item10:11
cladamwso select frame first, then add vector, etc ... at that those vectors would be included in 'marker' ?10:13
wpwrakwhen you add vectors, they're always added to the current frame10:14
cladamwhmm ... need to get used to .... seems the 'start' and 'end' point are hard to click. :(10:15
cladamwand period of clicking changes frame layer.10:16
wpwrakbut you'll have to connect the "marker" frame first. so you select the frame in which it should be placed, then drag "frame" on the location where its origin should be (the origin is the fat little circle)10:16
wpwrakso the complete sequence for creating a new frame and then adding things to it is:10:17
wpwrak1) "Add frame" on the left side10:17
wpwrak2) edit the name of the new frame (default name "_")10:17
wpwrak3) select the parent frame10:18
wpwrak4) drag the new frame (click and hold the name, then drag it into the canvas) to its starting location10:18
wpwrak5) select the new frame (on the left side again)10:18
wpwrak6) so to the canvas and add the things you want there, starting from the frame's origin10:19
wpwraknote that a frame can have multiple parents. e.g., if you have a vector (10, 0) and a vector (-10, 0) in the root frame, you can attach "marker" (or some other frame) to each vectors10:21
cladamwah ... seems 3) and 5) i missed or made wrong, alright, i try these 6 steps.10:21
wpwraklast but not least, the place where you attach a frame is the end of a vector in the parent frame or the parent frame's origin. so you don't have to make a vector (0, 0) to connect something to the origin10:22
wpwrak(3-5) yes, if you don't attach the frame, then the screen stays dark :)10:22
cladamwand why Fped generates many filename~#.fpd ?10:28
wpwrakthese are older versions you've saved. that's to make sure that you have backups :)10:30
LunaVoraxHello everyone!11:17
wolfspraulhi :-)12:25
qi-bot[commit] Maarten ter Huurne: Re-enable zram and zcache on MIPS. (jz-3.4) http://qi-hw.com/p/qi-kernel/262b3bc12:39
viricwhat is zram/zcache? compressed ram?12:39
mthwe use it as swap space on the Dingoo12:39
viricis that on vanilla linux?12:39
viricah is that a block device? a fs?12:40
mthit's in staging12:40
mthzram is a block device12:40
viricthe dingo has 32MB of ram?12:40
mthyes, most models do12:40
virichm how fast do snes emulators run there?12:40
viricI have a gp2x, and they are too slow12:40
mththe 3.4 kernel made it x86-only, but I looked at the code and the call it needed was available on MIPS too, but with a slightly different name12:40
mthdepends on the game, some SNES games run at 60 fps, some slower12:41
viric(I've seen the idff)12:41
mthbut afaik all are at a playable speed12:41
viricAren't 25fps enough for the eye?12:41
mthyes, but the originals run on 60 fps, so that's the holy grail12:41
viricso anything slower than 60fps will be noticeable12:42
mthask Ayla when he returns, he has ported several emulators12:42
mthif you're very sensitive to it, yes, but in practice everything above 30 fps is good enough12:42
viricah ok12:42
viricI never understood before about wanting 60fps :)12:42
viricdid you do any work for open2x or the gp2x in general?12:43
mthno, the Dingoo was my first handheld12:43
viricand the main os for dingo is openwrt?12:43
mthactually I ordered a Pandora first, but since it was delayed and delayed, I got a Dingoo so I'd actually have something to play with12:43
viricmain, as in most broadly deployed12:43
mththere is no openwrt for Dingoo12:44
mththere is the native OS, uCos-II I think12:44
viricand you are making openwrt work for the dingoo?12:44
mthand there is the original Dingux (we call it "legacy" now), which is buildroot + ported Ingenic kernel12:44
viricor you only update the kernel?12:44
mthand OpenDingux, which is a more recent buildroot + ported qi-kernel12:45
viricbuildroot = debian?12:45
mthno, it's a build system for embedded Linux12:45
viricah ok12:45
mthnot really a full distro, but it has a decent amount of packages12:45
viricI wonder if there is gp2x support in mailine linux12:46
mthI think OpenWRT started as a buildroot fork, or at least adopted its build system as a base12:46
mthwe try to get as much as possible included upstream, but the Dingoo specific drivers are not in good enough shape for that yet12:47
viricwell I meant gp2x this time12:47
mthI don't know about gp2x12:47
mthprobably not from GPH, but maybe open2x did?12:47
viricthe open2x repositories fell down some time ago12:48
viricah no they are there12:48
virichm 2.4 still12:49
viricnoone ported the gp2x kernel part to 2.6 I imagine12:49
mthI always try to port our kernel to new mainline releases as soon as possible12:50
mthto avoid getting a backlog and being stuck at an old kernel12:50
viric'a backlog'?12:50
mthdelayed maintenance12:51
mth"an accumulation of tasks unperformed or materials not processed" according to the dictionary12:51
mthgoing from 2.4 to 3.x is probably a few weeks worth of effort, so it's unlikely someone will do it12:52
mthbut spending a few hours per release is quite feasible12:52
mthalso, if you want to submit patches upstream, you've got to be on a recent release12:53
viricwell, it could be '2.4 upstream' :)12:54
mthdo they still accept patches other than bugfixes?12:54
viricI don't think so12:55
viricat what linux version linux moved to git?12:56
mthv2.6.11 is the oldest branch I have in my repo12:56
viricbut I think that's when linux made a new '.git' from scrathc12:57
viricto avoid using too much disk space12:57
mthand I have kernel.org history in there, so it should be the oldest existing version in git anywhere12:57
