**** BEGIN LOGGING AT Tue Mar 04 10:59:57 2008 Mar 04 11:00:00 * * OM Bug 1248 has been created by will(AT)openmoko.com Mar 04 11:00:02 * * Network connection alert is unstable Mar 04 11:00:03 * * http://bugzilla.openmoko.org/cgi-bin/bugzilla/show_bug.cgi?id=1248 Mar 04 11:12:06 lots abugz Mar 04 11:12:10 yum yum - crunchy Mar 04 11:12:26 hi folks. Anyone have an idea why mokomake build busybox for ARM 32bit architecture instead of x86_64? Mar 04 11:14:14 ... Because the Neo is ARM-based? Mar 04 11:14:35 ok Mar 04 11:14:57 but i can't run openmoko in my laptop Mar 04 11:15:47 thats kind of the idea... Mar 04 11:16:17 but i compile openmoko with x86_64 flags Mar 04 11:21:58 hi guys Mar 04 11:25:25 hi Mar 04 11:30:25 Man_of_Wax: i have never tried building for anything but arm Mar 04 11:31:36 i has a pidgin Mar 04 11:31:36 what is the sense of doing it for x86_64? Mar 04 11:32:04 chrooting it and using xoo/xephir Mar 04 11:34:53 Man_of_Wax, i did once try this and it worked great Mar 04 11:35:00 I did averything according to http://wiki.openmoko.org/wiki/Getting_OpenMoko_working_on_host_with_Xephyr Mar 04 12:07:04 openmoko: 03werner * r4165 10/branches/src/target/kernel/2.6.24.x/ (5 files in 2 dirs): Mar 04 12:07:04 openmoko: scripts/build, scripts/update: introduced variable PLATFORM to distinguish Mar 04 12:07:04 openmoko: gta01 from gta02 Mar 04 12:07:04 openmoko: config/Makefile: generate platform-specific defconfigs from a single master Mar 04 12:07:04 openmoko: config/defconfig-gta01, config/defconfig-gta02: generated defconfigs (for Mar 04 12:07:07 openmoko: convenience) Mar 04 12:42:16 cjb_ie, how's the openmoko community in ireland? Mar 04 12:43:22 tuukkah: dunno. i'm only a lurker here i'm afraid :) there's plenty of overlap between here and #osm (openstreetmap) though Mar 04 12:43:44 OSM's next annual conference is in limerick in july Mar 04 12:44:01 cjb_ie, did you see a neo yet? Mar 04 12:44:35 tuukkah: nope Mar 04 12:45:25 stupid me for not bringing mine to cork :-( but if you visit galway i'm happy to demo Mar 04 12:48:25 tuukkah: cool, thanks :) Mar 04 13:52:10 howdy all Mar 04 14:06:45 does anyone know whihc navigation app this is? http://scap.linuxtogo.org/files/e55809bf0e49edd7da61370e1fffdea0.png Mar 04 14:08:24 ScaredyCat, pyroute? Mar 04 14:08:37 or tangoGPS Mar 04 14:14:18 looks like pyroute... Mar 04 14:14:20 ta Mar 04 14:14:31 doesn't look like tango; my version of it, anyway Mar 04 14:14:37 ScaredyCat, now kneel before me Mar 04 14:14:51 suck my balls Mar 04 14:14:57 bleh Mar 04 14:15:20 i thought you would be scared..:S Mar 04 14:15:36 kneel before Zod! Mar 04 14:15:56 :D Mar 04 14:16:22 ScaredyCat, its pylgrim Mar 04 14:16:29 who? Mar 04 14:16:32 ...this looks like a job for SUPERCAT! Mar 04 14:16:32 me Mar 04 14:16:36 josch, hi long time Mar 04 14:16:40 hoi Sup3rkiddo Mar 04 14:17:11 got a url josch? Mar 04 14:17:13 a bird, a plane? no! it's supercat! Mar 04 14:17:25 ScaredyCat, http://neo1973-germany.de/wiki/pylgrim Mar 04 14:17:37 it's a python+evas application Mar 04 14:17:40 ta, does it use OSM ? Mar 04 14:17:43 yepp Mar 04 14:17:46 "But where is the mild-mannered, bespectacled ScaredyCat whenever SuperCat is around?" Mar 04 14:18:13 ScaredyCat, you can also have a look at this video demonstartion: http://rabenfrost.net/openmoko/pylgrim/pylgrim_drag&zoom&progress.ogv Mar 04 14:18:20 roflmao - mickey|sofa ... Mar 04 14:20:38 josch: nice Mar 04 14:20:39 :) Mar 04 14:22:06 can it route? Mar 04 14:23:36 ScaredyCat, no - it's only a map viewer but it connects to gps Mar 04 14:23:52 ah ok... looks good though ... Mar 04 14:24:09 i for myself needed nothing more than a map that shows where i am Mar 04 14:24:35 lol Mar 04 14:24:36 as you can see from the interface it's very minimal and this is the philosophy i wrote the tiny app with Mar 04 14:24:41 hehe pidgin is a little hard to use on the neo with the soft keyboard visible Mar 04 14:24:46 the .ogv crashed vlc Mar 04 14:24:52 zash, lol? Mar 04 14:25:00 zash, i can play fine with vlc Mar 04 14:25:12 zash, use mplayer..period Mar 04 14:25:23 mplayer and compiz are not friends :( Mar 04 14:25:30 use -vo Mar 04 14:25:46 -vo x11 -zoom :) Mar 04 14:25:55 works very fine Mar 04 14:26:36 aha Mar 04 14:26:39 tnx Mar 04 14:28:12 hm, the "downloaded %d of %d tiles" could be anoying Mar 04 14:28:40 yeah i will take care of this in feature releases Mar 04 14:29:09 :D Mar 04 14:29:15 one option atm is to dump all the osm tiles you need - it wont have to download anything Mar 04 14:29:46 or you just could use it for some time and because it cashes the tiles it will download very seldom Mar 04 14:30:11 a feature idea I'm going to implement is to pre-download tiles that /could/ be needed if the user drags further Mar 04 14:36:15 The right solution of course is to render directly from the OSM database - in a packed binary format. Mar 04 14:36:23 That's a whole bunch more coding though. Mar 04 14:37:16 this is what pyroute and navit are for i think Mar 04 14:37:33 and i didnt want o do a clone of those Mar 04 14:39:06 as I said: all I wanted was a map application as simple