**** BEGIN LOGGING AT Sun Nov 18 02:59:56 2007 Nov 18 03:09:52 hmm...wow this room is up and down Nov 18 03:10:03 looks like cable.rcn.com is not hacving a good nigt Nov 18 03:12:57 yeah Nov 18 03:15:22 anyone know about dbus on moko? Nov 18 03:15:27 I'm having real problems with it Nov 18 03:32:36 I've looked about on a few sites, and wondered if there are any images of openmoko bootable, say with qemu or something similar ? i'd like to play with things if it's possible for a while. thanks. Nov 18 03:35:06 bluh! Nov 18 03:35:22 run-parts on busybox has no --verbose option Nov 18 03:35:27 how to debug... grrr Nov 18 03:40:55 morning Nov 18 03:42:06 morning Nov 18 03:42:31 night Nov 18 03:43:01 access denied on dbus :\ Nov 18 03:45:32 Interesting that this O'Reilly onlamp article uses the Neo handset graphic in all it's diagrams: http://www.onlamp.com/pub/a/onlamp/2007/11/12/google-calling-inside-the-gphone-sdk.html Nov 18 03:47:55 blah, strace doesn't have a full dictionary on moko Nov 18 03:49:33 anyone know how I can get my phone back working? x-window-manager (matchbox-wm) isn't getting around to launching all the apps it's supposed to according to /etc/matchbox/session Nov 18 04:18:04 Hmm: | ERROR: Cannot satisfy the following dependencies for task-openmoko-ui: Nov 18 04:18:05 | gsmd-plugin-machine-gta01 Nov 18 04:18:16 (during do_rootfs) Nov 18 04:19:09 matchbox-window-manager, as far as I can tell... doesn't call /etc/matchbox/session during startup Nov 18 04:19:15 I can't figure out what's supposed to issue this Nov 18 04:22:06 rwhitby: Yep, been that way all day... I'm waiting for a clean build to finish, but I expect mine will fail the same way. Nov 18 04:22:27 libgsmd doesn't emit that package Nov 18 04:26:08 rebuilding libgsmd and task-openmoko-ui doesn't fix it Nov 18 04:26:22 and it's not obvious how task-openmoko-ui depends on that package Nov 18 04:29:58 am I supposed to be using matchbox-window-manager? Nov 18 04:30:10 perhaps there's something else that I'm supposed to be using and I don't know it Nov 18 04:31:36 mwester: gsmd-plugins depends on gsmd-plugin-machine-gta01 according to depends.dot (bitbake -g) Nov 18 04:32:29 ERROR: Cannot satisfy the following dependencies for task-openmoko-ui: Nov 18 04:32:36 hachi: matchbox-window-manager is certainly what's built and put in the image. Nov 18 04:32:41 gdk-pixbuf-loader-png gdk-pixbuf-loader-gif gdk-pixbuf-loader-xpm gdk-pixbuf-loader-jpeg gsmd-plugin-machine-gta01 Nov 18 04:32:58 That's a clean build. Looks like it's a bit more than just the plugin. :( Nov 18 04:33:21 Actually, probably only one other package, now that I look closer. Nov 18 04:33:46 rwhitby: matchbox-window-manager is launching just fine on my phone, talking over dbus... yet it never makes a call to /etc/matchbox/session to launch all the other necessary apps to make my phone usable Nov 18 04:34:05 I've checked out the matchbox-window-manager source, I can't see anything that calls this file Nov 18 04:34:15 grepped in /etc/X11/Xsession.d/ Nov 18 04:34:37 I can't find who the devil is supposed to be executing that file to set up my X interface for everything Nov 18 04:38:34 mwester: seems that latest gsmd svn source has "machine" gta01, rather than "vendor" gta01. Nov 18 04:40:00 so that matches what task-openmoko-ui is looking for. Nov 18 04:40:12 but machine gta01 plugin doesn't seem to be built Nov 18 04:40:15 I guess machine is technically more correct, but still... Nov 18 04:42:29 machine gta01 is in Makefile.am ... Nov 18 04:48:16 mwester: looks like an sane svnrev thing Nov 18 04:49:02 mwester: what version of libgsmd is being built for you? Nov 18 04:52:20 for some reason, snvr2806-r24 is building for me, whereas sane-srcrevs says 3074 should be built Nov 18 04:52:48 (and latest change in svn is r3430) Nov 18 04:54:53 * rwhitby blows away all trace of libgsmd and rebuilds Nov 18 04:56:01 ok, scratch that - seems it moved to machine-specific build Nov 18 04:56:32 so I am building svnr3074-r38 Nov 18 04:57:34 still no machine gta01 built in there Nov 18 04:58:26 looks like machine gta01 plugin didn't exist in Makefile.am in rev 3074 Nov 18 04:58:33 did the moko makefile start hiding the commands it's running in the last week? Nov 18 04:58:44 hachi: nothing has changed in a number of weeks Nov 18 04:58:48 ugh, running task 993 of 5051 Nov 18 04:58:57 I guess I might have a working build in a week or two Nov 18 04:59:27 ok, I thought it had been a lot more explicit, obviously wrong :) Nov 18 05:00:16 * rwhitby tries SRCREV_pn-libgsmd ?= "3434" Nov 18 05:04:10 does the gsm device in the neo identify itself to the network differently than other phones? Nov 18 05:05:05 cause I can't get gprs working on it... LCP is fine now, CCP is fine, auth is fine (CHAP for some reason) but IPCP is unable to get an ipv4 configuration from the far end Nov 18 05:05:25 mwester: machine gta01 plugin builds with rev 3434 of libgsmd Nov 18 05:05:54 mwester: and rootfs completes/ Nov 18 05:06:05 now, where is my neo to test it on ... Nov 18 05:06:23 you left it on top of your microwave Nov 18 05:08:09 doh! left them both at work :-( Nov 18 05:09:05 next error is in qemu: /home/moko/openmoko/trunk/src/host/qemu-neo1973/audio/audio.c:1912: fatal error\ Nov 18 05:09:05 : opening dependency file audio/audio.d: No such file or directory Nov 18 05:15:36 * mwester finds the missing gtk libs, and goes to figure out why they didn't get installed... Nov 18 05:37:13 * mwester finds the problem with the missing gtk libs. Nov 18 05:50:36 Hello! Nov 18 05:50:54 So - how is typing on the openmoko? Nov 18 06:10:00 aha, it was mickey_away who added the gta01 plugin to gsmd.inc, but forgot to update sane-srcrevs.inc Nov 18 06:10:20 (about 14 hours ago) Nov 18 06:10:58 And it was likewise who commited (in error) a hacked up gtk+ bb file along with his squashfs changes. Nov 18 06:15:54 pushing a fix for the gtk+ problem into OE now. Nov 18 06:30:50 mail them the commit and ditch pony Nov 18 06:30:53 ? Nov 18 07:29:15 I am... heavily impressed by this build system Nov 18 08:00:42 roh: I found out how I got that neo Nov 18 08:01:03 apparently my friend knows sean... sean gave it to him or something Nov 18 08:38:43 good morning Nov 18 09:54:09 good morning Nov 18 09:58:23 pushing the libgsmd fix now Nov 18 10:04:19 whats up folks. Nov 18 10:04:35 things clearing up with the build mess? Nov 18 10:07:55 hello folks Nov 18 10:08:27 rwhitby fixed it torpor Nov 18 10:08:29 howdy Nov 18 10:08:31 ok Nov 18 10:09:19 ScaredyCat: mwester fixed the other bit Nov 18 10:09:25 getting ready for a days worth of hacking finally .. Nov 18 10:09:26 together it should build now Nov 18 10:09:32 ok Nov 18 10:09:45 I seem to still be a bit far from running linux on my phone, but I'd like to know, is syncing of PIM data supported in OpenMoko for example, also with ActiveSync (which may be the only option sometimes, at the office)? Nov 18 10:10:14 Zoolooc: such features are possible, if there is activesync for regular linux, it can easily be ported to the openmoko environment Nov 18 10:10:24 but i dont think anyone is working on it as a priority as of today Nov 18 10:11:11 alright, I also doubt there is already something like this. I might ask the synce folks Nov 18 10:11:30 the thing is, its quite easy to port Nov 18 10:11:42 i think there is a python tool for handling these sync issues .. that'd be easy to get running Nov 18 10:12:18 treat the neo1973 as if its a regular linux workstation, running a linux distribution that supports compatability with a lot of already-existing linux software. Nov 18 10:12:30 more than likely there will be a way to do sync'ing properly.. but it could be that its just not there yet. Nov 18 10:12:50 you're welcome to fix that, though. ;) Nov 18 10:14:25 is there anyone who frequents here that could help me with my matchbox-wm issue? Nov 18 10:14:42 whats the problem? Nov 18 10:14:45 hachi: I never managed to get around it only using ipkg Nov 18 10:14:58 torpor: who knows, I might just need it and try to get active then ;) Nov 18 10:15:14 matchbox-wm launches, but the file /etc/matchbox/session never gets executed Nov 18 10:15:25 There's some conflict between machbox-common and openmoko-session2 I think. At least I had that before Nov 18 10:15:29 Zoolooc: thats what its all about Nov 18 10:15:30 so my phone shows 'please wait' all the time Nov 18 10:15:46 hachi: i think that problem is being worked on right now .. Nov 18 10:15:53 or was fixed in the latest build. Nov 18 10:16:01 hachi: If you have anything you need to save, move it to /media/card/, then reflash Nov 18 10:16:08 i'm doing a build now, and will see .. i also have that problem. Nov 18 10:16:14 It works now, I built and flashed this morning Nov 18 10:16:16 I don't have a jtag Nov 18 10:16:23 all the stuff says don't flash unless you have jtag Nov 18 10:16:28 hachi: Me neither, but flashing kernel and rootfs is safe Nov 18 10:16:32 Just don't flash uboot Nov 18 10:16:32 rwhitby: ahh sorry mwester... Nov 18 10:16:49 CM: but this bothers me Nov 18 10:16:55 the 'oh just reinstall the OS' is the windows method Nov 18 10:17:07 Yes, sure is.. Nov 18 10:17:13 I tried to figure out was wrong, read the matchbox-wm source, couldn't find what's supposed to invoke that file Nov 18 10:17:38 grepped /etc/X11/Xsession.d (since grep has no -r option in busybox) Nov 18 10:17:39 It's a combination of ipkg being a bit limited, and the libs being in an early state and changing a lot Nov 18 10:18:01 is there an image I can install from? Nov 18 10:18:09 because I've been compiling for like 24 hours now Nov 18 10:18:21 Running task 1518 of 5051 Nov 18 10:18:48 hachi: You can get the one I just flashed, just a sec Nov 18 10:18:55 okiedokie Nov 18 10:19:25 hachi: you do realize that the neo1973 is a *developers* phone right now? you're supposed to be updating the kernel+rootfs on a frequent basis, as things change daily, weekly, with the base system install. Nov 18 10:19:42 torpor: I just got the phone on... wednesday? Nov 18 10:19:50 hachi: and let your compile finish - its important. once its finished you can do "make update openmoko-devel-image" every day and get the new image for your own testing. Nov 18 10:19:51 hachi: http://claesmogren.com/~cm/neo/2007.2/ Nov 18 10:19:57 I'm trying to catch up fast Nov 18 10:20:01 hachi: but you shouldn't have gotten it without realizing that you're going to be flashing a lot Nov 18 10:20:06 just don't flash u-boot and you'll be fine. Nov 18 10:20:34 this 'windows method' thing is not applicable here, because in fact you are participating in the development of an OS. its not a user OS. yet. Nov 18 10:20:40 hachi: Yes, forget about the uboot I have there, It's just a folder where I have the same images as I have working on my Neo Nov 18 10:20:46 I will Nov 18 10:21:02 torpor: sorry, I landed this phone entirely by accident Nov 18 10:21:04 try to be patient; i'm only warning you because there may be wrinkles with the OS releases in the coming weeks which mean frustration if you're only trying to use the phone. Nov 18 10:21:10 I wasn't planning to buy one till the GTA02 myself Nov 18 10:21:17 so I'll try to calm down Nov 18 10:21:33 hachi: Don't worry, it's frustrating at times Nov 18 10:21:36 well, just be aware: things change a lot with Openmoko right now, and its a rocket ride through the progressive development stages of an OS. Nov 18 10:21:48 I'm fine with that Nov 18 10:22:07 I removed a transistor from my neo yesterday to make it work better Nov 18 10:22:08 it is frustrating. we all share that frustration. but if you've got a neo1973, it means you're a developer, and that should be enough of a clue that you've gotta be able to weather the storm like other developers do. its not a user phone yet. Nov 18 10:22:36 well, yes... that's why I was reading the source to matchbox-wm and trying to diagnose the startup Nov 18 10:22:41 it ain't windows, either. Nov 18 10:22:42 ;) Nov 18 10:22:53 I hadn't seen any notes about it being dead, nor anyone fixing it Nov 18 10:23:07 I don't know how to subscribe to this information yet, that's sorta where I'm getting stuck Nov 18 10:23:14 the thing you should be aware of is that things change a *lot* on a daily basis. you have to catch your build up with the rest of us, then do daily make update openmoko-devel-image to stay current. Nov 18 10:23:30 hachi: there definitely needs to be a release engineer on the openmoko project, but there isn't right now. Nov 18 10:23:53 I'm asking... in a roudnabout way Nov 18 10:23:57 one thing we have all suffered from is that there is no real 'official weekly build' thats been through a test state. we all have issues with that, which is why #openmoko is important. for now, we update each other daily. Nov 18 10:24:00 how did you know that it was being worked on? Nov 18 10:24:09 this channel? Nov 18 10:24:14 because i do a daily make update and i see the changes. Nov 18 10:24:18 also, because of this channel. Nov 18 10:24:38 is there an easy way to get a diff? *goes reading the makefile* Nov 18 10:24:40 this channel is a primary point of support for active developers on openmoko right now.. there really aren't any other resources - no openmokoslash.org or anything .. yet .. Nov 18 10:24:46 well, a commit log actually Nov 18 10:25:03 when you do 'make update' you'll see the differences. yes, there is a commit log - you can subscribe to the commits mailing list and get updates. Nov 18 10:25:31 hachi: http://lists.openmoko.org/pipermail Nov 18 10:25:57 Sorry, forget the pipermail part, I was reading the commitlogs Nov 18 10:26:07 I'll assume you meant mailman Nov 18 10:26:12 cause pipermail don't work :) Nov 18 10:26:17 Yes Nov 18 10:26:29 there are comit logs Nov 18 10:26:35 -devel is the commits list? Nov 18 10:26:38 http://lists.openmoko.org/pipermail/commitlog/2007-November/date.html Nov 18 10:26:41 wait, missed commit log Nov 18 10:26:42 sorry Nov 18 10:26:48 blind me Nov 18 10:27:13 hachi: -devel is ok, too, community is too much noise for me Nov 18 10:27:22 good opinion Nov 18 10:27:53 yay Nov 18 10:27:59 * ScaredyCat points at scap Nov 18 10:28:19 hi Nov 18 10:29:10 ScaredyCat: Ahh... Cool :) Nov 18 10:29:15 cool Nov 18 10:29:16 :) Nov 18 10:30:41 ScaredyCat: I know you have an enormous amount of song soo, you show-off! :P Nov 18 10:30:45 songs.. Nov 18 10:30:57 I'm gonna make frowny faces on this channel if I have to debrick my phone tonight :) Nov 18 10:31:24 hachi: Feel free to blame me if you do Nov 18 10:31:45 * CM is off to buy food Nov 18 10:31:45 I'll say your name and frown, OK Nov 18 10:31:49 Hehe Nov 18 10:31:55 CM: oooh yes.. Nov 18 10:32:50 should I do a nand erase? Nov 18 10:32:51 but the web browser is unstable in om Nov 18 10:32:58 since I don't know if this image is large or smaller Nov 18 10:35:25 hachi: yeh, unless you have newish u-boot which does that anyway Nov 18 10:35:31 aand it is a shame that fullscreen isn't :( Nov 18 10:37:14 hachi: Be very careful with nand erase Nov 18 10:37:26 hachi: If you do only "nand erase" you'll have a brick Nov 18 10:37:52 if I do that, then