mthat least the oldest developeed in git, you can import any version of course12:57
mthI can't imagine maintaining any software without revision control12:58
mthalthough Linus hates SVN, I still think SVN is much better than nothing at all12:59
viricof course12:59
viricthey were using bitkeeper no?12:59
virictime goes by...12:59
viricI don't know what did they use before bitkeeper12:59
viriccvs I imagine.13:00
mththey used nothing before bitkeeper13:01
mthjust mail + some scripts13:01
viricI can't believe it13:02
mthit's true, although I can't understand it either13:02
viricwell, at that time there were almost no DVCS13:03
mthespecially design criterium #113:05
mthfor me the distributed nature is not the major selling point of the new systems13:06
mthbut that they treat a commit like the fundamental unit, rather than a tree state13:07
mththat's what makes merging in SVN so painful: you can only merge differences between two tree states onto a third tree, you cannot merge actual commits13:07
mthso you lose history when merging13:08
mthin practice, people have to keep a lot of administration outside of SVN to be able to merge between different branches13:08
viricit looks like you speak of darcs, and not of git13:08
viricin git, every 'commit' hash refers to a tree.13:09
mthdarcs was the first system like that, but git takes more or less the same approach13:09
viricgit manages trees13:09
viricit simply has many tricks to transform one tree to another13:09
viric(hence, changing commit hashes)13:09
viric- that's my view of git :)13:09
mthit changes commit hashes only for cherry-picking afaik13:10
mthif you merge, you create a merge node that has two parents13:10
viricany rebase too13:11
viricCommit Object13:11
viricThe "commit" object links a physical state of a tree with a description of how we got there and why.13:11
viric*physical state of a tree*13:12
viricnot like darcs.13:12
viric- let me add that I don't like git much :)13:12
mthwell, I both like and dislike it ;)13:14
mthI like what it can do, but I think it is overly complex13:14
mththe git commit links a physical state with its history; an SVN commit is just a physical state without history13:18
viricwell, one revision always comes based on the previous revision13:20
mthyes, but as soon as you merge you only transfer state, not history13:20
viricwell, it knows that you merged and what you merged13:21
mthit knows the diff you merged, not the commits13:22
viricwell, since 1.5 or 1.6 it has the merge info13:26
mthah ok, I haven't used SVN in complex projects recently13:27
viricbut other than that, if you merge branches, it notes what the merge comes from13:27
mthopenMSX uses SVN, but we hardly ever branch13:27
viricwhat's openmsx about? :)13:28
virican emulator, or an OS for msx?13:28
mthMSX emulator13:28
mthwe have a custom firmware too, C-BIOS, but that is a separate project13:28
viricever run on hw?13:30
mthyes, we had someone put C-BIOS in an EPROM and it runs fine on real hw13:30
mthwell, the current version does, the early versions did I/O too fast and the video RAM contents would get corrupted13:30
mthhi Ayla13:34
mthwe got zram swap back :)13:34
mthalso, viric was interested in SNES emulation13:35
viric(for the nanonote, I don't have any dingoo)13:35
mththat should be just a compile away13:35
Aylaah, great13:37
AylaI have a build of pocketSNES, but you will need libao13:38
Aylamth: you found anything about the logo?13:41
mthno, I still haven't figured out how it works13:41
Aylathis might interest you guys: http://www.phoronix.com/scan.php?page=news_item&px=MTEwNTA13:46
qi-bot[commit] Maarten ter Huurne: MIPS: A320: defconfig: Re-enable zram in kernel config. (jz-3.4) http://qi-hw.com/p/qi-kernel/58d111115:05
qi-bot[commit] Werner Almesberger: b2/subst.c (prepare_re): fail on bad (#unit) syntax instead of ignoring (master) http://qi-hw.com/p/eda-tools/2946b8316:54
qi-bot[commit] Werner Almesberger: b2/: meta-variable $ for "previously matched variable" (master) http://qi-hw.com/p/eda-tools/3d2eca316:54
qi-bot[commit] Werner Almesberger: b2/: regexec puts first substring into pmatch[1] and not pmatch[0] (master) http://qi-hw.com/p/eda-tools/148d5e816:54
qi-bot[commit] Werner Almesberger: b2/: use re_nsub instead of maintaining our own equivalent (master) http://qi-hw.com/p/eda-tools/217027816:54
qi-bot[commit] Werner Almesberger: b2/subst.c (resolve_jump): use pointer comparison instead of strcmp (master) http://qi-hw.com/p/eda-tools/2ce2da519:21
qi-bot[commit] Werner Almesberger: b2/: make "again" include the match; make "break" pop the stack (master) http://qi-hw.com/p/eda-tools/bb809b019:21
qi-bot[commit] Werner Almesberger: b2/: rename "again" to "continue" (master) http://qi-hw.com/p/eda-tools/107280b19:21
qi-bot[commit] Werner Almesberger: b2/: make the match name after "break" and "continue" optional (master) http://qi-hw.com/p/eda-tools/694d48b19:21
qi-bot[commit] Werner Almesberger: b2/: for consistency, make "break/continue $" equivalent to "break/continue" (master) http://qi-hw.com/p/eda-tools/4f142c119:21
qi-bot[commit] Werner Almesberger: b2/Makefile: generate dependencies also for YACC- and LEX-generated files (master) http://qi-hw.com/p/eda-tools/c9bd26a19:21
qi-bot[commit] Werner Almesberger: Makefile: generate dependencies also for YACC- and LEX-generated files (master) http://qi-hw.com/p/fped/6623ae119:21
qi-botThe build was successful: http://fidelio.qi-hardware.com/~xiangfu/build-nanonote/openwrt-xburst.full_system-20120521-1428 20:22
--- Wed May 23 201200:00

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