as possible and the only features I'll add are things like preloading tiles and double click anywhere on the display that switches between fixed-to-gps-mode and free-to-drag-around-mode Mar 04 14:40:00 eg. there wont be more controls than those + and - buttons Mar 04 14:40:31 it also was tested on a n770 and there you can use the up/down/left/right and zoom keys to navigate around Mar 04 14:41:10 josch: a voice note taker'd be nice too Mar 04 14:41:18 but... Mar 04 14:41:19 :-) Mar 04 14:42:44 notes of any kind are difficult - I will /maybe/ add such a feature but now I'm more interested to make the current state more robust Mar 04 14:43:01 hell it took me only two days to hack this thing together - this is the power of python! Mar 04 14:43:09 josch: just please don't include a 'download a gigabyte of tiles around my location' feature :) Mar 04 14:43:31 this feature is included as an external cli script Mar 04 14:43:40 Or at least throttle it to 1k/s or so. Mar 04 14:44:34 instead of other tile dumping scripts you tell it not four coordinates but only one coordinate of your target and a range around it in km and it calculates the appropriate lat/lon for you and asks you if you really gonna download 10k tiles now Mar 04 14:45:55 openmoko: 03werner * r4166 10/trunk/src/host/devirginator/environment.in: Mar 04 14:45:55 openmoko: Correct booting from SD/MMC: Mar 04 14:45:55 openmoko: - environment.in: "rootfs=" has to be "root=". Silly. Mar 04 14:45:55 openmoko: - environment.in: commented out ext2-only boot since ext2load is broken in Mar 04 14:45:55 openmoko: u-boot Mar 04 14:46:18 fix it werner.. fix it.. Mar 04 14:46:23 :) Mar 04 14:46:23 Strip - two destinations and a radius is also of use Mar 04 14:47:24 what's the full osm db size? Mar 04 14:47:37 does this commit mean that 2.6.24 still is not working from sd? Mar 04 14:47:39 (rendered) Mar 04 14:47:46 ScaredyCat, too much ^^ Mar 04 14:47:54 you can use ext3 just not ext2 Mar 04 14:48:02 i did some benchmark with packaging osm tiles in squashfs Mar 04 14:48:04 ScaredyCat: about 40GB or so Mar 04 14:48:07 the results where quite good Mar 04 14:48:12 ScaredyCat: but that's silly - as it's XML Mar 04 14:48:14 oh, not that muh then Mar 04 14:48:17 SpeedEvil, yepp sounds about right Mar 04 14:48:18 much Mar 04 14:48:27 ScaredyCat: the packed binary representation fits in around a gigabyte Mar 04 14:48:27 40GB is ok for a CarPC Mar 04 14:48:38 Not compressed - just packed Mar 04 14:48:47 :O Mar 04 14:48:54 show me your car that goes everywhere on the world an needs a 40G dump ^^ Mar 04 14:49:27 Relatively few countries you'd even consider using it as a car nav system Mar 04 14:49:54 US, NL, UK, germany, and some other limited bits of europe. Mar 04 14:49:56 depends if it includes off road sutff too Mar 04 14:50:00 sure - calculating from xml is reasonable but much slower than using prerendered bitmaps Mar 04 14:50:30 ScaredyCat: mostly on-road at teh moment. Mar 04 14:50:42 how long to render a tile with , say a 1.2ghz box - are we talking 10 secodns or 10 hours? Mar 04 14:51:07 It uses inkscape as a backend much of it. Mar 04 14:51:17 cool Mar 04 14:51:24 Fundamentally - not very long. Mar 04 14:51:36 so tiles are svg's then ? Mar 04 14:51:36 With current software - much of it is very long though. Mar 04 14:51:50 No - tiles are png Mar 04 14:52:01 :( Mar 04 14:52:06 the osm data is xml but Mar 04 14:52:18 with xslt you can convert it to svg Mar 04 14:52:21 wouldn;t svg's zoom working? Mar 04 14:52:23 this is the magic of xml ;) Mar 04 14:52:31 work better.. Mar 04 14:52:37 on the neo? Mar 04 14:52:39 josch: and the horrible, horrible speed. Mar 04 14:52:42 generally Mar 04 14:52:48 not just on the neo Mar 04 14:53:25 just seems to like : if you have the cpu, use it, if not use prerendered tiles Mar 04 14:53:28 You really don't want to draw large tiles as vectors on slow hardware. Mar 04 14:53:58 A hundred thousand lines really slows stuff up. Mar 04 14:54:06 navit and pyroute are doingexactly this Mar 04 14:54:22 but they do it in a very optimised and complex way Mar 04 14:54:31 and both are slow (naturally) Mar 04 14:54:58 well, I guess you don't render the 100,000 lines, only the ones in the viewport, and only to a certain level Mar 04 14:55:12 of detai; Mar 04 14:55:46 yepp but this requires a lot of work and not only those two days it took me to throw pylgrim together ^^ Mar 04 14:57:51 ;) Mar 04 14:59:16 ScaredyCat: of course - you don't end up with something that looks as nice as a tile though. Mar 04 14:59:32 ScaredyCat: I mean if you zoom out of a major metropolis. Mar 04 15:00:01 * * OM Bug 1249 has been created by mickey(AT)openmoko.org Mar 04 15:00:02 * * No Wifi in Managed Mode Mar 04 15:00:03 * * http://bugzilla.openmoko.org/cgi-bin/bugzilla/show_bug.cgi?id=1249 Mar 04 15:00:07 !!! Mar 04 15:00:11 * * OM Bug 1250 has been created by mickey(AT)openmoko.org Mar 04 15:00:12 * * No Wifi in Managed Mode Mar 04 15:00:15 * * http://bugzilla.openmoko.org/cgi-bin/bugzilla/show_bug.cgi?id=1250 Mar 04 15:00:35 well, I've probably only ever zoomed out half a doezn times with