I'll buy a jtag Nov 18 10:37:56 but I'll frown first Nov 18 10:37:59 :) Nov 18 10:38:10 hachi: Since you have a P0, you probably have an old uboot though.. Nov 18 10:38:20 I' getting a version Nov 18 10:38:24 hachi: I've updated my uboot 4 times, and I don't have a jtag Nov 18 10:38:30 remember it's: Nov 18 10:38:32 :-(® Nov 18 10:38:43 1.2.0-moko9_r0 Nov 18 10:38:55 aug 19th Nov 18 10:39:06 Hmm, I think that one's ok Nov 18 10:39:07 I assum ethat's not 'recent ish' by definition in this setup? Nov 18 10:39:11 oh Nov 18 10:39:13 for me any uboot > 23rd August has issues... Nov 18 10:39:20 race condition... Nov 18 10:39:32 It's from the 2007.08 "official" release Nov 18 10:39:37 easily solved but usung ux button, but worth being aware of Nov 18 10:40:17 ScaredyCat: Did that august 19 uboot have the nand erase fix? Nov 18 10:40:22 I think it did.. Nov 18 10:40:30 CM, tell me Nov 18 10:40:32 yeah, methinks it might... Nov 18 10:40:42 does the u-boot in that directory you gave me work Nov 18 10:40:58 hachi: Yes, for me and ScaredyCat it did at least Nov 18 10:41:02 if so, I'll just flash it... I don't fear this really Nov 18 10:41:15 ScaredyCat: I think it's yours from august 31 Nov 18 10:41:25 CM: ScaredyCat: btw, i'm gonna use that uboot today to set up root-on-sd .. Nov 18 10:41:37 blibble Nov 18 10:41:43 :D Nov 18 10:41:49 CM : 31st? Nov 18 10:41:51 :O Nov 18 10:42:06 ScaredyCat: Yes, it's bad? August 31 that is Nov 18 10:42:12 It's been working ok for me.. Nov 18 10:42:29 But I haven't tried any root on SD though Nov 18 10:42:44 the sd one is in the root dir... Nov 18 10:43:13 hachi: ScaredyCat's files are in http://buildhost.automated.it/ Nov 18 10:43:17 none of the others seem to work properly... they return 0 size Nov 18 10:43:45 http://buildhost.automated.it/u-boot-good-for-sd-boot-r13_0_2632_0.bin works for SD cards Nov 18 10:43:49 I... don't have an SD card in my phone Nov 18 10:43:57 I put a piece of paper in to prevent shorting Nov 18 10:44:07 but the guy who gave me the phone kept the SD card :) Nov 18 10:44:40 hachi: you can use any sd card, at least up to 4gb will work Nov 18 10:44:51 I don't have any Nov 18 10:44:55 that's the problem Nov 18 10:44:57 ok :) Nov 18 10:44:59 I'll buy one next week Nov 18 10:45:05 but thanks for that info Nov 18 11:01:31 hey... uh CM Nov 18 11:01:44 dfu-util reports no devices to flash to Nov 18 11:02:40 is the neo in the right state to accept it (hold aux while power on)? Nov 18 11:02:41 reading the wiki pages again Nov 18 11:02:47 yeah, I'm at the uboot menu Nov 18 11:03:05 I can connect to it as ttyACM0 just fine right now Nov 18 11:04:23 I don't have any devices on the bus that are id 0x1457:0x5119 Nov 18 11:05:11 ahh, you have to be root on the host to do this Nov 18 11:13:27 hachi: also unplug the cable and put the cable back in .. Nov 18 11:14:58 nah, just needed permissions Nov 18 11:15:19 I'm so used to doing as little as I can as root that I forgot to even try it Nov 18 11:16:16 eh Nov 18 11:16:29 oO Nov 18 11:16:48 eh Nov 18 11:17:12 it just made a noise... clickie sound Nov 18 11:18:51 CM: ping Nov 18 11:20:22 ScaredyCat: pong Nov 18 11:20:40 did you get audio (ringback) on your calls Nov 18 11:20:43 test calls Nov 18 11:21:01 I did try to call, but I called myself Nov 18 11:21:12 I'll try and see if I get any audio Nov 18 11:22:56 Hmm.. Weird Nov 18 11:23:22 I was unable to answer, since the dialer didn't pop up on incoming calls, it just buzzed Nov 18 11:24:29 when you dial from the neo can you hear the ringingngingigng Nov 18 11:24:29 ScaredyCat: Ok, audio worked fine Nov 18 11:24:35 Yes Nov 18 11:24:37 k Nov 18 11:24:49 I did not get any boot up sound though Nov 18 11:25:01 make sure you have the right kernel Nov 18 11:25:01 Or ring tone when calling to the Neo Nov 18 11:25:23 torpor: I just used the latest built by the makefile Nov 18 11:25:36 hurray, outbound calls working for me Nov 18 11:25:43 hachi: Nice :) Nov 18 11:29:21 hmmmm Nov 18 11:30:34 * ScaredyCat points at scap again... Nov 18 11:30:53 I get that error, but it connects to the network fine Nov 18 11:31:05 ScaredyCat: I get that too Nov 18 11:31:44 But after another pop up with that "connected to network" test it works again Nov 18 11:31:59 yah... Nov 18 11:32:10 I get no audio though.. at all Nov 18 11:32:20 Strange Nov 18 11:33:34 if I want to program with python and use gps data, how do I do that? is there a gpsd module for python and is there and documentation? Nov 18 11:34:29 guys, i'm panicked, can't hear anything from neo. i suspect somehow i've burned the audio chip, or sth like that.. Nov 18 11:34:54 emre_: Do you have the right kernel for the rootfs? Nov 18 11:35:39 I'm trying both the latest (1 hour ago) openmoko build and one of the first qtopia images. Nov 18 11:36:04 both is silent when I cat .. > /dev/audio Nov 18 11:36:12 Ok. Have you tried to run alsamixer and unmuted the speakers? Nov 18 11:36:35 Does lsmod show a lot of loaded modules? Nov 18 11:36:35 i've run the alsamixer but i don't know which parameters i should play with there. Nov 18 11:36:47 emre_: audio volumes on my phone were jacked till I reloaded proper settings from /etc/gsmhandset.state Nov 18 11:37:01 lsmod says: Module Size Used by Nov 18 11:37:01 sco 10020 2 Nov 18 11:37:01 nls_iso8859_1 3808 1 Nov 18 11:37:01 nls_cp437 5472 1 Nov 18 11:37:01 vfat 10176 1 Nov 18 11:37:02 fat 46396 1 vfat Nov 18 11:37:04 nls_base 6720 4 nls_iso8859_1,nls_cp437,vfat,fat Nov 18 11:37:06 snd_soc_neo1973_wm8753 10040 0 Nov 18 11:37:08 snd_soc_s3c24xx_i2s 3560 1 snd_soc_neo1973_wm8753 Nov 18 11:37:10 snd_soc_s3c24xx 4448 1 snd_soc_neo1973_wm8753 Nov 18 11:37:12 snd_soc_wm8753 28752 1 snd_soc_neo1973_wm8753 Nov 18 11:37:14 s3cmci 11396 0 Nov 18 11:37:16 rfcomm 35580 10 Nov 18 11:37:20 ohci_hcd 20844 0 Nov 18 11:37:22 hidp 14784 0 Nov 18 11:37:24 l2cap 20164 10 rfcomm,hidp Nov 18 11:37:26 hci_usb 15260 2 Nov 18 11:37:26 emre_: nopaste.org Nov 18 11:37:28 bluetooth 48512 9 sco,rfcomm,hidp,l2cap,hci_usb Nov 18 11:37:30 i loaded /etc/gsmhandset with alsactl -f /etc/gsmhandset.state restore Nov 18 11:37:43 emre_: I don't see any snd_* there Nov 18 11:37:58 there are 4 snd_soc_xxx here Nov 18 11:38:07 Or rather I do, but not the pcm and mixer things Nov 18 11:38:15 Tried restarting alsa? Nov 18 11:38:20 I have mixer stuff Nov 18 11:38:42 I have too, and without snd_pcm I doubt there will be much sound Nov 18 11:38:43 http://nopaste.snit.ch:8001/11552 Nov 18 11:39:07 how to restart alsa? Nov 18 11:39:32 /etc/init.d/alsa-state restart ? Nov 18 11:40:17 still the same. Nov 18 11:40:22 mmm.. seems better now Nov 18 11:40:31 tried modprobe snd_pcm ? Nov 18 11:40:33 hachi: I have the same sound modules loaded as you Nov 18 11:40:38 to see if maybe there's an error loading the module Nov 18 11:40:50 it says snd_pcm not found. Nov 18 11:41:00 emre_: What image do you use? Nov 18 11:41:01 ahh, well. theres your problem then Nov 18 11:41:13 right at the moment qtopia.. Nov 18 11:41:36 Ah, heh. Maybe they expect it to be built into the kernel? Nov 18 11:42:01 but the thing is, it was working.. Nov 18 11:42:33 I plugged in another earphone (not the one shipped with neo) and I guess after that the sound is stopped. Nov 18 11:43:01 does it work via the earhpones? Nov 18 11:43:06 no. Nov 18 11:43:35 let me try with other images I have.. Nov 18 11:44:15 one thing i don't get is when i burn a freshly-built image, then do an ipkg update and upgrade, there's tons of packages upgraded. Nov 18 11:44:22 how come thats not all in my freshly built image, eh? Nov 18 11:45:11 hilariously, I woldn't be doing that :) Nov 18 11:45:40 yell ... Nov 18 11:45:45 errm, Nov 18 11:46:04 well all the packages installed from the image, they would be the latest - yes Nov 18 11:46:19 assuming your image was made after any updates.... Nov 18 11:46:31 yeah Nov 18 11:46:34 it was Nov 18 11:46:41 i.e. a freshly done 'make update all' this morning .. Nov 18 11:47:08 but where are your ipkg updatesfrom? Nov 18 11:47:23 the default sources. Nov 18 11:47:37 plus i have a little thing called scaredycat.conf too .. Nov 18 11:48:29 yeah, so when my stuff gets built, if mine were made after yours you'll get mine Nov 18 11:48:41 iyswim Nov 18 11:48:48 hmm. Nov 18 11:49:41 you built it today? Nov 18 11:49:44 i.e. an hour ago? Nov 18 11:49:47 i do lots of make update, make of openmoko feed and my own extrast too Nov 18 11:49:49 yes Nov 18 11:49:58 twice today Nov 18 11:50:03 yeah i do too .. but i guess its not just scaredy repo that gets updates. Nov 18 11:50:07 ScaredyCat: what kind of work do you do? Nov 18 11:50:14 Upgrading openmoko-theme-standard2 on root from 0.1.1+svnr3289-r5 to 0.1.1+svnr3425-r5... Nov 18 11:50:17 Downloading http://buildhost.openmoko.org/OM2007.2/tmp/deploy/glibc/ipk/all/openmoko-theme-standard2_0.1.1+svnr3425-r5_all.ipk Nov 18 11:50:20 see? Nov 18 11:50:26 3rd update (packages ) running now Nov 18 11:50:45 ok Nov 18 11:50:48 different svn revision numbers Nov 18 11:50:54 also why isn't the SD card being mounted? Nov 18 11:51:06 please don't tell me its because the image doesn't have modules. Nov 18 11:51:08 i hate that. Nov 18 11:51:44 edistar: atm I do web stuff and drivers and stuff - I work for a company called MultiThread, we also run linitx.com Nov 18 11:52:01 it should do Nov 18 11:52:07 ok Nov 18 11:52:09 what format is the sd in? Nov 18 11:52:12 fat Nov 18 11:52:20 k, should be ok then Nov 18 11:52:22 mmmm Nov 18 11:52:25 not mounting Nov 18 11:52:44 even if you ' mount /media/card '? Nov 18 11:52:59 did you reformat it? Nov 18 11:53:19 nope Nov 18 11:53:22 no mmc device found Nov 18 11:53:32 shutting down and checking its inserted properly .. Nov 18 11:53:46 edistar: I get to play with a lot of linux stuff... we import quite a bit of hardware too... Nov 18 11:54:26 and since a lot of it is 'windows only' we like to try to add something for our linux customers too Nov 18 11:54:33 ah, cool Nov 18 11:54:52 ScaredyCat: good job. got any small-form factor PC's you wanna recommend, with built-in screens? Nov 18 11:54:56 miniitx formfactor? Nov 18 11:55:10 i wanna put a PC in a Virus KB keyboard (synthesizer) Nov 18 11:55:34 when you say built in screen - you mean lcd? Nov 18 11:55:49 or tft Nov 18 11:55:53 yeah Nov 18 11:56:01 ie a 40 x 4 line thing or a real vga type thing? Nov 18 11:56:12 the minibox ones are small... Nov 18 11:56:31 real vga Nov 18 11:56:48 No results for 'minibox' Nov 18 11:56:59 ok... well the only vga built in screens we have are indash jobbies.. Nov 18 11:57:13 ScaredyCat: which country? Nov 18 11:57:26 ost of our customers go for the linitx screens, then prefab their dash, so I guess you could do the same Nov 18 11:57:40 edistar: we're in the UK, we serive the world :) Nov 18 11:57:42 service Nov 18 11:58:22 just got a new carrier for shipping in Europe, much better for the customer - cheaper... and service as good as ups/fedex Nov 18 11:59:06 people want a steampunk case for the neo? Nov 18 11:59:07 torpor: http://linitx.com/search.php?keywords=mini-box&Submit=Search Nov 18 11:59:14 * ScaredyCat does Nov 18 11:59:17 hachi: Nov 18 11:59:34 yeah, I do too Nov 18 11:59:54 our lcd screens (linitx ones get good reviews) Nov 18 12:00:03 I'd love to have a brass case on the thing Nov 18 12:00:07 ScaredyCat: do you have thin clients? Nov 18 12:00:28 some, most of them are a bit lardy like me... :P Nov 18 12:00:51 http://linitx.com/viewcategory.php?catid=198&pp=198 Nov 18 12:00:54 edistar: Nov 18 12:01:05 they're very easy to build tbh... Nov 18 12:01:07 ScaredyCat: how much do they cost? my school needs some, you wouldn't mind sponsoring a german school, would you? :P Nov 18 12:01:42 standard hardware + something like thinstation... Nov 18 12:01:51 hehe, if only Nov 18 12:02:30 it really depends, if you particularly want small boxes, you end up paying a premium. Nov 18 12:03:15 just look at via's el cheapo itx range of boards, you can build a thin client for 100 - 135 eur Nov 18 12:03:16 ScaredyCat: just want to boot from an LTSP server, and have these clients really cheap Nov 18 12:03:45 iirc the mobo is 60 usd, which is like 40 eur or somehting Nov 18 12:04:07 ScaredyCat: which one? Nov 18 12:04:49 the via PC-1 Nov 18 12:04:54 k, off for food, bbiab... Nov 18 12:05:13 just msg me if you want some info Nov 18 12:05:17 ScaredyCat: thx Nov 18 12:13:54 i want an *anything* case for the neo Nov 18 12:13:57 sick of the plastic. Nov 18 12:14:13 I wan't a brushed aluminium case Nov 18 12:14:17 and a rugged one Nov 18 12:14:19 and a pony. Nov 18 12:14:21 okaY? Nov 18 12:16:12 everyone wants a pony Nov 18 12:16:45 Ima try and figure out why screen doesn't work on my neo, anyone done that yet? Nov 18 12:17:28 hachi: Not that I know Nov 18 12:17:44 aww, I think my phone just went to sleep Nov 18 12:18:55 CM, cool... I like it a lot better than 'cu' as a tty endpoint Nov 18 12:20:05 I run screen all the time everywhere :) Nov 18 12:20:28 but you don't run it on your phone Nov 18 12:20:32 has anybody tired to use qemu-neo1973/openmoko/download.sh and flash.sh today? The openmoko/flash.sh is failing for me, quemu gets stuck when it uses rootfs image. Nov 18 12:20:33 so, I'm gonna fix that Nov 18 12:20:42 Ecellent :) Nov 18 12:21:30 well carp Nov 18 12:21:32 it worked this time Nov 18 12:21:34 dinamich: Been problems with that lately, I'll try and see if I can get it running here Nov 18 12:22:15 maybe image `OpenMoko-openmoko-devel-image-glibc-ipk-P1-Snapshot-20071113-fic-gta01.rootfs.jffs2` is broken? Nov 18 12:22:32 dinamich: Are you using the moko makefile? Nov 18 12:22:49 hachi: screen works, it'll just take ages if your termcap is massie, like the default one Nov 18 12:23:00 massive Nov 18 12:23:07 ahh, that's what it is Nov 18 12:23:18 I just checkedout svn trunk and follwed wiki pages for quemu Nov 18 12:23:18 dang, I was hoping to fix something that I could handle :) Nov 18 12:24:42 dinamich: Ok. I'm building it with the MokoMakefile and that seems to work, booting now at least Nov 18 12:26:05 Heh, takes about 4 minutes to boot OpenMoko in qemu on my comp.. Nov 18 12:26:16 CM: i didn't use MokoMakefile, because i don't have 12G free space Nov 18 12:26:30 dinamich: Ah, ok. Hmm.. Nov 18 12:26:54 It also uses the latest svn trunk, so it should be the same Nov 18 12:27:03 CM: maybe i should increase timeout then in openmoko/flash.sh script!? Nov 18 12:27:14 No, that's not the problem I think Nov 18 12:27:38 Sometimes it doesn't find the rootfs at all, but it still gives that error message Nov 18 12:28:37 CM: well, i have celeron laptop about 1.6GHz and quemu is using full cpu, then script aborts with timeout Nov 18 12:29:40 Hmm.. then maybe. I have 4 year old amd64, and it just takes seconds to flash, not 5 minutes Nov 18 12:29:59 CM: the u-boot and kernel phases of flash.sh get done, so my quemu is probably working ok Nov 18 12:30:38 CM: but quemu with rootfs gets stuck, 100% cpu.... Nov 18 12:30:40 dinamich: openmoko/flash.sh has been getting stuck for a few days now.. Nov 18 12:30:58 dinamich: I get a timeout too Nov 18 12:31:17 edistar_: maybe it is not me after all Nov 18 12:31:34 dinamich: you didn't use the makefile, did you? Nov 18 12:32:10 edistar_: no, dont have enougs HDD space, just svn of trunk Nov 18 12:32:33 dinamich: thats what I did too, what does flash.sh say when it times out? Nov 18 12:35:59 edistar_: output of flash.sh at http://pastebin.com/d3b10fdbf (i modified rootfs timeout to 400 seconds) Nov 18 12:38:23 dinamich: I just did a "make qemu-local" and it flashed my locally built image Nov 18 12:38:36 It took a while to flash the rootfs, but it worked. Booting now.. Nov 18 12:39:26 dinamich: yep, thats the same problem that I have Nov 18 12:39:31 CM: maybe the snapshot that I am using is not good Nov 18 12:40:43 ok, thanks both edistar_ and CM, i think i will go makefile route (just need to free some HDD space:)) Nov 18 12:40:55 dinamich: Prepare for some buildtime then.. Nov 18 12:41:08 I'm guessing 12h or so on your laptop Nov 18 12:41:10 jeah;) Nov 18 12:42:01 bye Nov 18 12:42:04 later Nov 18 12:45:35 Heh, really sucks to run OpenMoko in qemu. Awfully slow.. Nov 18 12:45:55 CM: true Nov 18 12:46:05 CM: how do you run it if you don't have a device? Nov 18 12:46:18 I do have one.. :) Nov 18 12:46:36 Has anyone ever gotten Xoo to work properly? Nov 18 12:46:41 CM: that wasn't my question :) Nov 18 12:47:52 CM: I did once, yes Nov 18 12:47:56 CM: but it doesn't work anymore Nov 18 12:48:09 CM: I used Xephyr Nov 18 12:48:17 I tried a few times this summer, also using xephyr Nov 18 12:48:25 Got it to start, but no further than that Nov 18 12:49:04 the pixels were much to big, couldn't work with it ;) Nov 18 12:49:11 Hehe Nov 18 12:49:36 I am still waiting for someone to build OM for maemo devices.. then I could test it on my N770 ;) Nov 18 12:59:42 Huge but interesting post on version control systems: http://blogs.gnome.org/newren/2007/11/17/adoption-of-various-vcses/ Nov 18 13:04:05 someone earlier before I did my flash run told me where to properly store things I don't want to have blown away Nov 18 13:04:08 where is that? Nov 18 13:04:56 hachi: Heh, on that card you don't have yet Nov 18 13:05:08 I have a lot of stuff in /media/card/ :) Nov 18 13:05:25 ahh, okay, that explains why I don't see a logical mountpoint Nov 18 13:06:13 I'd like to see some attempt to run OpenMoko as a user instead of everything being root, but I guess that's quite some time away Nov 18 13:06:22 iPhone runs everything as root Nov 18 13:06:44 Android is cool though, does the total opposite. Every process supposedly is it's own user Nov 18 13:07:01 Of the user started processes that it Nov 18 13:07:02 is Nov 18 13:10:26 will openmoko run everything as root :-o Nov 18 13:10:28 ? Nov 18 13:10:53 It's supposed to be changed, but right now all is run as root Nov 18 13:11:23 ldap auth on phone Nov 18 13:11:27 wait no Nov 18 13:11:30 then one bad app can erase the root partition? thats very insecure :P Nov 18 13:11:31 kerberos auth Nov 18 13:11:43 hachi: ldap would be fun :P Nov 18 13:11:51 hachi: companies could enforce ldap login Nov 18 13:12:43 okay, anyone wanna explain to me why there's an ircd user in the passwd file on my phone? Nov 18 13:12:53 are you begging me to load an ircd on this thing? Nov 18 13:13:26 hachi: why not? Nov 18 13:13:27 also an lp user Nov 18 13:13:37 I need a USB dot matrix printer please Nov 18 13:14:21 hehe Nov 18 13:14:26 hachi: download one ;) Nov 18 13:14:50 aww, no perl built on moko Nov 18 13:15:02 guess I'll have to write things in C Nov 18 13:15:34 How big is perl? I assume there's an ipkg in oe for it Nov 18 13:16:15 too big, it comes with all the unicode tables and crap built into it these days Nov 18 13:16:31 [-------------------] <----- not to sale Nov 18 13:16:42 Ok. Python isn't that big Nov 18 13:16:49 ScaredyCat: I think your c key is broken Nov 18 13:16:58 bleh Nov 18 13:17:00 [-------------------] <----- not to scale Nov 18 13:17:18 python isn't evil enough for me Nov 18 13:17:19 just aking sure you were paying attention Nov 18 13:17:42 Hehe.. No c, no m, what are you trying to tell me? Nov 18 13:17:50 :P Nov 18 13:17:53 ;) Nov 18 13:20:23 awww... gps doesn't have a driver yet Nov 18 13:20:24 bleh Nov 18 13:20:29 that shoots my first idea Nov 18 13:20:50 hachi: Did you see ScaredyCat's scap? Nov 18 13:21:01 obviously not Nov 18 13:21:04 what's a scap? Nov 18 13:21:07 http://scap.linuxtogo.org/ Nov 18 13:21:12 hachi: Second image Nov 18 13:22:22 hachi: if you've configured networking on your neo, and take a screenshot (aux button), you can automaticly upload it to scaps.ltg.o just by clicking a button Nov 18 13:22:40 did I read the wiki wrong, or is that a result of the reverse engineering efforts? Nov 18 13:23:02 I'd say it's a wrapper around the oabi gllin binary driver.. Nov 18 13:39:06 cool screenshots Nov 18 13:58:50 how goes bluetooth on gta01? Nov 18 13:58:58 fine Nov 18 13:59:10 it's like bluetooth on any other linux machien Nov 18 14:00:47 thats great to hear Nov 18 14:01:17 It's using the normal bluez libs :) Nov 18 14:04:53 hello, is there any static qemu binary for linux as there is for windows? Nov 18 14:05:07 * mmp is too lazy to compile one himself:) Nov 18 14:06:35 mmp: ask your distro Nov 18 14:06:40 not the one that emulates neo1973 afaik Nov 18 14:07:02 but if you roll a package, feel free to upload it to the wiki Nov 18 14:07:18 well, I'll have to overcome my laziness:) Nov 18 14:07:26 The win binary is 5 months old.. Yikes Nov 18 14:08:08 perhaps dfu-util, devirginator etc could also have packages for some main distros put up on the wiki Nov 18 14:09:14 argh, again the gcc-3.3 limitation :-( Nov 18 14:09:49 Yes, qemu problem Nov 18 14:09:54 mmp: What distro are you on? Nov 18 14:10:04 mmp: under debian/ubuntu you can install gcc3 and gcc4 Nov 18 14:10:54 hum on the other hand maintaining the package up-to-date will be a bit of pain... Nov 18 14:11:34 * balrog_ will see about upstreaming Nov 18 14:11:56 CM: gentoo:) Nov 18 14:12:40 on gentoo i think both gcc's are present by defalt Nov 18 14:13:20 mmp: Me too. You might have to do a ln -s /usr/bin/gcc-3.4.6 /usr/bin/gcc-3.4 though, because qemu just looks for gcc-3 and gcc-3.x Nov 18 14:13:55 CM: well, if I had gcc-3.x at all :-/ Nov 18 14:14:03 I recently depcleaned whole system... Nov 18 14:14:14 Hehe, just emerge it again then :) Nov 18 14:16:16 * mmp mumbles something about wasted processor time and burnt legs caused by compilation on his athlon64 laptop Nov 18 14:16:27 CM: not anymore Nov 18 14:16:42 balrog_: Ah, nice :) Nov 18 14:17:13 balrog_: I did say "might" because it hasn't been as many questions about that lately as it was before Nov 18 14:17:26 ah okay Nov 18 14:17:42 you can also use --cc=gcc-7.5.15 Nov 18 14:18:01 7.5.15? Nov 18 14:18:04 What's that? Nov 18 14:18:23 i mean if you had a compiler under that name that ./configure doesn't know about :) Nov 18 14:18:37 Ah, ok.. :D Nov 18 14:25:44 Cool game for a tablet: http://www.youtube.com/watch?v=QsTqspnvAaI Nov 18 14:42:38 CM: cool Nov 18 14:42:41 is this opensource? Nov 18 14:43:59 borg_: can't find any mention of any licence on the website Nov 18 14:44:08 me too Nov 18 14:44:17 just for the pictures and sounds Nov 18 14:44:23 Music: _ghost - Lullaby. The song is used under the Creative Commons Attribution-NonCommercial 2.5 -license. Nov 18 14:44:26 The game uses Erin Catto’s Box2D physics engine. Nov 18 14:44:28 The graphics of the game are under the Creative Commons Attribution-NonCommercial-ShareAlike 2.0 -license. Nov 18 14:44:51 yes Nov 18 14:45:20 borg_: there is no sourcecode in the download Nov 18 14:45:31 but i think it isnt that hard to make a clon of it ;) Nov 18 14:46:23 borg_: no idea, a friend of mine tried to write a physics engine once, and failed.. but maybe he is just not such a genius ;) Nov 18 14:47:40 edistar: we had to program something similar at the animations course here in the university Nov 18 14:47:53 it is pretty easy to fake this gravity effects Nov 18 14:47:58 borg_: cool, do you have the source somewhere? Nov 18 14:48:01 heh Nov 18 14:48:22 i do Nov 18 14:50:16 edistar: http://cs.wcsu.edu/upb/ Nov 18 14:50:24 look at the assignments Nov 18 14:50:36 borg_: ok Nov 18 14:50:49 i can send you the source for this if you like Nov 18 14:50:53 but its java Nov 18 14:50:54 :\ Nov 18 14:51:13 hm.. I'm not into java that much, we did that at school, but I don't like it ;) Nov 18 14:51:21 yes, me too :) Nov 18 14:51:24 the hard part is collisions in physics engines Nov 18 14:51:44 ah ok, the first assignment is about collisions :) Nov 18 14:52:27 borg_: at least you get nice assignments, we got something like: make the program draw a table and make an input box, write a list class and sort the input Nov 18 14:52:40 borg_: that was fun... we did that kind of thing for about 2 years Nov 18 14:53:17 borg_: the kind of collisions where more than two bodies touch, for example when a ball is lying stable on three other balls Nov 18 14:53:47 edistar: not everybody gets to write games for a living ;) Nov 18 14:54:13 balrog_: ah ok Nov 18 14:55:25 * balrog_ wants to see videos of the bees assignment results :) Nov 18 14:55:51 balrog_: the results of assignment 2 were most amazing Nov 18 14:56:32 ewww .ppt Nov 18 14:59:48 Stephmw: true again.. Nov 18 15:00:40 how can I stop my build generating locales? Nov 18 15:02:44 lol Nov 18 15:02:45 i hate java Nov 18 15:02:56 i cant get my assignments running anymore :) Nov 18 15:02:58 edistar: ENABLE_BINARY_LOCALE_GENERATION = "0" in your local.conf Nov 18 15:03:13 abraxa_: how much time would you think does that save? Nov 18 15:03:16 abraxa_: thx Nov 18 15:03:24 who ever invented classpath Nov 18 15:03:28 "time" is sooo relative :) Nov 18 15:03:35 he should surely be killed :) Nov 18 15:04:18 borg_: the inventor of the classpath concept was a genius! talk about job security :) Nov 18 15:04:27 :D Nov 18 15:04:27 borg_: classpath? Nov 18 15:04:46 borg_: it's also damn useful for overlaying paths (unit testing, etc) Nov 18 15:05:16 Stephmw: it is usefull Nov 18 15:05:19 *ahem*, but back on topic Nov 18 15:05:23 but it is too complicated :) Nov 18 15:05:36 Stephmw: which topic? ;) Nov 18 15:06:01 edistar: http://en.wikipedia.org/wiki/Classpath Nov 18 15:06:55 borg_: thanks Nov 18 15:21:16 edistar: Saved over 1h for me Nov 18 15:24:55 CM: I'll see how much it saves for me Nov 18 16:22:59 is there a way to blend out the buttons in the browser? a lot of space is being wasted by that.. Nov 18 16:23:21 edistar: I think there's a bug filed against that Nov 18 16:23:30 CM: ok Nov 18 16:23:44 Default, only 52% or so of the browser window is showing the actual page ;) Nov 18 16:23:49 hehe Nov 18 16:25:11 I really like safari in the iphone, it's the only thing I like about it Nov 18 16:25:35 scrollbars will disappear I hope? Nov 18 16:25:49 yes Nov 18 16:25:59 good Nov 18 16:26:15 I thought kinetic scrolling was implemented systemwide already.. Nov 18 16:26:23 it was Nov 18 16:26:33 But it crashed webkit Nov 18 16:26:52 CM: how that? Nov 18 16:26:54 any chance we'll be able to have a module to encrypt the comunication ? Nov 18 16:26:55 I think koen re-enabled it a week ago or so, but it's not in the locked src-rev in openmoko Nov 18 16:27:17 So finger scrolling should be implemented already Nov 18 16:27:20 CM: how could it crash webkit? Nov 18 16:27:34 edistar: I have no clue, I only see rumours here and in #oe ;) Nov 18 16:27:42 ok :) Nov 18 16:27:47 CM: thanks anyway for all the info Nov 18 16:27:48 Btw, if you try to surf to wikipedia.org, the browser crashes too Nov 18 16:27:55 Annoys the hell out of me Nov 18 16:28:08 CM: is that a problem of webkit or of the OM implementation? Nov 18 16:28:32 I'd say either the version of webkit used in openmoko-browser2 or the browser itself Nov 18 16:28:46 ok Nov 18 16:29:33 * CM is away: making dinner Nov 18 16:30:23 what other repo's are good as well as scaredy's? Nov 18 16:30:29 you guys sharing repo's? Nov 18 16:34:00 still ... I wonder if there will be a way to encrypt the conversation .. Nov 18 16:35:12 BBishop: you mean scramble/descramble the audio? Nov 18 16:35:33 BBishop: it's non-trivial Nov 18 16:35:39 yeah .. but I'd like to be able to really encrypt the conversation :) Nov 18 16:35:42 BBishop: you can't do it nicely Nov 18 16:35:54 SpeedEvil, well .. doesn't matter .. I just want to know if I'll be able to make a module or something Nov 18 16:35:58 BBishop: in that the GSM chipset only has audio in/out Nov 18 16:36:05 so you have to do it in data mode Nov 18 16:36:12 yeah Nov 18 16:36:26 And many providers don't support data mode, and charge more. Nov 18 16:36:41 seriously, does somebody not support it? Nov 18 16:36:43 there are ways to encrypt analog audio Nov 18 16:36:45 Or you do it with VOIP, and eat a 2s or so delay - which workd fine for PTT Nov 18 16:36:49 charge more, yes... Nov 18 16:36:49 not really balrog. Nov 18 16:37:13 well .. let's just say I'd like to change the pitch or amplitude of the voice Nov 18 16:37:28 balrog-kun: the fundamental problem is that GSM is designed to throw away non-speech-like signals, and on errors to make something that doesn't sound much different to the input voice Nov 18 16:37:39 I'd like to be able to work with the imput voice data flow Nov 18 16:37:39 with non-voce signals, that' reaqlly brittle Nov 18 16:37:46 BBishop: oh you can do that Nov 18 16:38:00 stuff like pitch alteration on calls from designated numbers Nov 18 16:38:11 for example if you find someone has an annoyingly high voice Nov 18 16:38:17 (in principle, SMOP) Nov 18 16:38:24 no no .. I meant modifyin' my voice Nov 18 16:38:25 SpeedEvil: they sometimes make the signal lose some quality after decoding, but you can encode audio so that it remains a speech-like audio signal but is incomprehensible with no key to decode Nov 18 16:38:27 to any number Nov 18 16:38:33 BBishop: yhes - that way too Nov 18 16:38:41 brb Nov 18 16:38:43 like balrog_ is sayin' Nov 18 16:38:55 balrog-kun: yes - that's messy - but 1300 bits/s over GSM has been done Nov 18 16:39:07 BBishop: or that - yes, you can vary the voice Nov 18 16:39:23 balrog-kun: that way you do VOIPOV Nov 18 16:39:36 and strong encryption/one-time-pad in the middle Nov 18 16:40:05 kewl .. I can barely wait for 02 then :) Nov 18 16:40:39 All of this is in principle of course - nobodies written the software Nov 18 16:40:50 I'lll try :) Nov 18 16:41:44 There is a paper online somewhere about the data encoding over GSM voice links. Nov 18 16:41:46 realistically, encrypted calls with gsm-data without a huge effort Nov 18 16:41:53 mjr: yeah Nov 18 16:42:00 mjr: or VOIP with push-talk Nov 18 16:42:28 I do wonder if there are any speech recognitions that we can get on the neo. Nov 18 16:42:36 even speaker tuned Nov 18 16:42:37 as long as I can get the voice somehow .. and apply an algorithm to it Nov 18 16:42:43 same thing to the other end .. Nov 18 16:42:48 BBishop: you can't Nov 18 16:43:00 damn Nov 18 16:43:05 BBishop: basically you can do analog transforms that make it sound voice like Nov 18 16:43:22 BBishop: but the GSM voice channel is _not_ a transparent analog channel Nov 18 16:43:37 dang Nov 18 16:43:38 