my tomtom go... Mar 04 15:00:48 and the info in that is pretty much al you need Mar 04 15:01:07 mickeyl: same bug twice? Mar 04 15:01:20 did you search bugzilla first? :P Mar 04 15:01:50 heh Mar 04 15:01:54 ~lart 'reload' Mar 04 15:01:55 * apt whips out his power stapler and staples 'reload''s foot to the floor Mar 04 15:03:59 mickeyl: but the big box that says "do you realy want to POST again?" Mar 04 15:04:36 hello Mar 04 15:05:04 zash, mickeyl uses plain tcp to browser the net and hit enter twice after he typed POST Mar 04 15:05:11 like josch says Mar 04 15:05:16 hardcore browsing Mar 04 15:05:31 telnet bugzilla.openmoko.org 80 Mar 04 15:05:35 probably python-efl-webkit :-) Mar 04 15:05:39 schhhhhhhhh Mar 04 15:05:40 roflmao Mar 04 15:05:44 don't you dare telling everyone Mar 04 15:06:10 but webkit re-implented in py-vala Mar 04 15:07:03 heh Mar 04 15:07:03 curl ftw Mar 04 15:08:20 I bet he's really using Vista :P Mar 04 15:08:26 and ie Mar 04 15:23:19 openmoko: 03mickey * r4167 10/trunk/src/target/OM-2007.2/artwork/themes/openmoko-standard-2/gtk-2.0/gtkrc: gtkrc: add note about TakuIconTile features Mar 04 15:24:29 actually no. i got rid of Windows lately Mar 04 15:27:42 real men use set the universal constants and wait for the reload event to unfold naturally Mar 04 15:27:53 ..copied from xkcd of course Mar 04 15:28:03 we have video footage of mickeyl and his windows lovage :-) Mar 04 15:28:19 hehe Mar 04 15:28:54 xkcd pwnz Mar 04 15:41:50 XorA, url or it didn't happen :-P Mar 04 15:42:09 josch: look at koens OEDEM photos from a couple of years ago Mar 04 15:42:38 *cough+ Mar 04 15:49:17 got a windows virus mickeyl? Mar 04 15:50:32 damn no pictures to be found :-/ Mar 04 15:51:03 * Sup3rkiddo has a deja vu on this conversation Mar 04 15:51:47 * ScaredyCat passes out pictures of Sup3rkiddo with a goat and some lard.... Mar 04 15:52:15 somehow i was expecting ScaredyCat to pwn me Mar 04 15:52:25 the matrix is hre Mar 04 15:54:48 <`pwgen`> hello Mar 04 15:56:58 <`pwgen`> there are rumours that there should be an gsmd daemon scripted in python not useing the openmokogsmd2. where can i find it ? Mar 04 15:58:32 `pwgen`, svn checkout svn://projects.linuxtogo.org/svn/smartphones/trunk/software/pyneod Mar 04 15:58:48 it's python and works great Mar 04 15:59:16 josch: is your svn server running on a C64? Mar 04 15:59:30 ScaredyCat, only a vserver Mar 04 15:59:41 ScaredyCat, you can make donation - then i can switch :-P Mar 04 15:59:49 pyneod isn't a gsmd daemon Mar 04 16:00:02 that uses mickeyl's gsmmux Mar 04 16:00:08 <`pwgen`> hmm i hope .. it is the 3 try getting someting that can speak with the gsm ... Mar 04 16:00:28 ScaredyCat, no it also runs standalone Mar 04 16:00:43 ScaredyCat, and the gsm muxer is also from emdete Mar 04 16:01:31 you can run pygsmd standalone and with the moxer - as standalone version it will provide you a dbus interface to doing all the phone stuff Mar 04 16:01:51 aye, mybad - sorry emdete. Mar 04 16:02:06 * ScaredyCat slaps mickeyl Mar 04 16:03:51 <`pwgen`> hmm and where can i get moxer aka gsmmux ? Mar 04 16:04:41 `pwgen`, svn checkout svn://projects.linuxtogo.org/svn/smartphones/trunk/software/gsm0710muxd/ Mar 04 16:04:54 josch: what are the python deps for pylgrim ? Mar 04 16:05:09 gsm0710 is gprs thingies...right? Mar 04 16:05:10 <`pwgen`> josch: THX Mar 04 16:05:34 `pwgen`: you can download a ready to run image of emdete's stuff from my buildhost Mar 04 16:05:40 ScaredyCat, you additionally need optparse and dbus Mar 04 16:05:56 ta Mar 04 16:06:37 <`pwgen`> *G* and what ist the url of your buildhost ? Mar 04 16:06:37 Sup3rkiddo, gsm0710 is X times gsm plus Y time gprs at the SAME TIME Mar 04 16:06:50 http://buildhost.automated.it/ Mar 04 16:07:38 http://buildhost.automated.it/OM2007.2/moko-underground-image-fic-gta01.jffs2 Mar 04 16:07:38 josch, danke :D Mar 04 16:07:38 is a recent image.. Mar 04 16:07:38 underground? Mar 04 16:07:46 emdete's stuff Mar 04 16:07:56 thats a snazzy name Mar 04 16:08:06 you don't read my blogposts or mickeyl's do you... Mar 04 16:08:19 ScaredyCat, you have a blog? Mar 04 16:08:27 * ScaredyCat sobs Mar 04 16:08:41 ooh, ooh, there is progress on the multiplexing front? Mar 04 16:09:09 mjr, ask emdete about this - iirc there are still major problems Mar 04 16:09:14 ScaredyCat, gee, whats the link? Mar 04 16:09:20 planet.openmoko.org Mar 04 16:09:21 ;) Mar 04 16:09:26 dang Mar 04 16:09:27 http://blog.automated.it Mar 04 16:09:29 so obvious Mar 04 16:09:34 it's on planet Mar 04 16:09:34 ~lart Sup3rkiddo Mar 04 16:09:35 * apt does a little 'dpkg -P Sup3rkiddo' action Mar 04 16:10:36 josch, okay. Well. It's just that the neo would be so much more useful if I could keep gprs on ~continuously Mar 04 16:10:39 <`pwgen`> thx to all, i will try it out Mar 04 16:18:34 mjr: AIUI the gprs can only be suspended during calls - a hardware limitation. Mar 04 16:18:49 yes yes, ~ Mar 04 16:18:56 the mux means calls can interrupt a gprs session Mar 04 16:24:29 mjr: what's the question about the muxer? Mar 04 16:24:58 I was mostly generally enthusiastic that there's progress on that front