It's very lossy codec designed to throw non-voice components away Nov 18 16:44:10 you can send data over it at reasonable rates only by basically building a synthetic vocal tract, and exciting it with data - not speech Nov 18 16:44:19 you have to use an algorith that doesn't just encode data into a voice-like signal, Nov 18 16:44:27 so that each 'syllable' carries 20-40 bits Nov 18 16:44:32 it needs to take advantage of the fact that input signal is also voice-like Nov 18 16:44:51 .. so it'll be a pain in the ass .. Nov 18 16:45:40 balrog-kun: it can't realistically do that - otherwise you compromise the encryption. It's got to be a general purpose voice-data codec, that runs over the data-voice codec that you connect to the GSM modem Nov 18 16:46:07 BBishop: basically, yes. You can get toy 'voice changer' functionality out of it easily - but no actual encryption without much effort Nov 18 16:46:38 SpeedEvil: you can, i just don't remember the names of the algorithms because i wasn't interested in that, but generally "encoding" doesn't mean digital Nov 18 16:46:46 can't I transform it into a hissing thing .. like the old dial-up modems did Nov 18 16:46:51 ? :) that would be cool Nov 18 16:46:53 balrog-kun: in this case unfortunately it does Nov 18 16:47:20 balrog-kun: because the encoding schemes that you can do result in 'data' that the GSM codec discards as it's not voice like Nov 18 16:47:26 the hissing has very information content which codec will discard Nov 18 16:47:35 SpeedEvil: damn your nimble fingers Nov 18 16:47:40 BBishop: just invent your own language Nov 18 16:47:53 we need a navajo gsm codec ;-) Nov 18 16:47:56 Try playing music or whatever over GSM and see how it copes. Nov 18 16:48:06 Simple single note stuff just about gets through OK Nov 18 16:48:12 Or a vocal Nov 18 16:48:19 mix the two, and quality plummets Nov 18 16:48:32 pulseaudio needs a good sample-rate convertor. Nov 18 16:48:33 SpeedEvil: there are algorithms that make the signal sound speech-like, but i can't remember the algorithms so i have no good argument to back this up Nov 18 16:48:36 add polyphonic stuff and it just goes to crap Nov 18 16:49:07 damn .. Nov 18 16:49:43 balrog-kun: yes - they are generally basically half a voice synth - just synthesising the vocal tract bit that the GSM codec tries to derive the state of. Nov 18 16:51:09 balrog-kun: you basically don't want a signal that just 'sounds' voice-like though. You want a signal that the GSM codec can efficiently and accurately extract all the information about the synthetic vocal tract that it's designed to. Nov 18 16:51:27 This is annoyingly hard, and I'm not aware of published code. Nov 18 16:51:36 SpeedEvil: well, true Nov 18 16:52:36 i do think it should be doable though, especially when the GSM algorithm is generally known and you can experiment with it and reverse the equations etc Nov 18 16:53:04 Reversing the equations doesn't work Nov 18 16:53:12 At least the simple solution. Nov 18 16:53:35 Consider for a moment if you have absolutely perfect knowledge of the input state of the GSM codec and what it's transmitting Nov 18 16:53:56 yup Nov 18 16:54:14 Your task is to derive from teh output of the analog GSM modem on teh other side what the input was Nov 18 16:54:26 However, there are errors in the datastream Nov 18 16:54:55 it's a lossy channel so you can't basically Nov 18 16:55:04 which means that you've basically got to run 1024 cases for single-bit errors per frame Nov 18 16:55:13 and 1 million for 2-error cases Nov 18 16:55:30 and it rapidly gets out of hand - the GSM function is basically functioning as a (bad) hash Nov 18 16:55:46 If you don't feed it very speech-like data Nov 18 16:56:18 but you're not transmitting digital data, you're transmitting voice Nov 18 16:56:33 you have to find an encoding in which the errors don't harm the signal much Nov 18 16:57:42 the contraint is that the encoded signal contains only frequencies in some range and the resolution is not too high Nov 18 16:57:48 constraint Nov 18 16:58:14 balrog_: /j hackndev? :P Nov 18 16:58:24 why? Nov 18 16:58:34 c'mon :) Nov 18 16:59:13 * balrog_ doesn't think "c'mon" is a good argument Nov 18 16:59:32 Mhm Nov 18 17:00:59 balrog-kun: it's not that simple Nov 18 17:01:16 balrog-kun: the GSM codec attempts to work out the state of the vocal tract of a human being Nov 18 17:01:22 this is what it's optimised to do. Nov 18 17:01:53 As you depart from what it's expecting to model, signal/noise (in the binary sense) drops Nov 18 17:02:18 In that the amount of data you can pass through the channel drops as you depart from what the GSM codec expects to be modeling. Nov 18 17:02:34 SpeedEvil: well, yes, the constraints are slightly more complex than that, but still you need to explit the fact that the input signal and the part of the input signal that is relevant, fits with this contraints Nov 18 17:03:17 By input signal, are you meaning the voice input to the encryptor - before it's sent over the GSM channel? Nov 18 17:03:26 SpeedEvil: right Nov 18 17:04:36 Basically you can't exploit the input signals charachteristics to send it encrypted over the channel, as it then leaks information about that input voice signal Nov 18 17:05:35 SpeedEvil: only if you do it poorly Nov 18 17:06:11 there's part of information that is relevant in the signal and there is part of information that the GSM channel passes through Nov 18 17:06:19 in normal scenario this is the same subset Nov 18 17:06:45 you need to enocde the signal so that the information stays in this subset Nov 18 17:07:24 oops - afk Nov 18 17:07:45 laters Nov 18 17:10:18 * BBishop flips Nov 18 17:18:04 BBishop: you see, it will be a bit of work ;) Nov 18 17:18:21 yeah .. but it can be done ! Nov 18 17:18:22 :) Nov 18 17:18:34 BBishop: I would settle for VoIP and encryption ;) Nov 18 17:18:45 BBishop: or implement it on a 5 ghz quad core ;) Nov 18 17:19:06 edistar: GTA0120 Nov 18 17:19:18 SpeedEvil: I am looking forward to it ;) Nov 18 17:19:34 wouldn't using voip mean using the internet connection ? Nov 18 17:19:38 SpeedEvil: maybe it will be GTB0120 by then ;) Nov 18 17:19:40 BBishop: yes Nov 18 17:19:42 BBishop: data Nov 18 17:20:17 edistar: GTB would mean changing the case Nov 18 17:20:40 SpeedEvil: oh, I am not aware of how they name Nov 18 17:20:56 using the internet is not an option :) Nov 18 17:21:13 well .. is there a way then to encode the voice between 2 moko devices ? Nov 18 17:21:23 BBishop: via gsm? Nov 18 17:21:25 yeah Nov 18 17:21:35 ...and how do you tell the device at the other end is a moko? Nov 18 17:21:38 BBishop: didn't you read? '_ Nov 18 17:21:49 cjb_ie, that's not the problem .. Nov 18 17:21:55 I don't care about other phones :) Nov 18 17:22:02 BBishop: in principle yes. In practice, it's going to be _serious_ amounts of coding Nov 18 17:22:19 edistar, I kinda' did ... that's why I was asking .. damn Nov 18 17:22:29 BBishop: no, doesn't work Nov 18 17:22:36 the gsm chipset filters a lot Nov 18 17:22:44 and you have no influence on the gsm firmware Nov 18 17:22:45 GRRRR .. damn filters Nov 18 17:22:54 yeah .. closed source .. Nov 18 17:23:00 why is it closed source anyway ? Nov 18 17:23:07 univac: i can't answer on msg because it asks for a registered nick Nov 18 17:23:26 BBishop: So you don't fiddle with the chipset and do malicious things Nov 18 17:23:47 balrog_: ah Nov 18 17:23:49 BBishop: it's a law Nov 18 17:23:55 =)) .. damn Nov 18 17:23:55 balrog_: jabber? :) Nov 18 17:29:40 wtf: mount: special device /dev/mmcblk0p1 does not exist Nov 18 17:29:45 is this busted in the latest image? Nov 18 17:29:50 (just sick of fixing it..) Nov 18 17:31:08 torpor: looks like a memory card.. Nov 18 17:32:30 * mwester is having a bug-battle with S.C. (http://bugzilla.openmoko.org/cgi-bin/bugzilla/show_bug.cgi?id=1003) Nov 18 17:34:54 * BBishop will h4x Th3 GsM FiRmWarE >:) Nov 18 17:35:30 mwester: yeah, the folks assigned to gsmd don't like to consider input from people :( Nov 18 17:35:47 * balrog_ is having a battle at http://bugzilla.openmoko.org/cgi-bin/bugzilla/show_bug.cgi?id=894 Nov 18 17:36:51 mwester: wel.. looking good. sending at-commands without gmsd is a very bad idea and generally a layering violation Nov 18 17:38:20 it's a work around for a layering violation in GTA01 design Nov 18 17:38:51 the AT command is in this case the only way to switch power to the gsm, and power is not manage by gsmd Nov 18 17:39:01 *managed Nov 18 17:39:54 balrog_ true. but even with a real load switch powering the firmware doen before is a good idea (save states in flash, whatever). Nov 18 17:40:33 since by design gsmd is the only thing that should ever open that serial and run over the whole time, its right to send the power-off via gsmd instead of weird shell-hacks Nov 18 17:40:45 roh: yeah, perhaps, but it should also be done on echo 0 > /sys/.../power_on and on kernel shutdown Nov 18 17:41:46 balrog_ how? sending serial chars to the modem without knowting the state of the interpreter only makes it worse. Nov 18 17:42:09 sending them from kernelspace would be even more ugly than a shellscript Nov 18 17:42:38 roh: you don't know the state of the interpreter at the beginning of executing the echo 0, but the end result should always be the same Nov 18 17:42:48 balrog_ wrong. Nov 18 17:42:52 the modem should be completely down after the echo 0 > ... command finishes executing Nov 18 17:43:13 you do not even know if the modem is in powersave mode, or not, or even in the midst of an at-commandline still in the buffer Nov 18 17:43:51 roh: right, but that doesn't change the fact that it shold be down when the command finishes executing Nov 18 17:44:25 balrog_ the echo 0 should never be neccessary. thats just not a job for a shellscript. also switching the serial between console and modem should be seperated from the power-control in my eyes. but we will see. Nov 18 17:44:46 atleast we will never have that hassle with gta02. real load switch and dedicated serial Nov 18 17:45:20 roh: well, i'm not saying anything about shell scripting, but the kernel interface through /sys should work nevertheless Nov 18 17:45:22 btw. i cannot see anything bad in having the option to powercycle the app-cpu while having gsm up and working. Nov 18 17:45:43 if 0 is written to the node names "power_on" then the device should be down after that Nov 18 17:45:47 *named Nov 18 17:46:15 balrog_ yes. there are bugs (the crash) which needs to be fixed properly. 1 and 0 only toggle a line. it is not a STATE they toggle bug a signal it sends. Nov 18 17:46:27 so power_on IS labeled correctly. Nov 18 17:46:37 it sends a power_on signal. no power_off Nov 18 17:46:58 roh: hmm, okay Nov 18 17:47:44 i still think that power should be managed by the kernel though, not userspace, so there maybe should be a different node that performs a complete gsm shutdown Nov 18 17:47:52 i know thats all ugly. but i think we need to not mess it up even more. Nov 18 17:48:10 there is fundamentally a huge blob of crap that needs to manage power Nov 18 17:48:21 the operation should look as a single operation to the userspace, the details of it should be handled in kernel Nov 18 17:48:24 everything from suspending between X events, to ... Nov 18 17:48:32 the kernel cannot manage power in that case. gsm is a subsystem outside its responsibility. the kernel does hardware abstraction and management of everything thats not self-sufficient Nov 18 17:48:49 balrog-kun: you can't - without putting a _huge_ bunch of policy in the kernel. Possibly X too. Nov 18 17:49:20 roh: if you consider AT@POFF a part of hardware interface, rather than an AT command, you can Nov 18 17:49:22 the gsm is self sufficient. it does not depend on the kernel to keep working. config the mixer right and make a call. and you can kill your kernel, the call will stillbe on Nov 18 17:49:32 balrog-kun: consider that you want to suspend between touches if you can without turning off any other functionality. But playing an mp3, or doing voicemail, or ... may preclude this Nov 18 17:50:12 balrog_ i consider sending a 0-1 trigger to power_on the go to on signal and sending at@poff the off signal. abstract speaking. Nov 18 17:50:43 its like asymetric routing, in that case one part of the interface is a at-command and one a file in sysfs, exposed to userspace Nov 18 17:50:58 SpeedEvil: the part "other functionality" is not something kernel is conerned anymore, so this case is not a purely "power management" task Nov 18 17:51:03 roh: this is all theoretical, "in the pure world" stuff. Nov 18 17:51:05 Incidentally. Could one run gprs through the current gpsd in passthrough mode, while having it handle incoming calls? Nov 18 17:51:05 roh: agreed Nov 18 17:51:19 roh: I wish to solve a real-world problem: the GSM modem remains powered up when the phone is off. Nov 18 17:51:23 wrapping it in kernelspace just makes it more ugly. using gsmd to bond it together is the right call since there we also have hardware abstraction already (the plugins) Nov 18 17:51:43 roh: we CANNOT rely on gsmd because a) it is often shutdown by the user, and b) it is unreliable. Nov 18 17:51:43 mwester: you have compared how other phones do this? Nov 18 17:51:55 mwester we need to rely on gsmd. Nov 18 17:51:56 roh: I don't need to know how other phones do it; other phones are closed. Nov 18 17:52:04 a) is false and should never happen. Nov 18 17:52:06 roh: or are you just disagreeing with me becuase it's me. Nov 18 17:52:07 ? Nov 18 17:52:14 mjr: you mean gsmd? no Nov 18 17:52:17 a) is true, and often happens rith now Nov 18 17:52:21 b) is beeing fixed partially already. i have not seen it crash very recently Nov 18 17:52:22 indeed, gsmd should be relied on, and b) should be fixed to be worthy of that... Nov 18 17:52:24 mjr: not current gsmd at leat :) Nov 18 17:52:27 a) is a bug. Nov 18 17:52:50 roh: a) is a bug. So is the wiring of the gsm modem. We can fix neither at the moment. Nov 18 17:53:13 mwester no i am not disagreeing with you because of you. on the contrary. i highly regard your comments. but this is about not messing the messed up architecture up even more by adding hw abstraction where it should not be Nov 18 17:53:18 Some day, maybe, gsmd will be reliable enough that your argument will hold water. Right now, hacks are required, becasue the modem wiring is a hack, and gsmd is not ready yet. Nov 18 17:53:46 mwester no. shutting down gsmd is a bug. we need that ts07.10 and its gprs-line discipline. Nov 18 17:54:02 Well, I'll just keep re-opening the bug until someone fixes it, and since it's open source, I'll keep an alternate firmware about if necessary. Nov 18 17:54:18 the wiring will not change so fast, atleast gta02 will not have anything besides power-on triggers and a loadswitch for safety. Nov 18 17:54:32 and its own serial for more ease of development Nov 18 17:54:37 gta02 is the solution to all our problems, but I don't have one right now. Nov 18 17:55:06 So I work with what I have, and work around unreliable components such as gsmd as best as I can, and solve immediate problems. Call it a hack