Mar 04 16:25:17 mjr: not progress, it's finished since weeks :D Mar 04 16:25:23 emdete, so it works? Mar 04 16:25:27 mjr: shure Mar 04 16:25:37 mjr: om does not like it because it's user-space Mar 04 16:25:42 ich sehe Mar 04 16:25:48 mjr: and it uses dbus for channel allocation Mar 04 16:26:37 mjr: you /may/ try to use it with libgsmd as someone already tried but this is a little unstable i think if gsmd does not know dbus Mar 04 16:27:07 mjr: i also used pppd directly with it (without pypppd) Mar 04 16:27:08 i wouldn't state it like that Mar 04 16:27:22 mickeyl: then correct me ;) Mar 04 16:27:34 OM Inc. loves gsm0710muxd. Mar 04 16:27:42 do they? :D Mar 04 16:27:46 although they still would prefer an in-kernel solution Mar 04 16:28:01 hey, they didn't tell me! i like love-stories! Mar 04 16:28:24 hehe Mar 04 16:28:25 mickeyl: if the neo does umts i start with that, versprochen Mar 04 16:28:31 hahah Mar 04 16:28:34 righto Mar 04 16:28:42 one step at a time Mar 04 16:28:45 but honestly Mar 04 16:28:54 for now the user-space-daemon does not eat any cpu with my setup Mar 04 16:29:04 given the trouble we have with modem control, i start thinking the userland daemon makes much sense Mar 04 16:29:35 with an in-kernel muxer it's tough to get the layering right Mar 04 16:30:18 yes, for now we at least know how that thing has to work. if someone moves that to the kernel at least that is solved Mar 04 16:30:47 mickeyl: the ipk of gsm0710muxd seems to be broken - it tries to install init.d scripts... Mar 04 16:31:20 ok, i'll look at it Mar 04 16:32:02 mickeyl: also some dependencies are missing when trying to move to "underground" Mar 04 16:32:44 k Mar 04 16:32:48 a full rebuild is due today Mar 04 16:33:03 since TMPDIR changed ABI Mar 04 16:33:09 ? what does that mean? Mar 04 16:33:14 :D That'll be fun... Mar 04 16:33:23 fat ass changes landed today in OE.dev Mar 04 16:33:30 for the good, of course Mar 04 16:33:37 when - i started a build some hours ago Mar 04 16:33:39 but needs a full rebuild Mar 04 16:33:49 That's what they all say... "This is for your own good"... :D Mar 04 16:33:56 hehe Mar 04 16:34:05 Right up there with: "This will hurt me more than it will hurt you!" :D Mar 04 16:34:12 mickeyl: can you give me a time when you did that Mar 04 16:34:12 heh Mar 04 16:34:13 ? Mar 04 16:34:18 emdete: a few hours ago Mar 04 16:34:23 :/ Mar 04 16:34:26 emdete: if you could start a build you don't have the changes yet Mar 04 16:34:34 OE will warn you if you have an old TMPDIR Mar 04 16:34:38 Seriously, though, it looks to be a good change, but I think there will be some trauma in getting things sorted. Mar 04 16:34:48 absolutely Mar 04 16:35:04 it's not for free, but increasing the metadata is the top goal of OE Mar 04 16:35:10 +quality Mar 04 16:35:11 mickeyl: how can i check if i have a good "TMPDIR"? Mar 04 16:35:23 emdete: OE will not build if your TMPDIR is too old Mar 04 16:35:25 it will say Mar 04 16:35:29 TMPDIR ABI 0 found Mar 04 16:35:43 just don't mtn pull for a while Mar 04 16:36:49 can't follow you... old TMPDIR is bad, i shouldnt pull the new, so what? Mar 04 16:36:54 FYI: it was in rev. ce72f9f467a7c0cf9bc1e280749f8f8259ef44b9 Mar 04 16:37:15 ooops Mar 04 16:37:27 well, if you don't have time to do a full rebuild, don't pull Mar 04 16:37:32 that's it Mar 04 16:37:48 okay... Mar 04 16:37:49 and not a good 'oops' either Mar 04 16:38:22 mickeyl: you could have said that 8 mins ago... Mar 04 16:38:30 right before I hit enter Mar 04 16:38:38 :D Mar 04 16:39:09 ScaredyCat: monotone is a SCM, just update to another rev Mar 04 16:39:24 46a0a75288d26760fdfab6d87d48fb71e729d3ea is the last one before Mar 04 16:40:07 ~lart webkit for failing to build Mar 04 16:40:07 * apt says "boot to the head" and knocks webkit over for failing to build Mar 04 16:41:26 * mwester cuts the rev mickeyl posted, and pastes it into a file for future reference... Mar 04 16:41:42 I wish someone had tagged that version. Mar 04 16:42:14 (that would be OE that would do that, I guess, not the OM folks) Mar 04 16:42:39 This only applies to the openembedded repo, not the openmoko repo yet, correct? Mar 04 16:43:50 yep Mar 04 16:44:12 you can't build bleeding edge, and not be at the edge mickeyl :) Mar 04 16:44:23 indeed Mar 04 16:46:41 bleeding edje... Mar 04 16:47:16 hehe Mar 04 16:47:25 #opEnmoko Mar 04 16:50:58 mine was more funny :-P Mar 04 16:53:42 true Mar 04 17:00:41 So is it possible to shut down an 'underground' install right now w/o an ssh cxn? Mar 04 17:01:28 no Mar 04 17:01:33 well Mar 04 17:01:33 k, cool ;-) Mar 04 17:01:34 wurp2: you mean you are looking for the switch-off in the gui? no, not currently Mar 04 17:01:34 yes Mar 04 17:01:40 rip the battery out Mar 04 17:01:52 emdete: thx Mar 04 17:01:58 that doesn't require ssh Mar 04 17:02:01 ScaredyCat: yeah, I knew that :-P Mar 04 17:02:05 wurp2: sry, it's just 6 weeks old ;) Mar 04 17:02:12 slacker Mar 04 17:02:15 :P Mar 04 17:02:15 emdete: Oh, it's very impressive Mar 04 17:02:24 I was just making sure I didn't miss a piece of the UI Mar 04 17:02:44 It looks as if I could add an option to shut down in like 5 minutes if I was willing to just devote a 'thing' to it Mar 04 17:03:10 :D yes, the power daemon already can do it ;) Mar 04 17:03:11 there's a free button 'hole' Mar 04 17:03:12 I glanced over pyneog & pyneod, or whatever those /usr/share dirs are called Mar 04 17:03:37 yes, its for daemon & gui. Mar 04 17:04:07 ScaredyCat: you mean to put "off" to that button? :D Mar 04 17:07:06 Any plan to make one python "app server" type of application that multiple python apps can live in? Mar 04 17:07:25 because devoting 4mb to each little process could get to be a bit much Mar 04 17:08:05 4mb? Mar 04 17:08:17 where do u gt that # from? Mar 04 17:08:29 from running a python program that does nothing but wait Mar 04 17:08:34 and looking at top output Mar 04 17:08:42 ands how do u calculate that number? Mar 04 17:08:46 what in top? Mar 04 17:08:48 2.5 mb resident, 4 mb total Mar 04 17:08:56 I didn't calculate, just look at the # in the output Mar 04 17:09:07 wurp2: yes, if we are really going to do a full python-UI, then it's going to be an app server / app client model Mar 04 17:09:09 4 mb VIRT for a python process Mar 04 17:09:09 so it's more 2.5MB, much of which would presumably be shared between python instances Mar 04 17:09:36 wurp2: whats the share # ? Mar 04 17:09:49 Maybe I'm badly misreading top, but it shows a separate entry for shared Mar 04 17:09:52 and it's like 1.5 mb Mar 04 17:10:01 so at least 2.5 mb per process Mar 04 17:10:01 so really its moe like 1.5m Mar 04 17:10:10 with a base cost for all python apps of 1m Mar 04 17:10:12 er, 4 - 1.5 = 2.5 Mar 04 17:10:15 no Mar 04 17:10:17 1 Mar 04 17:10:26 ok res is ? Mar 04 17:10:31 whats res or rss? Mar 04 17:10:40 res is just 2.5 Mar 04 17:10:49 ok Mar 04 17:10:50 but still, the total memory matters Mar 04 17:10:53 and share is 1.5? Mar 04 17:11:08 no - your'e reading it wrong Mar 04 17:11:09 I agree that it is 1 mb res per process Mar 04 17:11:12 1.5m of it is shared pages Mar 04 17:11:20 that means they are common with other processes on your system Mar 04 17:11:23 they are the same memory pages Mar 04 17:11:27 they ar enot duplicated Mar 04 17:11:30 1.5 of the 2.5 res, right? Mar 04 17:11:33 they are re-used in-place Mar 04 17:11:36 yes Mar 04 17:11:39 1.5 of the 2.5 Mar 04 17:11:43 so u run 1 python process Mar 04 17:11:45 but the total mem is still over 4 mb Mar 04 17:11:54 it will almost guaranteed use the exact same 1.5 m shared base Mar 04 17:12:01 so while each process may only take 1 extra mb resident Mar 04 17:12:03 this is also shared with other procs like c/c++ etc. based ones Mar 04 17:12:06 no Mar 04 17:12:09 vsize is just virtual space Mar 04 17:12:20 tats unimportant (unless u get close to 4g on a 32bit box) Mar 04 17:12:23 res/rss is your key Mar 04 17:12:27 share is what u subtract Mar 04 17:12:37 That swapped space has to go in & out sometimes Mar 04 17:12:38 so your real memory footprint is RSS-SHR Mar 04 17:12:41 one would assume Mar 04 17:12:41 i.e 1mb Mar 04 17:12:46 so IO matters too Mar 04 17:13:09 VSIZE/SIZE is a superset of all of it Mar 04 17:13:14 no Mar 04 17:13:15 I wasn't sure whether share was after res, or part of it Mar 04 17:13:17 it wouldnt Mar 04 17:13:21 it is not touched or used Mar 04 17:13:36 RES is your resident memory Mar 04 17:13:44 that is the amount of memory you are actually using Mar 04 17:13:46 in pages Mar 04 17:13:53 that are actually active and live in real memory Mar 04 17:14:05 they can either be private or shared Mar 04 17:14:08 OK, if I have an app that actively uses 64 mb on a 32 mb box, it will run Mar 04 17:14:15 the SHR is the amount of ram in RSS/ Mar 04 17:14:17 and will only have (at most) 32 mb res Mar 04 17:14:22 but that other 32 mb still matters Mar 04 17:14:26 because it's getting swapped in & out Mar 04 17:14:27 the SHR is the amount of ram in RSS/RES you SHARE with other processes or can share Mar 04 17:14:38 OK Mar 04 17:14:48 no it doesn't - if u arent swapping out Mar 04 17:15:01 in the case of a neo - u arent swapping Mar 04 17:15:04 there is nowhere to swap to Mar 04 17:15:14 (unless u went and put swap on flash.. or sd) Mar 04 17:15:29 most apps have much more vsize/size than res/rss Mar 04 17:15:55 So an app can have memory in VIRT that doesn't have any real correlation at all in hw? Mar 04 17:16:00 unused memory in the virtual memory space is just not used Mar 04 17:16:11 its a virtual mapping that is not used/neded etc. Mar 04 17:16:19 correct Mar 04 17:16:22 it can Mar 04 17:16:28 Sorry, I was unaware of that... I thought it all had to be mapped somewhere physical Mar 04 17:16:32 yeah... eat my ram! Mar 04 17:16:37 Is there a way to see how much mem is swapped out, then? Mar 04 17:16:44 if i mmap() a 2g file on disk Mar 04 17:17:29 my SHR will go up by 2g Mar 04 17:17:35 and my RSS will also go up by 2G Mar 04 17:17:42 but i actually will not allocate that much Mar 04 17:17:56 But in that case, the memory does have a real corollary in hw Mar 04 17:17:58 the file Mar 04 17:18:07 yes Mar 04 17:18:08 Not that I'm arguing against your general point Mar 04 17:18:09 but not ram Mar 04 17:18:25 