if you will, but I fix things. Nov 18 17:55:21 roh: any rumors of an 850Mhz run of a triband GTA02 in parallel with the first run? Nov 18 17:55:46 mwester: why is #1003 assignet to gsmd component? Nov 18 17:56:04 from my perspective unifying the power-on/off 'switch' in gsmd is the right call from sean_c Nov 18 17:56:18 SpeedEvil: i think the last we heard was that FCC approval would take significantly longer than CE Nov 18 17:56:36 Is it available to play in GTA 3 on Neo1973? :) Nov 18 17:56:36 so euromoko would be a month or two ahead of amerimoko Nov 18 17:56:40 imho the kernel should expose an identical interface for turning the modem off on all phones Nov 18 17:56:55 as i said, abstraction for hw is already done there and not only in the kernel, and it should run all the time. just gprs/ppp needs 'fixing' and ts07.10 mux/demux Nov 18 17:57:37 balrog_: It seemed reasonable at the time, since it involved the GSM modem. But in retrospect, it should have been against the hardware as a whole, since the impact of the problem is that the device sucks the battery flat when it is supposed to be powered off. Nov 18 17:57:53 SpeedEvil well.. i can only say that we now can predict that its doable to build gta devices with exactly this layput, no gerber changes for triband 850/1800/1900 Nov 18 17:58:04 mwester: it can be reassigned Nov 18 17:58:13 roh: so it's just a matter of swapping a filter or something on the production run? Nov 18 17:58:15 hooray for bom-only changes Nov 18 17:58:19 hxd8 proved that (uses the same gsm layout) Nov 18 17:58:22 :) Nov 18 17:58:29 balrog_: based on roh's reaction, I'm not sure it would change anything to reassign it. Nov 18 17:58:34 SpeedEvil different filter, different calibration Nov 18 17:58:39 okay Nov 18 17:58:43 Great. Nov 18 17:58:52 SpeedEvil do not know about different firmware, but i think thats the least of problems in that case Nov 18 17:59:11 we know it works, we know because we have fcc for hxd8 for 850/1800/1900 Nov 18 17:59:19 So it's build one, submit to FCC, and wait? Nov 18 18:00:26 I remain ready to battle this issue of shutdown; when someone proves that gsmd is worthy of being given sole rights to power-off the GSM modem then I'll shut up. Until then, we cannot entrust something as unreliable and often-changing as gsmd to do this vital task. That's my opinion, and if necessary I'll maintain an alternate image that won't flatten the battery - but that would be a shame. Nov 18 18:00:30 yes and no. gta02 needs to be done first. then its a management decicion what to do. Nov 18 18:01:04 mwester: there is no reason you can't put echo AT@poff>/dev/foo into the shutdown script Nov 18 18:01:09 mwester adding more hacks will not make it more stable. only fixing bugs. Nov 18 18:01:37 roh: will OM in a year still work on GTA-01 or will there be separate OM version for each decive? Nov 18 18:01:46 s/decive/device Nov 18 18:01:49 SpeedEvil there is. you do not know if the buffers are empty. you do not know if the firmware is in sleep mode and needs to be awakened first, etc. only gsmd knows that. Nov 18 18:02:06 * balrog_ notices how insignificant this issue would be on a normal phone which takes three hours to charge and then is constantly kept on for four days Nov 18 18:02:09 roh: yes - I mean as a backup case. Nov 18 18:02:23 edistar currently as much as possible is done portable, so i currently see no reason for differenciation there Nov 18 18:02:27 roh: the reset of the modem in the script that I put in the bug report fixes all the sleep issues and the buffering is taken care of by the kernel and the script. Nov 18 18:02:41 roh: what you are describing is a red herring. Nov 18 18:02:42 edistar we do not plan dropping all support for gta01 as soon as gta02 comes out. Nov 18 18:02:52 roh: good, just asking.. Nov 18 18:03:28 SpeedEvil still-.. when i buy this phone after its completed the last thing i want to see is any at-command outside of gsmd or gprs/ppp stuff. Nov 18 18:04:02 roh: I notice that the GSM chipset draws 30mA or so from the battery. About 4* what my cheap nokia quadband uses. Is this just a less optimised for power chipset than the nokia ones? Nov 18 18:04:06 we need _less_ scripts which fork less often. this device is much too slow already. Nov 18 18:04:14 roh: agreed. So take it out ont he gta02 images, if that one is wired correctly. Nov 18 18:04:16 roh: I quite agree for a final phone. Nov 18 18:04:17 SpeedEvil which firmware do you use? Nov 18 18:04:36 roh: I wasn't aware there was other firmware for the GSM modem than the shipped one Nov 18 18:04:52 I mean with modem on and phone shut down. (including amp powered off) Nov 18 18:04:53 mwester you cannot wire it with a real off switch. you can only disconnect if from mains. this chipset does not have any hardware-power-off caps. Nov 18 18:05:56 If you wire it on the phone side of the PM chip, then it should power off when the PM chip powers off the phone hardware. Nov 18 18:06:36 balrog_: I checked out the bug you highlighted -- I completely agree with you on all points. Nov 18 18:07:10 mwester: :) Nov 18 18:07:17 well.. that only messes up the design even more. we have a switch then. but i think we should shut it down and switch power off afterwards. so everything i said still stands. Nov 18 18:07:30 balrog_: which is especially interesting considering how the current discussion on gsmd and power-off is dependent upon bugs being fixed in gsmd.... Nov 18 18:07:39 its just a safety net, not the solution itself Nov 18 18:08:14 Agreed. gsmd *SHOULD* power off the modem. The issue is not about that, the issue is about what happens if gsmd is not able to do so. Nov 18 18:08:34 mwester then the modem stays on. which is a bug. Nov 18 18:08:48 have you _ever_ taken a look at real world phones? Nov 18 18:09:14 every cellphone has the same problem in case firmware crashes on powerdown or shortly before. Nov 18 18:09:22 The use has every right to expect that if the phone has powered off, that the GSM modem is off as well. We need to ensure that this happens, and gsmd is not able to do so reliably. Hence we need something to do this. I provided same. OM can remove that as soon as gsmd grows up. Nov 18 18:09:27 so making it not crash is priority. Nov 18 18:09:37 Yes, but that is not happening, roh Nov 18 18:10:04 that IS happening. finally there is a team working on gsmd and you still nag Nov 18 18:10:06 Not for lack of the community effort, I must add. Nov 18 18:10:29 I'm "nagging", roh? I have a perfectly valid issue, and you call that nagging? Nov 18 18:10:51 roh, you are incredibly rude. Nov 18 18:11:05 the user has no right to expect everything to work properly after a crash. not even power_off. same rules like for all devices are valid: if it crashes and you want a defined state, remove power, readd power, get state you want, be happy. Nov 18 18:11:16 making a program not crash is quite trivial compared to making it work correctly Nov 18 18:11:24 what the user has a right to is not to have crashes. Nov 18 18:11:34 but with the current team that controls the svn it may take a couple of years Nov 18 18:11:35 Thank you for your opinion, I shall cease wasting my time arguing this with you. I can see that you have made up your mind, and will resort to insults and rudeness when logic fails your arguments. Nov 18 18:12:24 * mwester ceases to "nag", lest it bother roh. Nov 18 18:14:44 mwester i am sorry if you interpret me as rude today, that wasnt my intention. i am just stating the facts, that none of us can fix such problems since it would involve builing own gsm chipsets in the end Nov 18 18:15:06 so we have to work with what we have and make it the best and cleanest we can. Nov 18 18:15:36 regular user behaviour is 'remove battery' in case of crash. which is ok and good. also solves all problems. Nov 18 18:15:52 if the gsmd crashes we need to find out why. Nov 18 18:16:29 roh: regular linux behaviour is restart the module/daemon.. ;) Nov 18 18:17:40 regular linux daemons don't crash on me Nov 18 18:17:57 balrog_: they do it the firmware in the gsm chip does Nov 18 18:18:02 edistar bad code does. Nov 18 18:18:11 why does the firmware in the gsm crashing bother linux? Nov 18 18:18:15 to be honest gsmd also doesn't crash on me recently, although it leeks tons of memory and sometimes talks to the modem nonsense Nov 18 18:18:20 roh: ok, that way too ;) Nov 18 18:18:33 is gsm so thorny? Nov 18 18:18:35 torpor: it doesn't sry, my mistake Nov 18 18:18:36 torpor we are talking about gsmd, not the firmware crashing Nov 18 18:18:53 gsmd is crappy? i haven't noticed. Nov 18 18:19:17 torpor: yes it is :) Nov 18 18:22:15 ermm .. is there an alternative for buildhost.automated.it for stuff, or should i just get over it? Nov 18 18:22:21 mwester/balrog: i hope you now know what the way i imagine to solve that mess would be right and that our engineers like sean_c are on it. Nov 18 18:25:19 torpor: scaredycat. Nov 18 18:25:53 edistar: maybe its time i set up my own feed then. ;) Nov 18 18:26:06 roh: this somehow doesn't raise my hopes :) Nov 18 18:26:32 hopes on proper working gsmd you mean? Nov 18 18:26:50 roh: yes, on anything good coming out of the official gsmd svn Nov 18 18:27:39 you know sean_c is our gsm firmware guy? Nov 18 18:27:52 torpor: sounds good Nov 18 18:28:06 i really hope some other distro forks it and starts maintaining like a normal opensource project and then maybe the fork is accepted as official like it was with gcc Nov 18 18:28:18 torpor: where do you get the webspace/server from? Nov 18 18:28:58 Could gsmd have a "--cleanup" mode where it would reset the modem to a known state, power it off, then exit cleanly? This could be called from shutdown scripts to make sure things were off even gsmd had crashed before. Nov 18 18:29:12 what is is you want to see? adding more hacks? Nov 18 18:29:48 mmontour: that's a bit of a weird function for a daemon program, don't you think? Nov 18 18:30:51 gpsd has similar Nov 18 18:31:01 I don't think it's too big of a stretch, if gsmd in its normal daemon role is responsible for all control of the GSM hardware Nov 18 18:31:24 roh: no, perhaps cleaning up whatever mess the recent commits by tick introduced (or perhaps revertin gall of it) and applying in order incorporating the comments from community like "here's, there is a leak" or "this approach is completely" wrong which they routinely ignore Nov 18 18:31:36 BTW, my $0.02 is that gsmd should have full control of the GSM modem, including power and serial-port 'stty' stuff. Nov 18 18:32:08 * mwester notes that some person(s) here, if stranded at the side of the road due to a broken fuel line on the automobile, rather than fix the fuel line with whatever was at hand, they would sit there and wait until the auto maker redesigned and re-engineered the automobile so the fuel line wouldn't break anymore. Nov 18 18:32:21 and perhaps assigning someone who knows how to code C to a project that is in C Nov 18 18:32:36 can anyone tell me how it works, that the windows in openmoko/matchbox become smaller if the vkbd is opened? is this a special matchbox feature, or is it a NetWM-function? Nov 18 18:34:48 balrog_ i cannot comment on his code right now. have not read it yet. too much other stuff happening. but i agree that gsmd need to work properly, not crash, not leak for being 'fixed' Nov 18 18:36:35 roh: there are constantly cases where a patch is submitted by one of the internal openmoko hackers and people on the list comment pointing out obvious errors or even supplying a corrected patch, and the original patch still being committed with no comentary Nov 18 18:37:16 mwester: balrogg: but be assurd that i take both of your concern very seriusly and share some of them. Nov 18 18:37:38 (while code that has been tested and reviewed by community is dropped with no discussion also) Nov 18 18:37:52 i think one important point on mwester's side is that the current issues have hindered the community Nov 18 18:38:28 a bug should be worked around if it stops people from contributing Nov 18 18:39:15 and when a bug is known, what bad is a simple workaround until the bug eventually gets fixed? Nov 18 18:40:25 guys, if you are so pissed at gsmd's current state just fork the damn thing, build your own images and be done with it Nov 18 18:41:18 CoreDump: the issue with forks is that it's good when the various hackers use the same repository, a central one, that's what cvs or svn are invented for Nov 18 18:41:54 huh? since when do true forks use identical repos? Nov 18 18:42:23 i don't want to build an openmoko fork Nov 18 18:42:50 CoreDump: i'm not saying identical repos, but if various people work on a project they need means for cooperation, cvs was invented for that Nov 18 18:43:20 I don't get it. Then use your own repo for your fork -> problem solved Nov 18 18:44:22 I "nag" here in hope of inspiring action, rather than a fork. A fork would be a last resort IMO Nov 18 18:44:23 CoreDump: what part you don't get? i use my repo for my fork, and person X also writes code in the same project, if they use a seaprate repo they will waste time fixing the same bugs Nov 18 18:45:31 cant we just put gsmd in a git or hg repository? Nov 18 18:45:43 they are more made for forking things and put together later on Nov 18 18:45:44 balrog-kun: ....that is the point of a fork, isn't it? If your stuff is better than the original, people are going to use your fork (and repo). If not...well one less fork Nov 18 18:46:13 i think if someone would just pop up with a completely rewritten and working gsmd, which has all neccessary features and abstractions as well as the right license, i would be the last to stand in the war of throwing away the gsmd we have Nov 18 18:46:52 roh, i don't think there's a public spec for gsmd, or is there? Nov 18 18:47:43 CoreDump: gsmd still needs a couple of months work to be usable, it won't be done by one person, and indeed several people are happy to work on it, but they have no means to cooperate Nov 18 18:48:23 set up a repo, contact these persons and try to get them to cooperate Nov 18 18:48:29 CoreDump: they need a common repository with commit access to it, and if a big enough projects starts such a repository then it's very likely that people will switch to it, but not if an individual sets it up Nov 18 18:48:37 tuukkah: not really. but from what i hear the people who care know about everything they need already. Nov 18 18:49:29 roh, i don't know about everything :-( Nov 18 18:49:48 roh you know thats not true. om has the gsm proprietary docs Nov 18 18:50:30 with the official svn it seems there is no problem because it looks to me like no persona that understands gsmd code has commit access to it so it won't win with any 3rd party fork, but the issue is that the 3rd party forks shouldn't compete against each other Nov 18 18:50:35 mwester yes, but you know that you should not need them for 99% of gsmd Nov 18 18:50:38 *person Nov 18 18:51:02 only for vendor specific stuff in some plugins (and then its true for htc and othe rplugins as well Nov 18 18:51:41 1% is