the point is - your mem footprint is RSS-SHR Mar 04 17:18:38 I can see that the processor could provide a virtual memory space of 4 gig or whatever and not physically back it up except when you use it Mar 04 17:18:46 the SHR amount u need to be careful of as it may be split among many current processe sand be a BASE SYSTEM COST DUE TO EVERYONE USING IT Mar 04 17:19:01 surely Mar 04 17:19:02 oops Mar 04 17:19:05 caps Mar 04 17:19:23 yes Mar 04 17:19:26 correct Mar 04 17:19:33 I should say, not physically back it up *until* you use it Mar 04 17:19:50 in fact each process can have up to 4g (in theory minus some hw mappijgn and other stuff - so in reality about 3.5g) mapped to it Mar 04 17:19:59 u can have 100 procss each with 3.5g mapped to it Mar 04 17:20:12 and it may use almost not actual sstem resources Mar 04 17:20:20 And that tells you nothing about how much mem + swap it uses Mar 04 17:20:27 correct Mar 04 17:20:34 but, I still dunno how you see how much real mem + swap an app uses Mar 04 17:20:34 RSS is the only thing that tells u Mar 04 17:20:38 if u have no swap Mar 04 17:20:43 if u have swap - then it gets tricky to tell Mar 04 17:20:49 (well have swap and are using it) Mar 04 17:20:59 as i said Mar 04 17:21:07 i am assuming u are not knee-deep in swap Mar 04 17:21:07 OK, well thanks very much for the tutorial ;-) Mar 04 17:21:16 that your numebrs came from a neo - and thus had no swap Mar 04 17:21:18 :) Mar 04 17:21:21 u are right Mar 04 17:21:34 if u have a chunk of your app swapped out the rss is now not a useful number Mar 04 17:21:59 the absolute best measurment u can do is on a whole-system basis Mar 04 17:22:11 boot up to point X Mar 04 17:22:14 eg a gui Mar 04 17:22:16 or whatever Mar 04 17:22:19 do a: Mar 04 17:22:27 free Mar 04 17:22:30 look at: Mar 04 17:22:37 -/+ buffers/cache: 135992 1939680 Mar 04 17:22:51 the first numebr there is the actual in-ram memory footprint of your whole system Mar 04 17:23:09 then run your app Mar 04 17:23:12 whatever it is Mar 04 17:23:15 then do it again: Mar 04 17:23:22 sure Mar 04 17:23:22 -/+ buffers/cache: 134616 1941056 Mar 04 17:23:31 and your app takes negative memory ;-) Mar 04 17:23:36 so i went down there Mar 04 17:23:43 something on my system freed up resources Mar 04 17:23:46 Yeah, I understand, oother stuff going on Mar 04 17:23:49 as its a whole system - its complex Mar 04 17:23:51 :) Mar 04 17:23:57 but it at least means that your app is at the noise level Mar 04 17:24:03 let me take a better example than xterm Mar 04 17:24:04 :) Mar 04 17:24:27 Just do it a hundred times, and average the result :-) Mar 04 17:24:57 But at any rate, devoting 1 mb of ram to each python process on a system w/ 128 mb total seems excessive Mar 04 17:25:08 113468 -> 132660 Mar 04 17:25:12 that is to run gimp Mar 04 17:25:35 so 19192kb Mar 04 17:25:47 sounds reasonable Mar 04 17:25:51 is the "cost" of gimp to bring its gui up to its initial state Mar 04 17:25:56 from a baseline system Mar 04 17:26:11 and highlights the questionability of 1 mb per do-nothing python process Mar 04 17:26:28 python do-nothing < 1/20th of the gimp Mar 04 17:26:30 that would be all real malloc()ed ram (ram that would need to be swapped out) Mar 04 17:26:36 (should be, that is) Mar 04 17:26:54 u can measure increas in cache/buffer usage too to see what level of file IO has been needed to get it up Mar 04 17:27:04 this is stuff that can be paged to/from disk dynamically not needing swap Mar 04 17:27:11 ut it does impact performance Mar 04 17:27:39 u are right tho Mar 04 17:27:41 wow python uses more memory at start that ... clisp Mar 04 17:27:46 but 1mb is probably a base setup cost for python Mar 04 17:28:03 you could make your app have 1000 tims the code and u'd probably hit 2mb Mar 04 17:28:05 for example Mar 04 17:28:11 Yeah, that was why I was asking about client/server Mar 04 17:28:18 so u need something more than a trivial example Mar 04 17:28:19 well, app server Mar 04 17:28:32 although I thought the problem was more than twice as bad as it is :-) Mar 04 17:28:41 yeah Mar 04 17:28:44 shame python doesn't have real threads so you can't load multiple apps into the same image Mar 04 17:28:49 thats why i questioned your numbers :) Mar 04 17:28:56 I appreciate it Mar 04 17:29:12 i just like to help peolpe do the right reading of their memory output Mar 04 17:29:13 I'm not sure what python does when you fork Mar 04 17:29:19 its almost alwmis-readays Mar 04 17:29:21 err Mar 04 17:29:24 mis-read Mar 04 17:29:26 always Mar 04 17:29:29 almost Mar 04 17:29:47 I'll play with checking memory on forked python processes Mar 04 17:30:09 unknown_lamer: You can do multiple apps in the same image... it's just not very clean Mar 04 17:30:12 beware the copy-on-write monster Mar 04 17:30:20 wurp2: it should have the same SHR/RES Mar 04 17:30:20 it will do interesting things to your stats Mar 04 17:30:21 :) Mar 04 17:30:27 Unless someone has done something with it that I don't understand Mar 04 17:30:52 unknown_lamer: Really? You think that if you do it multi-threaded (not forked) that it will occupy full memory? Mar 04 17:30:59 I can't see how that would be the case Mar 04 17:31:17 wurp2: no if you fork then it will use full memory (by the nature of fork), but threads should be lightweight Mar 04 17:31:17 raster: Sure - the fork could go nuts on memory usage as soon as you try to actually do something Mar 04 17:31:37 wurp2: initially after fork u should only use 1 or 2 more pages of ram Mar 04 17:31:40 or therabouts Mar 04 17:31:43 unknown_lamer: I haven't done explicit multithreading in python, but I've seen it Mar 04 17:31:45 but yes Mar 04 17:31:55 after that it depends how many more pages of ram u write to Mar 04 17:32:58 unknown_lamer: I just dunno what you mean when you say python doesn't support real threading Mar 04 17:33:01 wurp2: isn't there some interpreter lock that makes it kind of useless except for making blocking syscalls? Mar 04 17:33:40 unknown_lamer: That may be... but on a single cpu system, is multithreading really useful except during blocking syscalls? Mar 04 17:33:43 and for time-sharing Mar 04 17:33:57 timesharing naturally Mar 04 17:34:11 you don't want one thread to enter some computation loop that never calls something that yields Mar 04 17:34:33 unknown_lamer: I do think I've read about the issue you're talking about, but it sounded to me as if it only really applied to taking advantage of multiple cpus Mar 04 17:34:37 dunno how python does that though (perhaps after every interpreter loop?) Mar 04 17:34:53 python is one of the languages I looked at and decided it didn't do anything worth learning it for Mar 04 17:35:05 because I am an elitist jerk Mar 04 17:35:55 I think a well supported language that lets you write 10 lines of code that do the work of 50 in another language is worth learning ;-) Mar 04 17:36:18 Of course, lisp would probably be a better choice, except that I want to also get paid contractor $$$s to write code Mar 04 17:36:26 wurp2: unless you already know a language that lets you write one line of code to do the work of those ten lines :-) Mar 04 17:36:33 I am paid to write common lisp Mar 04 17:37:06 The market didn't look good enough for lisp to try to switch to it professionally :-/ Mar 04 17:37:10 you just have to find clients who want you to handle everything Mar 04 17:37:11 unknown_lamer: very nice Mar 04 17:37:12 they seem to pay more too Mar 04 17:37:29 "just write the damn thing and set up hosting and stuff for it, or install it on our intranet server now GET A CODING" Mar 04 17:37:48 True, but finding such clients consistently seems a trick Mar 04 17:37:55 when I tell them I can do it in half the time for 1.5x the hourly rate and then actually pull it off ... Mar 04 17:38:01 yeah Mar 04 17:38:15 since those are typically the people who don't know how to find anyone in the industry Mar 04 17:38:22 this is why I like longer jobs (e.g. "hello please replace our secretary with a set of fifteen programs") Mar 04 17:38:56 wurp2, I think the idea of comparing languages based on the number of lines of code required to do something is fairly flawed Mar 04 17:39:09 the only downside is that I have nothing public to show potential new clients :-( Mar 04 17:39:14 wurp2, as you have just pointed out in your memory analysis anyway :-) Mar 04 17:39:15 just "I swear I did this you can email these people" Mar 04 17:42:17 thos: Really? On the desktop, unless you're writing a 3D game, cpu and memory are probably the last of your worries Mar 04 17:42:39 wurp2, and what's that got to do with lines of code? Mar 04 17:42:55 Time to market and # of dev hours is the concern Mar 04 17:43:02 wurp2, and I hardly think cpu and memory should be the least of your worries Mar 04 17:43:14 And while it's not *always* true that writing 1/10th the code takes less time, it usually is... Mar 04 17:43:15 wurp2, oh, and there was me thinking about quality of software :-) Mar 04 17:43:26 Surely Mar 04 17:43:46 lines of code != quality of software Mar 04 17:43:52 You haven't gotten to market in any meaningful way unless your sw does what it's supposed to Mar 04 17:43:57 exactly Mar 04 17:44:08 and that has no relation on the number of lines of code you can do it in Mar 04 17:44:10 But more lines of code is more places for it to go wrong Mar 04 17:44:19 Not true Mar 04 17:44:19 nope Mar 04 17:44:47 wurp2, you think the number of lines of code in python + your app is less than the number of lines in just C and libc? Mar 04 17:45:02 thos: Often, yes Mar 04 17:45:09 I doubt it Mar 04 17:45:15 how many lines of code are there in python? Mar 04 17:45:20 thos: the lines in the standard library don't count Mar 04 17:45:20 But even when it isn't, the code in python is tested WAY more than the code in my app Mar 04 17:45:25 nor in any library Mar 04 17:45:46 unknown_lamer: I would say in any widely used library Mar 04 17:46:04 aye, I wanted to qualify that with "well written and tested" Mar 04 17:46:09 both OO and imperative sux....do it in a functional language ;) Mar 04 17:46:11 reused code is freeeeeee Mar 04 17:46:32 That's a myth. Mar 04 17:46:52 so you are basically saying you can't write good code Mar 04 17:46:59 thos: I'm basically