the magic that is necessary Nov 18 18:52:12 mwester-n jap. and thats already in the gsmd we have. or are you missing any command still? Nov 18 18:52:40 mwester-n, based on the differing views, wouldn't you also consider it a problem to come up with something considered "proper"? Nov 18 18:52:51 i don't know what i don't know. Nov 18 18:52:51 does GSMD handle SMS/ Nov 18 18:53:06 matthew_: yes Nov 18 18:53:15 * balrog_ agrees that the Calypso specs are not needed for further gsmd development Nov 18 18:53:28 i need to go out and find some food now, will be back later. i do not want to kill this conversation, if you wish we can continue later. Nov 18 18:53:45 you can build a 100% functional phone with what is currently known publically Nov 18 18:55:56 is special functionality like *123# for getting my remaining minutes already possible with the gsmd? Nov 18 18:56:09 gamin: yes Nov 18 18:57:36 balrog_: using a centralised vcs really sounds like the wrong approach here Nov 18 18:58:02 roh: imo, not sucking power is way more important right now than a layering violation Nov 18 18:58:19 why dont we use git for gsmd, so that people cann pull from each other? Nov 18 18:58:22 gsmd fork with tricky license like 'you can't use our patches until you fire this and that person or at least forbid them from writing code' would be interesting solution. Resulting binaries won't be redistributable due to GPL, but it is possible to write script for one-step building them. Nov 18 18:58:34 borg_: folks.o-hand.com/andrew/gsmd.git/ Nov 18 18:58:44 dottedmag: eh, that doesn't sound right Nov 18 18:58:58 o_O Nov 18 18:59:03 it is in git? Nov 18 18:59:05 nobody stops you from setting up your git repo and point out interesting patches on the gsmd-devel list Nov 18 18:59:07 where is the problem then? Nov 18 18:59:08 LarstiQ: this is the crude hack for the situation when meritocracy does not apply. Nov 18 18:59:11 LarstiQ: i didn't use the word centralised, i'm saying using *one* repo Nov 18 18:59:41 balrog_: ehm, how is that _not_ centralised? Nov 18 19:00:07 dottedmag, blergh Nov 18 19:00:09 LarstiQ: well, it is, but centralised is a requirement for cooperating, isn't it? Nov 18 19:00:15 balrog_: no it isn't Nov 18 19:00:32 borg_: the problem is that nobody will base their patches on this git tree, people will still base patches on the svn and never have them applied Nov 18 19:00:58 balrog_: cooperation requires communication, if things flow in a mesh like way that is fine Nov 18 19:01:16 balrog_: but having one central repo where no one has access to is counter productive Nov 18 19:02:03 LarstiQ: the idea behind using a repo is that one person can commit with a single command, like "commit" and the other people can do a "update" and immediately start working on top of the code written by other people, without rediffing by hand Nov 18 19:02:36 LarstiQ: obviously everyone in the mesh needs commit rights Nov 18 19:02:42 balrog_: bzr have this feature, and still distributed. Nov 18 19:02:54 dottedmag: git also Nov 18 19:02:59 yep Nov 18 19:03:04 balrog_: they can to their own branches, and others can just merge their work when they want to Nov 18 19:03:07 i'm happy to give access to folks.o-hand.com/andrew/gsmd.git/ to anyone Nov 18 19:03:13 balrog_: no need to rediff by hand Nov 18 19:03:30 (who can code) Nov 18 19:03:34 balrog_: :) Nov 18 19:03:34 balrog_: heh :) Nov 18 19:04:10 LarstiQ: for this they need to know about each other's trees Nov 18 19:04:18 balrog_: right, communication Nov 18 19:05:13 and it's much easier to do if there's one tree, for example because .git allows you to have a single "origin" tree to/from which it pushes/pulls on commit/update Nov 18 19:05:22 balrog_: I do agree you're likely to get one "blessed" repo people pull from Nov 18 19:05:55 but which one that is fluctuates Nov 18 19:06:42 so lets promote this repo, it doesnt sound that complicated if there is so much frustration with the persons in charge of the openmoko inc. svn Nov 18 19:07:11 sounds much better to fork gsmd then to fork the whole project :) Nov 18 19:08:16 the nice thing about git is that it works well together with svn Nov 18 19:08:22 i mean in a bidirectional way Nov 18 19:08:54 Sorry if this is sort of noobish, but how serious are the problems with gsmd? Are they going to be sorted out by the time GTA02 becomes available? Nov 18 19:09:39 (Or does my question violate the "don't ask about GTA02?) Nov 18 19:09:43 haakeyar, depends when the time is =) Nov 18 19:10:19 gsm and power management issues don't seem to be trivial at least Nov 18 19:10:55 but we just get our act together, put the needed work in them and we'll be ready in no time :-) Nov 18 19:11:15 i've been questioning power management issues since june Nov 18 19:11:21 people didn't think them to be important at the time Nov 18 19:11:42 matthew_, must depend on people Nov 18 19:12:54 power management and gsmd go together though Nov 18 19:13:25 or any daemon, for that matter Nov 18 19:15:09 why is gsmd so problematic? Nov 18 19:15:43 ruxpin, it's made from scratch Nov 18 19:16:18 ruxpin: because it's crucial for a phone and it seems to be currently regressing instead of progressing Nov 18 19:16:24 there are lots of different opinions on how it should look like Nov 18 19:17:15 that's why i was talking of having a public spec Nov 18 19:17:46 someone might like to create one, perhaps in the lines of http://wiki.openmoko.org/wiki/Power_management_requirements Nov 18 19:18:24 then extract the current features from gsmd source code and see if they fit Nov 18 19:18:31 tuukkah: http://wiki.openmoko.org/wiki/Gsmd/document Nov 18 19:18:38 i guess one could just extend this Nov 18 19:18:57 "Gsmd document" =) Nov 18 19:20:44 indeed, that's a start Nov 18 19:23:13 although, without requirements it can't really be filled in Nov 18 19:24:09 what should happen if two apps try to place a call at the same time? how should cancelling a call work? etc. Nov 18 19:25:10 can i have an app that steals incoming calls during the times i'm sleeping? Nov 18 19:25:25 tuukkah: yes Nov 18 19:25:31 tuukkah: at least - the hardware permits it Nov 18 19:25:48 many of these requirements are already dictated by the hardware and gsm specs Nov 18 19:25:53 tuukkah: bluetooth heartrate monitor would probably work OK for a sleep-detector Nov 18 19:27:42 i don't see what hardware has to do with these Nov 18 19:29:01 for example if the hardware doesn't allow two calls from two different apps at the same time then you know that gsmd can't provide that to its client apps Nov 18 19:29:13 wrt gsmd requirements, i see the importance in making the abstractions something that meets the requirements we're going to have Nov 18 19:29:44 balrog_: yes, but what should gsmd do in such a situation? return an error, just do nothing, explode, ...? Nov 18 19:29:46 balrog_, i know the hardware doesn't allow two calls. the point is, *what* should happen? Nov 18 19:30:00 it would be nice to have a paper that describes all those things Nov 18 19:30:16 tuukkah: i'm not sure that it doesn't... Nov 18 19:30:23 and it would be much easier for people to participate then Nov 18 19:31:31 if my script (app b) places a call, will gsmd relay the info to dialer (app a) and will dialer provide a gui to cancel this call? Nov 18 19:32:07 tuukkah: If the d-bus bindings will be designed properly then yes Nov 18 19:32:15 tuukkah: yeah, many decisions should be specified ahead Nov 18 19:32:42 abraxa_, "properly" is dependent on the requirements and design we decide on Nov 18 19:33:06 and i'd be happy if LaF0rge had put up a document detailing it because he had many good ideas while he worked on the project and his ideas brought gsmd to be able to do anything at all Nov 18 19:33:38 balrog_, he still might... Nov 18 19:33:46 unfortunately he left it and the currently controlling team dosn't know what it is that they want Nov 18 19:34:50 and especially doesn't seem to be able to come up with any new idea, while gsmd still needs some new ideas in a couple of areas because it's not only bugfixes that are left to do Nov 18 19:36:15 if i've learned something about designing software it's that there is no one right way to implement product x, or command y, or module z. it all depends on the requirements of the context Nov 18 19:37:35 balrog_, to me this would seem like a good time to work on this with the community Nov 18 19:43:19 hi there Nov 18 19:43:28 * mwester-laptop reluctantly agrees that perhaps a separate community effort for gsmd is required, as it seems that O.M. is unwilling to accept community input on gsmd. :( Nov 18 19:44:03 this is why i work solo ;) Nov 18 19:44:06 why not? Nov 18 19:45:55 * mwester-laptop lacks the skills to work solo on something as complex as gsmd. Nov 18 19:46:00 so, is there someone beside of LaF0rge who is able to make some sort of gsmd specification which describes in detail how the workflow of gsmd should look like? Nov 18 19:46:45 i guess there would be a lot of people who would help implementing gsmd, if they only whould know what has to be done... Nov 18 19:46:49 this wouldn't be that much of a problem if they had a really skilled crew assigned to the project internally, even a one person crew composed of a person like one of the core leaders, but that isn't the case Nov 18 19:48:00 can't the community help somehow? Nov 18 19:48:35 edistar: yes, community can solve perhaps most of the gsmd problems if they are allowed to Nov 18 19:49:13 balrog_: and what are the problems, why aren't they allowed to? Nov 18 19:50:18 mwester-laptop: so do I, which is why i'm not working on a project that complex ;) Nov 18 19:51:09 sry, didn't catch that Nov 18 19:52:23 edistar: see list/bugzilla for details Nov 18 19:53:34 ok Nov 18 19:56:09 edistar, integrating a community into a corporate project isn't trivial Nov 18 20:26:03 it's the mirrors Nov 18 20:27:59 they make everything look bigger. Nov 18 20:37:56 not necessarily, a concave mirror makes things look smaller up to a certain point :o Nov 18 20:38:22 or convex? .. hmm Nov 18 20:38:25 * anrp thinks for a second Nov 18 20:42:40 anrp: http://de.wikipedia.org/wiki/Bild:Spiegel.jpg street mirrors are convex, they are probably so to make things bigger Nov 18 20:42:50 so a concave mirror will make things smaller Nov 18 20:42:54 i guess ;) Nov 18 20:43:06 nah, street mirrors cover a very wide area Nov 18 20:43:09 and make things smaller Nov 18 20:43:19 shaving mirrors are concave, to magnify Nov 18 20:43:22 ah ok Nov 18 20:43:31 so it is the other way around? :) Nov 18 20:43:41 yup Nov 18 20:44:15 but of course with concave mirrors everything changes when you pass the focal point Nov 18 20:44:34 or you could skip the analog step entirely and use software processing! Nov 18 20:48:18 the key to fixing gsmd problems: the mirrors Nov 18 21:01:07 hmm, is it usual that flashing image in qemu-neo1973 takes much more than two minutes? Nov 18 21:01:16 on amd64 Nov 18 21:01:28 mmp: do you get an erro? Nov 18 21:01:29 error? Nov 18 21:01:54 edistar: only timeout Nov 18 21:02:05 and 100% cpu load when doing something Nov 18 21:02:12 mmp: I get that too, but not only athlon64 I think Nov 18 21:03:23 well, I have no idea now long does it take to flash whole distribution under emulator:) Nov 18 21:03:50 timeout means error Nov 18 21:04:32 yes, but the error exactly is: U-boot failed to finish writing in 240 seconds, giving up. Nov 18 21:04:46 so it looks like just it doesn't manage it in reasonable time Nov 18 21:04:47 mmp: yes Nov 18 21:05:13 I gave it 2400 seconds now, so I hope it will be enough... Nov 18 21:05:58 mmp: tell me if that solves it... would be great, but I suspect it won't Nov 18 21:10:40 hmm, u-boot really could write flashing progress :) Nov 18 21:11:06 it does Nov 18 21:11:25 if it doesn't then something is wrong Nov 18 21:11:49 balrog-kun: hmm, even in flashing script for qemu-neo1973 ? Nov 18 21:11:54 unless you're flashing remotely through dfu-util (but then dfu-util shows progress) Nov 18 21:12:01 mmp: yes Nov 18 21:12:26 I'm using openmoko/upload.sh Nov 18 21:12:33 ech, flash.sh rather :) Nov 18 21:12:47 mmp: does u-boot print *anything*? Nov 18 21:12:50 hmm, just now I noticed: openmoko/flash.sh: line 88: 17632 Ukončené ( sleep $2; kill $pid; sleep 1; kill -KILL $pid ) Nov 18 21:13:05 yes, wait a while Nov 18 21:14:01 http://www.pastebin.ca/781405 Nov 18 21:14:26 that 'Ukoncene' stands either for Terminated or Quit, I'm not sure now :) Nov 18 21:15:04 hello Nov 18 21:15:28 guys, i have kind of a doubt Nov 18 21:15:42 mmp: is it reproducible? Nov 18 21:15:42 When do you guys think the GTA02 will be available to buy ? Nov 18 21:15:45 balrog-kun: maybe that 'kill' should not be there Nov 18 21:15:54 balrog-kun: from three attempts, three failures Nov 18 21:15:58 i'm intrested in getting a gta01, it is supposed to be 300USD Nov 18 21:16:01 mmp: i.e. everytime the output is the same? Nov 18 21:16:04 how much will the GTA02 be? Nov 18 21:16:12 alex88ita: 450$ Nov 18 21:16:13 balrog-kun: yes, it looks like Nov 18 21:16:36 mmp: it looks like qemu isn't even running Nov 18 21:16:37 how much is the iphone? Nov 18 21:16:52 don't bother asking, just bend over Nov 18 21:17:04 thanks cjb_ie Nov 18 21:17:15 balrog-kun: I wonder whether qemu wasn't killed by that command above... Nov 18 21:17:24 mmp: can you try changing "-clock unix" to "-clock dynticks" in openmoko/env ? Nov 18 21:17:27 balrog-kun: but qemu is running Nov 18 21:17:30 cjb_ie: hehe Nov 18 21:17:31 similar money, plus a contract of 18 months in the UK Nov 18 21:17:37 it takes at least 100% of CPU time now:) Nov 18 21:17:45 uhm Nov 18 21:17:58 mmp: for me everything is ok with -clock unix, and works everytime Nov 18 21:18:10 cool Nov 18 21:18:21 you can also try "-clock hpet" or "-clock rtc" if that doesn't help Nov 18 21:18:28 balrog_: qemu is compiled with gcc-3.4.6, running on amd64 Nov 18 21:18:29 alex88ita: GTA02 will be out soon, and I'm sure it will be worth the higher price to have wifi, faster cpu, gpu, more flash and accelerometers :) Nov 18 21:18:33 neo1973 would have lots of customers if it would cost 100 usd less Nov 18 21:18:42 i will brb Nov 18 21:19:02 balrog_: it did not even finish in 20minutes, so I guess time is not the issue here... Nov 18 21:19:03 cool CM Nov 18 21:19:48 alex88ita: The reason it doesn't cost 100 USD is that it doesn't come with any long contract Nov 18 21:20:12 i think i'll be getting one as soon as they arrive in europe Nov 18 21:20:23 balrog_: hmm, neo poweroff is called from some weird reason... Nov 18 21:20:24 Me too, even if I have a GTA01 ;) Nov 18 21:20:32 sell it lol :P Nov 18 21:20:39 2 days before gta02 comes out Nov 18 21:20:51 Hehe.. Na, I'll keep it around. :) Nov 18 21:21:12 eheh Nov 18 21:21:24 i think i'm getting it mainly for touch-screen and wifi Nov 18 21:21:38 i need a wifi connection while i'm at uni Nov 18 21:22:09 do you guys also follow the 3ePc project from Asus Nov 18 21:22:26 ? Nov 18 21:22:29 A bit, I follow everything ;) Nov 18 21:22:47 do u like it Nov 18 21:23:16 It seems nice. I think I'll skip the first gen though, as I'm getting one of them OLPCs ;) Nov 18 21:23:30 I think it seems really nice, but the screen looks a bit small, and I'd like a larger disk too Nov 18 21:23:33 the eep could use bt though Nov 18 21:23:42 mjr: For sure, the XO is cooler :) Nov 18 21:23:45 When do you guys think the GTA02 will be available to buy ? Nov 18 21:23:53 CVirus: Still no news Nov 18 21:24:04 CVirus: So december is starting to look unlikely.. Nov 18 21:24:09 :-( Nov 18 21:24:09 CM, yeah well, I pretty much figure the a large SD card to accompany both the XO and the eep Nov 18 21:24:11 mjr u getting a olpc? Nov 18 21:24:22 i thought they were only for third-world countries Nov 18 21:24:27 CM: when do you personally expect ? Nov 18 21:24:31 http://xogiving.org Nov 18 21:24:36 CVirus, see topic Nov 18 21:24:57 mjr: I'm asking for a personal opinion that's all :-) Nov 18 21:25:13 CVirus: From GTA01 being manufactured until the store opened was about 3 weeks, and another 10 days before I got mine Nov 18 21:25:38 mmp: power-off is called by the script when it finishes flashing to stop qemu and start flashing the next file Nov 18 21:25:40 CM: when did GTA02 started being manufactured ? Nov 18 21:25:40 Now they have the store already, but I guess they will do it the same way, ship them to the US and then distribute from there. Nov 18 21:25:50 I haven't seen any reports of P0 GTA02 yet, so there is no chance that P1 GTA02 will happen in December. Nov 18 21:26:00 CVirus: I don't think it has yet Nov 18 21:26:07 there have been allusions to GTA02 EU shipping center Nov 18 21:26:35 rwhitby: The've had the pre-series versions for a bit more than 2 weeks, but that's all I've heard Nov 18 21:26:52 neo1973 gui rocks btw Nov 18 21:26:55 mjr: Even for P0? Nov 18 21:26:57 I hope so.. Nov 18 21:26:59 CM: yeah, that's P -1, not P0. Nov 18 21:27:05 Yes Nov 18 21:27:06 balrog_: aha... well, I'm looking on it Nov 18 21:27:26 alex88ita: If you try it in qemu it's sloooow.. But works ok on the actual phone Nov 18 21:27:26 if they manage to get a gps module they'll sell lots of these toys here in europe :P Nov 18 21:27:34 CM, P0 *shrug* Nov 18 21:27:42 eheh Nov 18 21:27:47 alex88ita, it has gps Nov 18 21:27:57 already? Nov 18 21:28:03 what, of course Nov 18 21:28:09 wtf lol Nov 18 21:28:13 :) Nov 18 21:28:16 they'll sell tons of these here Nov 18 21:28:18 just no public software to drive it ... Nov 18 21:28:30 uhm Nov 18 21:28:30 yet Nov 18 21:28:30 (the proprietary gps chip that is) Nov 18 21:28:37 that's enough Nov 18 21:28:40 lol Nov 18 21:28:41 balrog_: looks like when qemu is started second time, script isn't correctly talking to uboot... Nov 18 21:28:51 i think it would be really nice if the would sell some maps which can be used with navid in their shop too Nov 18 21:28:54 CM: lawyers are involved - "yet" can mean years. Nov 18 21:28:59 s/the/they/ Nov 18 21:29:00 borg_ meant: i think it would be really nice if they would sell some maps which can be used with navid in their shop too Nov 18 21:29:31 and in the meantime, let's keep our fingers crossed for openstreetmap... ;-) Nov 18 21:29:42 eheh Nov 18 21:29:46 that's a good project also Nov 18 21:29:50 What I can't understand is that in this day of email, Sean & Co. are trying and failing to reach people by telephone, and that is the reason why things are being held up ... Nov 18 21:29:57 rwhitby: Yes, sadly.. Nov 18 21:29:59 no fingers, let's improve openstreetmap in our neighbourhoods Nov 18 21:30:11 mmp: same with dynticks? Nov 18 21:30:18 mapping party in dublin next weekend if anyone wants to come ;-) Nov 18 21:30:22 no, have not tried dynticks yet, going to try Nov 18 21:30:30 what is the feature of neo1973 u guys prefer Nov 18 21:30:37 rwhitby, a phonecall is a better kick for reluctant to answer types than e-mail... Nov 18 21:31:17 mjr: true, but if email's not working, then it's a power relationship where OpenMoko has no clout to get anything done. Nov 18 21:31:36 (i.e. they are too small fry for the vendor to even bother answering an email) Nov 18 21:31:44 it does seem they have a bit of that problem Nov 18 21:31:44 balrog-kun: still the same, no u-boot prompt, just waits for timeout Nov 18 21:31:59 mmp: weird Nov 18 21:32:01 rwhitby: Q: What do you call 200 lawyers at the bottom of Lake Michigan? A: A good start Nov 18 21:32:03 mmp: how about rtc? Nov 18 21:32:07 -clock rtc Nov 18 21:32:10 2 thousand devices is a "sample order" for these vendors. It's in the noise. Nov 18 21:32:16 balrog-kun: hmm, do you think it's race condition in qemu? Nov 18 21:32:24 mmp: no idea Nov 18 21:32:33 mmp: it never happens to me Nov 18 21:34:04 again doesn't boot the third time... Nov 18 21:34:27 I wonder whetner resulting image doesn't have u-boot corrupted somehow (don't ask me how could that be possible, just guessing:) Nov 18 21:35:07 mmp: in the piece you pasted, it's running at least one time Nov 18 21:35:38 I guess even three times -- it programs boot splash and some other things.. Nov 18 21:35:43 guys, where can i get a qemu image of openmoko Nov 18 21:35:55 ? Nov 18 21:36:05 alex88ita: MokoMakefile, "make qemu" Nov 18 21:36:54 22:16:40 < traedet> Piotten, har jag fått allt om bakfoten men visst har du illrar? Nov 18 21:37:02 gah Nov 18 21:37:20 thanks rwhitby i'll try Nov 18 21:37:27 balrog-kun: should the script run qemu-system-arm from system, or qemu-img from qemu-neo1973 ? Nov 18 21:37:27 touchpads and random middlecklicks Nov 18 21:41:14 but it still doesn't seem to work Nov 18 21:42:04 mmp: qemu-system-arm three times Nov 18 21:42:07 its downloading now :D Nov 18 21:42:07 and qemu-img once Nov 18 21:42:20 140mbz :O Nov 18 21:54:22 uhm i get an error Nov 18 21:54:40 openmoko/flash.sh: line 77: /home/alex/qemu-neo1973/arm-softmmu/qemu-system-arm: No such file or directory Nov 18 21:55:07 i'm getting all the stuff from https://svn.openmoko.org/trunk/src/host/qemu-neo1973 Nov 18 21:55:25 hi all Nov 18 21:55:29 did you run "make"? Nov 18 21:55:47 ehm Nov 18 21:55:59 i forgot Nov 18 21:56:26 i'm such a noob sometimes Nov 18 22:02:37 pfff i get a timeout error Nov 18 22:03:07 it locks up at ' Please wait, programming the NAND flash...' for 2 minutes and then it leaves Nov 18 22:04:36 "U-boot failed to finish writing in 120 seconds, giving up." Nov 18 22:06:06 any idea, guys? Nov 18 22:08:06 * balrog-kun wishes he had a computer where that happens Nov 18 22:08:40 alex88ita: get the latest flash.sh Nov 18 22:08:56 U-boot failed to resurface? :-) Nov 18 22:09:00 balrog-kun: I've been unable to reproduce that too Nov 18 22:09:13 i've got the latest flash.sh Nov 18 22:09:23 balrog-kun: uhm, looks like the third time qemu even doesn't start... Nov 18 22:09:38 pfff Nov 18 22:09:38 alex88ita: the same at me:) Nov 18 22:09:43 maybe is the sdl? Nov 18 22:09:43 balrog-kun: do you need ssh access? Nov 18 22:09:50 balrog-kun: I could provide some Nov 18 22:10:09 balrog-kun: and so far I know that the last time, qemu gets started, but doesn't boot to uboot Nov 18 22:10:41 mmp: hmm, ssh access would be cool Nov 18 22:11:00 alex88ita: it's possible, there have been problems with sdl befoe that looked like qemu problems Nov 18 22:11:16 ok let's see if i can get those libZ Nov 18 22:11:17 balrog-kun: on my way, I just need to re-configure network and create you proper environment :) Nov 18 22:11:53 cool :) Nov 18 22:12:32 what's the phonesim thing flash.sh mentions? Nov 18 22:12:53 Vegar: modem emulator with a nice gui Nov 18 22:13:01 nice Nov 18 22:13:32 lol Nov 18 22:13:43 http://unadventure.files.wordpress.com/2007/10/screen-qtopia-qvfb.png <-- the gui Nov 18 22:14:07 About the qemu problems: Nov 18 22:14:12 nice Nov 18 22:14:21 I've had the same problems too, on a Fedora 7 machine (amd athlon or whatever it's called, I'm not a processor guy), while it worked on Ubuntu 7.04 on another computer Nov 18 22:14:25 so it emulates the gsm modem? Nov 18 22:14:51 (dual processor intel something) Nov 18 22:14:54 haakeyar, i've followed a ubuntu tutorial Nov 18 22:15:01 for installing the emulator Nov 18 22:15:28 i'm on debian sid Nov 18 22:15:52 i've checked for sdl libraries and these are the latest avaible Nov 18 22:15:56 1.2..... Nov 18 22:17:52 alex88ita, yes, I don't know what the problem is, but maybe you could try from a livecd with ubuntu 7.04? Ubuntu 7.04 worked fine with me, so if it doesn't with you, it's probably hardware related. Nov 18 22:18:17 yeah sure i got a ubuntu 7.04 live cd Nov 18 22:18:22 i'll try Nov 18 22:18:33 but i will try tomorrow Nov 18 22:18:36 it's quite late now :) Nov 18 22:18:42 11.20 pm Nov 18 22:19:02 Oh, you are in the same time zone as me. :) Nov 18 22:19:37 where r u from Nov 18 22:19:41 Norway Nov 18 22:19:51 cool Nov 18 22:19:54 Rome here Nov 18 22:20:22 Ok, well, good luck and good night. :) Nov 18 22:20:43 thanks ;) Nov 18 22:20:49 see u tomorrow guys, thanks for help Nov 18 22:43:39 the qemu flashing should be fixed now Nov 18 22:54:13 mmmm never remove libgcc1 :( Nov 18 22:54:45 balrog-kun: What was the problem? Nov 18 22:55:53 Ghiottone: Ouch.. :/ Nov 18 22:56:25 Ghiottone: oops! Nov 18 22:57:09 CM: in addition to bug #937, for which i committed a workaround on friday, the limit for any image size to flash was 48 MB due to how u-boot uses memory Nov 18 22:57:25 it turns out recent *-devel-* images grew to 50 MB Nov 18 22:57:43 Ahh.. Hmm, I haven't noticed Nov 18 22:57:56 is there the way to open a serial console to the neo? Nov 18 22:58:14 Ghiottone: While it's on? Nov 18 22:58:19 yes Nov 18 22:58:24 CM: so if i had downloaded an updated the rootfs i would see the issue too perhaps Nov 18 22:58:34 thats what the debug board does Nov 18 22:58:48 because the gsm modem and the usb-console is on the same Nov 18 22:58:59 no debug board for me Nov 18 22:59:34 balrog-kun: I guess so. I haven't seen that they're that big, mine are usually around 44 mb Nov 18 23:00:42 CM: they grow every now and then, i remember when the same issue was reported when they gre from below 32 MB to over 32 MB and i increased the limit to 48 MB squeezing some things as possible Nov 18 23:01:35 actually then the symptom was completely different because u-boot was overwriting part of the rootfs in memory instead of refusing to run like it did this time Nov 18 23:01:38 I wonder if they've added python in the latest devel images or what it is that's made them grow with at least 8mb Nov 18 23:02:07 and after flashing some files on the rootfs had wrong contents and it was really hard to track it down because it seemed like the flash had errors Nov 18 23:02:27 Heh, I can imagine Nov 18 23:02:43 That happened to my uboot env.. The kernel was written over it by a bad uboot Nov 18 23:03:18 yeah, i had that too whne i flashed a debugging kernel Nov 18 23:03:48 the new limit for images is 56 MB but that's near what the rootfs partition size is on real neo so this time it shouldn't hit Qemu more than real Neo Nov 18 23:04:52 Good night, it's after midnight here.. :) Nov 18 23:05:06 night CM Nov 18 23:49:44 hi Nov 18 23:50:21 where do i find which video codecs/resolution are supported in media player? Nov 18 23:59:33 shred: ask abraxa_ Nov 19 00:00:11 shred: For now anything you can throw at gstreamer and the plugins you installed Nov 19 00:00:26 Later on, anything you can throw at mplayer Nov 19 00:00:43 MUCH later on, anything you can throw at the SMedia 3224 Nov 19 00:01:11 what resolution does the cpu support? full 640x480? Nov 19 00:01:24 of course Nov 19 00:01:47 no Nov 19 00:01:50 480*640 Nov 19 00:02:09 * SpeedEvil has stainless steel nitpicks. Nov 19 00:02:27 so mpeg4 480x640 should work with sound? Nov 19 00:02:29 Unfortunately, it's not a meaningless distinction for the current hardware Nov 19 00:02:29 you just may drop some frames Nov 19 00:02:30 * thomasg slaps SpeedEvil with a DVI-cable Nov 19 00:02:33 :) Nov 19 00:02:40 as rotation is _very_ expensive Nov 19 00:03:17 As a ballpark - the current hardware can pretty much keep up with a VCD rate video at 320*240 or so Nov 19 00:03:26 without rotation, skipping the occasional frame Nov 19 00:03:44 add rotation, and it drops a fair way. Nov 19 00:04:00 ok cool Nov 19 00:04:05 I also tried 320x240 with 18 fps at 500 bps or so and it was ok Nov 19 00:04:11 *kbps Nov 19 00:04:11 * raster cheers for software rotation - yay! Nov 19 00:05:07 Hmm. Nov 19 00:15:43 cheers? Nov 19 00:15:47 sw rotation is a necessary evil Nov 19 00:15:53 not something to be happy about ~_~ Nov 19 00:16:45 hopefully the drivers on 02 will allow hw rotation Nov 19 00:17:51 It's only needed if you assume a major use-case is to play completely random videos Nov 19 00:18:09 if you're happy wiht an automatic transcoding thingy on the desktop, then... Nov 19 00:23:19 what's so bad with software rotation? Nov 19 00:23:31 slow Nov 19 00:23:36 and a terrible memory access pattern Nov 19 00:23:51 so its hard to accelerate Nov 19 00:32:59 anrp: Software video decoding will probably be a no-go on GTA02 so we may be left with only hw-MPEG4 Nov 19 00:35:15 a no-go? Nov 19 00:35:17 why? Nov 19 00:38:18 it should be better than it is on GTA01 Nov 19 00:38:21 Cause the channel between CPU and framebuffer will have less bandwidth - I doubt anyone has numbers, though, so worst case scenario is that even 320x240x20fps will demand too much Nov 19 00:38:59 less bw? Nov 19 00:39:03 I dunno. If it's attached to the memory bus, there shouldn't be a problem Nov 19 00:39:04 that seems .. unlikely Nov 19 00:39:11 pretty sure its just mapped in the address space Nov 19 00:39:25 and btw many codecs are built on the same ideas that MPEG and if the hardware can do fourier transformations then the decoders for other codecs can use that too Nov 19 00:39:32 it can't Nov 19 00:39:34 reportedly Nov 19 00:39:40 it's not general purpose DSP Nov 19 00:39:44 wrong term Nov 19 00:39:46 it's specified codecs only Nov 19 00:39:53 (inverse) discrete cosine transform Nov 19 00:39:56 fft is something else D: Nov 19 00:39:57 apparantly Nov 19 00:40:16 which is annoying, as it can't do sound Nov 19 00:40:31 which means you've gotta have the CPU on to do sound, which is a pity Nov 19 00:40:34 i think someone wrote an integer fft for arm .. i vaguely recall looking for it Nov 19 00:40:38 *at it Nov 19 00:40:40 anrp: oops, right Nov 19 00:40:53 you'd use the fft for equalization Nov 19 00:40:56 usually Nov 19 00:41:52 and for mp3 coding/decoding afaik Nov 19 00:41:52 what'd be lovely would be a chip that could take a list of blocks off the SD card to play video/audio from, and it'd just do it, with the CPU completely off Nov 19 00:42:18 but... Nov 19 00:42:25 that would almost be a general purpose cpu Nov 19 00:42:26 :p Nov 19 00:42:55 Getting close. Nov 19 00:43:14 As I understand it all of the MP3 decoder chips available are in fact little DSPs running canned programs. Nov 19 00:43:26 optimised of course, but... Nov 19 00:43:32 i'm pretty sure there's enough power in the lower speed modes in the CPU to do sw decoding of audio Nov 19 00:43:37 so it shouldn't be too bad a hit Nov 19 00:43:41 andred_: 60Mhz Nov 19 00:43:48 or so for 128K mp3 Nov 19 00:43:56 the glamo already does SD and MPEG in one chip, it's pretty qualiied to be extended to do what SpeedEvil proposes too Nov 19 00:44:16 Umm. extension means you fab a whole new chip prolly Nov 19 00:44:21 hehe Nov 19 00:44:23 are they going to put 802.11.b on these phones? Nov 19 00:44:23 yeah.. Nov 19 00:44:26 or pay more for a flash version Nov 19 00:44:29 if it exists Nov 19 00:44:33 A substantial chunk of a million. Nov 19 00:44:42 probably. Nov 19 00:45:00 i'd think over a million Nov 19 00:45:02 More than a million if it doesn't work first time Nov 19 00:45:06 unless they're designed around custom fab rubs Nov 19 00:45:07 *runs Nov 19 00:45:52 on hte order of a million is probably around right. Nov 19 00:46:02 .1M-10M certainly covers it Nov 19 00:46:04 basically, A Lot(tm) Nov 19 00:46:08 yeah Nov 19 00:47:49 i've been keeping track of the phone for a while now and i just can't seem to justify dumping 450 for non-wireless :-/ Nov 19 00:47:58 i suppose i will have to remain patient :-) Nov 19 00:48:07 you can wait Nov 19 00:48:09 and wait Nov 19 00:48:10 and wait... Nov 19 00:48:10 ha Nov 19 00:48:34 that sounds like i should just bite it and get one. Nov 19 00:48:55 well, Supposedly, december is when 02 is coming out Nov 19 00:49:05 from the copious amount of details released :p Nov 19 00:49:31 well thats not long :-) Nov 19 00:49:32 02 will be better in most respects Nov 19 00:49:44 well not most, i guess, but some Nov 19 00:49:50 better cpu, gfx Nov 19 00:49:51 my phone is about dead anyway after a year :-/ Nov 19 00:50:19 will you still be able to get the "developer" box even though its commercially ready? Nov 19 00:50:27 i want all the cool stuff to play with it as well Nov 19 00:50:42 you well Nov 19 00:50:43 it won't be buyable in december Nov 19 00:50:51 but the dev box of 02 will be $600 Nov 19 00:50:57 shit. Nov 19 00:51:20 also, that's an .. optimistic .. outlook, if you think it's going to be ready for consumers :/ Nov 19 00:51:25 i just read the /topic aparently i am not to ask about 02 :-) Nov 19 00:51:34 by december Nov 19 00:51:36 although i'd like to be wrong Nov 19 00:52:24 so right there it said the 02 will have wifi Nov 19 00:52:33 so for 600 bucks ill be set apparently. Nov 19 00:52:37 nod Nov 19 00:52:42 hmm. Nov 19 00:52:58 if you don't play with u-boot you might be able to get away with the non-dev-box ver Nov 19 00:53:15 i just want to code on it Nov 19 00:53:16 (afaik it also won't be buyable in january) Nov 19 00:53:24 i don't know what u-boot is Nov 19 00:53:31 balrog-kun: on a hunch, or a source? :o Nov 19 00:53:35 the bootloader Nov 19 00:53:45 basically if you fuck the bootloader you need the dev board to reload it Nov 19 00:53:46 like grub for the phone Nov 19 00:54:03 i shouldn't have to mess with that i would imagine Nov 19 00:54:07 probably not Nov 19 00:54:39 cool. thanks for the info. Nov 19 00:54:43 ill keep that in mine Nov 19 00:54:46 (d) Nov 19 00:55:08 anrp: i read the first run on 500 devices for developers would be on Dec 17 if nothing goes wrong then there's no more parts for at least two months Nov 19 00:55:13 s/on/of/ Nov 19 00:55:14 balrog-kun meant: anrp: i read the first run of 500 devices for developers would be on Dec 17 if nothing goes wrong then there's no more parts for at least two months Nov 19 00:55:45 i could believe that Nov 19 00:55:52 that's pretty sad though, since stock will likely run out Nov 19 00:56:01 quick ... Nov 19 00:56:44 the parts they have for the 500 units is the ones that they have in store now which were used for testing and design etc and some of them are not the final versions that will ship in the buyable units Nov 19 00:59:35 (for example the cpu) Nov 19 01:36:28 SpeedEvil: what does it mean when a GPS chip is "20 channel" or "32 channel"? Nov 19 01:37:43 * mjr hazards a guess that the more channels, the faster it finds something if it doesn't know the almanac Nov 19 01:38:11 hmmm' Nov 19 01:39:07 i'm looking at data sheets, and all the chips have ~40s warm start, irrespective of the number of channels Nov 19 01:40:49 DukeOfURL: number of satellites it can track Nov 19 01:41:04 the 40s is because the navigation messages shortest cycle is 30s Nov 19 01:41:26 but ther aren't 32 satellites--why 32 channels? Nov 19 01:44:52 how many satellites are there? Nov 19 01:46:38 the simulation in qemu has 16 satellites and i was wondering if it would make sense to take the date from system clock and increase the number of simulated satellites as years pass Nov 19 01:46:44 I think 24 Nov 19 01:47:21 ah ok, i will set it to 24 :) Nov 19 01:58:51 er Nov 19 01:59:03 you're never going to see more than maybe 12 sats at once in the real world Nov 19 01:59:07 open sky, flat ground Nov 19 01:59:12 in a city is less... Nov 19 02:00:46 12 is actually quite easy these days Nov 19 02:00:55 especially if you factor in WAAS Nov 19 02:01:04 ?? Nov 19 02:01:19 waas doesn't affect how many satellites you get, it affects (increases) your precision Nov 19 02:01:22 wide area augmentation system. Nov 19 02:01:45 WAAS is transmitted from geosynchronous satellites Nov 19 02:01:52 huh? Nov 19 02:01:53 dude Nov 19 02:02:02 ublox receivers can track them, and use them as part of the navigation solution Nov 19 02:02:12 hmm Nov 19 02:02:16 .. Nov 19 02:02:36 the one in my window is using PRN122 now Nov 19 02:02:40 you're right Nov 19 02:03:05 still, it shouldn't increase the # of satellites you can recieve Nov 19 02:03:19 depends on the chipset ;) Nov 19 02:03:24 just tell you the probable delay variation Nov 19 02:03:50 SiRF chips don't use ranging data from the WAAS birds, ublox does. Nov 19 02:04:38 ahh Nov 19 02:04:39 so one sunny afternoon, i was tracking 11 NAVSTAR PRNs and 1 WAAS PRN on my sirf2, Nov 19 02:04:45 that's different, but that makes sense Nov 19 02:04:57 the ublox right beside it was tracking the same 11 NAVSTAR, and 2 WAAS Nov 19 02:05:15 there just isn't more than 12 normal sats to see from the surface of the earth Nov 19 02:05:23 but since there's more than just the normal sats ... Nov 19 02:05:27 that won't be true for long Nov 19 02:05:34 its true for now :) Nov 19 02:05:39 there are plenty of spares in orbit Nov 19 02:06:06 .af.mil is/has been testing what happens with a larger almanac size Nov 19 02:06:59 well more sats is good for all Nov 19 02:07:06 multiple sat networks is almost better Nov 19 02:07:29 gps/glonass/galileo Nov 19 02:07:38 i'd prefer multiple frequencies Nov 19 02:07:45 then multiple networks Nov 19 02:07:56 ... multiple networks are on different freqs, generally Nov 19 02:08:04 although galileo is multiple freq to begin with Nov 19 02:08:09 GPS is getting upgraded iirc Nov 19 02:08:48 galileo civil service was designed to run in two bands from the get go - receivers are probably being built with that assumption Nov 19 02:09:14 it'd be nice if my $39 usb gps would work with navstar L1C and L5C... Nov 19 02:09:27 too bad that that increases receiver complexity .. but the accuracy is sometimes worth it Nov 19 02:09:29 alas, that's probably 5-7yr down the road Nov 19 02:09:43 and god only knows when galileo will actually make some progress Nov 19 02:09:54 too much political crap is holding them back Nov 19 02:09:59 always fun Nov 19 02:10:06 the us got that part right with navstar Nov 19 02:10:31 well its military-turned-everyone Nov 19 02:10:40 we-the-gov't, hereby order you-the-military to go forth and build us a navigation system Nov 19 02:10:43 the military doing it for itself, theres no politics involved for the most part Nov 19 02:10:48 just budgeting Nov 19 02:12:06 the problem with navstar is that it works too well Nov 19 02:12:32 that's hardly a problem Nov 19 02:12:39 there's not enough wrong with it to cause people to step up with bags full of money to build the next new system Nov 19 02:13:07 L1+correction data is good enough for most purposes Nov 19 02:13:15 exactly Nov 19 02:13:17 and if its not good enough then you probably already have the $ to spend on better receivers Nov 19 02:13:22 so eh, whatever Nov 19 02:13:41 see above... no real incentive to build something much better Nov 19 02:13:58 no one in europe wants to fund galileo, so the project office is trying to appropriate funds from agricultural budgets *snicker* Nov 19 02:14:23 "wants" -> concerted political will Nov 19 02:14:33 well, gps is free, and it works... just comes down to do you trust the US, which is a valid question to ask Nov 19 02:14:49 indeed Nov 19 02:15:30 although russia and india are working to refurbish glonass, so if that actually happens then you'll have two military-funded-everyone-accessible networks to use Nov 19 02:15:59 i think that not enough people would be severely hurt (or they haven't realized it yet) by the disappearance of GPS to really put some effort in. Nov 19 02:16:00 even less reason for galileo D: Nov 19 02:16:17 not enough? Nov 19 02:16:21 india's kicking in to glonass? i though theirs was called "GAGAN" Nov 19 02:16:29 pretty sure i read about that Nov 19 02:16:58 i know that .ru has committed to constellation completion and replenishment by end of 2009 Nov 19 02:17:04 but.. gps isn't going anywhere, the FAA pushed the military to make it permanent so that they can allow it to be used with IFR Nov 19 02:18:00 "not enough." europe likes to whine about the dependence on gps, but no one is actually funding the necessary work. to me that says they don't really care. Nov 19 02:18:11 s/gps/navstar/ Nov 19 02:18:11 ckuethe meant: "not enough." europe likes to whine about the dependence on navstar, but no one is actually funding the necessary work. to me that says they don't really care. Nov 19 02:18:38 i guess .ru will have an easier time finishing glonass if .in helps Nov 19 02:18:46 the only reason to would be a sense of political independance Nov 19 02:19:05 plenty of upscale vendors (topcon, javad, novatel) making glonass capable gear Nov 19 02:19:14 i wonder if they'll switch to CDMA Nov 19 02:19:21 switch Nov 19 02:19:22 ?? Nov 19 02:19:27 that's not the kind of thing you just switch :P Nov 19 02:19:29 yes, glonass is FDMA Nov 19 02:19:32 that would take decades Nov 19 02:19:48 on 9/11/2001 in the afternoon, I was sitting on top of Poison Spider Mesa in Moab. I had a Garmin on the handlebars of my mountain bike. The receiver regsitered zilch, nada, nothing. Nov 19 02:19:51 i'd love to see dual glonass+navstar capable chipsets hitting the mainstream Nov 19 02:19:53 the rumor mill says it's being considered Nov 19 02:20:20 i gather glonass should be better at higher latitudes too as it has 65% declination instead of 55% Nov 19 02:20:27 probably Nov 19 02:20:35 that is, the satellite orbits do Nov 19 02:20:44 .ru understands the concept of "it's gotta work up north" Nov 19 02:20:44 It was absolutely quiet. No planes, no contrails, no noise at all. Nov 19 02:21:03 But I did assume they turned the GPS system off. Nov 19 02:21:35 doubt it Nov 19 02:21:36 strongly Nov 19 02:21:49 DukeOfURL: i think it's safe to assume that one of the primary reasons that the new satellites are incapable of SA is that there are more subtle techniques Nov 19 02:22:03 .mil has said as much Nov 19 02:22:04 well, if all flights were grounded that day... Nov 19 02:22:12 SA? Nov 19 02:22:27 selective availability Nov 19 02:22:28 they have the capability to regionally deny the use of GPS without messing the clocks too bad Nov 19 02:22:53 SA = an attempt to degrade the accuracy of civilian GPS by adding noise/jitter to the clock Nov 19 02:22:58 not much mind you Nov 19 02:23:08 The FAA is publishing GPS ILS approaches. Nov 19 02:23:08 just about 100m Nov 19 02:23:08 just an order of magnitude Nov 19 02:23:10 yeah, a friend of mine reported that using an etrex in shenyang in china, over a period of months, the best fix he ever got was of the order of 100 metres Nov 19 02:23:16 a football field or so Nov 19 02:23:20 the same etrex happily gave 5m fixes in ireland. Nov 19 02:23:42 without knowing the noise environment, that doesn't mean much Nov 19 02:23:58 it could be degraded on purpose, it could be loaded with EM Nov 19 02:24:02 *EMI Nov 19 02:24:16 Anyhow, my Garmin was unable to acquire any satellites that afternoon... Nov 19 02:24:17 well, he tried it out in the countryside too, not just in the city - same result Nov 19 02:24:44 it wouldn't surprise me in the least if .mil had global zones of navstar quality Nov 19 02:24:57 ditto - it wouldn't surprise me at all Nov 19 02:24:58 north america gets 2m Nov 19 02:25:04 south america gets 10m Nov 19 02:25:29 ckuethe: well, isn't that because there's a bunch of waas monitoring stations in NA ? Nov 19 02:25:29 countries engaged in trade wars get 100m Nov 19 02:25:30 etc. Nov 19 02:25:52 we got high accuracy in singapore with some consumer units shipped there on a robot Nov 19 02:26:04 so it's gotta be pretty precise denial if that's happening Nov 19 02:26:23 i think the military would like us to arrive at that conclusion Nov 19 02:26:31 hoo boy, a conspiracy Nov 19 02:26:43 too lazy to go fish through my journal articles for a quote from the brass Nov 19 02:29:44 the logic is pretty sound though: Nov 19 02:29:44 a) selective availability was both a pain in the ass and fairly easy to defeat Nov 19 02:29:44 b) selective availability will no longer be used or even possible Nov 19 02:29:44 c) they're not going to deploy a system without some sort of way to deny precision gps Nov 19 02:29:44 d) there must be some new technique they're not talking about... Nov 19 02:29:47 The more recent satellites have enhanced regionalisation Nov 19 02:30:04 so they can selectively reduce accuracy to some areas Nov 19 02:30:13 also they have 'spot beams' Nov 19 02:30:24 which are military code, and 100* the normal power Nov 19 02:30:31 taking a page from the WAAS playbook, eh? Nov 19 02:30:42 looking at this: http://www.micro-modular.com/ Nov 19 02:30:59 directional antennae Nov 19 02:31:02 To some extend if you have IP connectivity, you don't care about the navigation messages the satellites broadcast Nov 19 02:31:21 as all you need to know is that the signal is the same over an area Nov 19 02:31:34 and several sites delay reports Nov 19 02:33:57 hm. that micromodular looks like a unav part Nov 19 02:34:09 orion is one of unav's navigation engines Nov 19 02:36:27 yep. the orion proprietary messages are $PUNV Nov 19 02:37:04 unav's hardware is pretty good, but the firmware is lam Nov 19 02:37:06 lame Nov 19 02:37:08 says orion in the datasheet :/ Nov 19 02:37:26 right... but orion is a unav product Nov 19 02:38:08 also, the orion firmware doesn't support 1PPS output even though the hardware can do it Nov 19 02:38:44 the itrax03 from fastrax is a unav chip, but running the isuite firmware... which is much nicer Nov 19 02:39:00 ..."Orion" is always overloaded with the drive in my mind Nov 19 02:41:22 hm. it looks like they unbroke 1pps Nov 19 02:41:53 last time i looked, orion couldn't do 1pps Nov 19 02:42:31 * SpeedEvil wants a neo with Orion 1pps 10kt drive. Nov 19 02:59:48 * hachi attemtps the moko makefile build again **** ENDING LOGGING AT Mon Nov 19 02:59:56 2007