saying that I'm smart enough not to write code that's already been written and well tested Mar 04 17:47:04 thos: Do you unit test? Mar 04 17:47:12 If you do, you'r saying you can't write good code Mar 04 17:47:22 who needs unit testing when you can write a formal correctness proof Mar 04 17:47:24 And if you don't, you're an idiot (for any real project) Mar 04 17:47:34 wurp2, too right Mar 04 17:47:41 unknown_lamer: People who aren't 100% confident in their test ;-) Mar 04 17:47:59 s/test/proof/ Mar 04 17:47:59 wurp2 meant: unknown_lamer: People who aren't 100% confident in their proof ;-) Mar 04 17:48:00 wurp2, but I don't think it automatically follows that just because you can write less code in one language, it automatically means the applications will be better Mar 04 17:48:05 I agree Mar 04 17:48:15 It is only a rule of thumb Mar 04 17:48:24 a pretty bad one Mar 04 17:48:27 Lack of type definitions in python is a big deal Mar 04 17:48:52 wurp2, if you're a good programmer, you can write the same quality of code in any language Mar 04 17:49:05 thos: hardly Mar 04 17:49:07 wurp2, just probably python doesn't allow it to blow up in your face as much Mar 04 17:49:16 but, when I write an app in TurboGears it takes less than half the time to get the same quality as when I use java/tomcat Mar 04 17:49:26 I could never write as quality code in C as in say SML Mar 04 17:49:34 thos: agreeing with unknown_lamer on that one Mar 04 17:49:36 owing to the utter lack of language features necessary for clean design Mar 04 17:49:41 yup, so experience too Mar 04 17:49:48 forcing me to use terrible terrible hacky imperative crap Mar 04 17:49:54 python makes it so that when it blows up, it tells me why Mar 04 17:49:59 so really, it's all down to the experiance and quality of the programmer Mar 04 17:50:04 not the language Mar 04 17:50:11 wurp2: and CL makes it so that you can restart where the computation left off ;-) Mar 04 17:50:13 Disagree Mar 04 17:50:20 plus just use the right language for the job Mar 04 17:50:36 Agreed Mar 04 17:50:54 although, for me, the right language for prototyping is python Mar 04 17:51:09 and often the prototype works well enough that I don't rewrite Mar 04 17:51:27 I like CL because my prototyping and production language are the same Mar 04 17:51:31 I'm sure that if I knew common lisp as well as unknown_lamer, that would be my choice Mar 04 17:51:44 wurp2, sorry, I've just seen so many people say "python is better because I don't have as many lines of code", and yet as we have just pointed out, that is only one small factor Mar 04 17:51:53 'tho I keep meaning to write some /refactoring/ (just "rename class / method / swap arguments around" stuff really) tools for slime Mar 04 17:52:04 But I also suspect I would still be doing java for $$$s :-( Mar 04 17:52:13 wurp2, and for embedded devices, I have yet to hear any *user* advantages for using it Mar 04 17:52:22 wurp2, as opposed to advantages for developers Mar 04 17:52:35 it == python? or Java? Mar 04 17:52:45 wurp2, anything :-) Mar 04 17:52:52 developer advantages translate into user advantages of having more and hopefully better software available Mar 04 17:52:54 wurp2, but I was thinking about python really Mar 04 17:52:57 well, the advantage would be getting an app of reasonable quality sooner Mar 04 17:53:04 *hopefully* Mar 04 17:53:09 true Mar 04 17:53:20 lua seems to be a win for embedded devices due to small footprint Mar 04 17:53:26 wurp2, but you just said, each python process is currently taking up about 1mb for no apparent reason Mar 04 17:53:33 pjz, right Mar 04 17:53:35 Every app would be better written in microcode, if we had the time & brainpower to spend on it Mar 04 17:53:50 wurp2: myth Mar 04 17:53:52 wurp2, I think there's a balance Mar 04 17:53:59 thos: Yup, and that's an issue that would have to be solved Mar 04 17:54:02 anything that compiles to C is also a good choice like Scheme ! Mar 04 17:54:09 pjz: Not a myth, that's a clear fact Mar 04 17:54:12 hah, scheme Mar 04 17:54:26 wurp2: compilers are better at optimization than most programmers Mar 04 17:54:28 does anyone use that outside of the odd comp sci course? Mar 04 17:54:38 pjz: Take an app in any other language in the best possible way, xlate it to the microcode that executes, find one optimization Mar 04 17:54:55 thos: I agree that there's a balance Mar 04 17:54:57 wurp2, the point is, we know python has memory issues, so why are people still trying to argue for it's use on embedded devices based on factors such as lines of code Mar 04 17:54:58 that was my point Mar 04 17:55:10 And I think C is on the far side of that balance for most applications Mar 04 17:55:44 the lack of GC in C results in terrible memory use issues Mar 04 17:55:53 owing to the inability to feasibly use structure sharing Mar 04 17:56:23 it makes life very annoying to have to keep copying stuff to make simple nondestructive updates Mar 04 17:56:42 agh, I just stood someone up for a lunch date Mar 04 17:56:45 argue with you later :-) Mar 04 17:57:20 ewon well yeh but in specialized apps mainly....scientific.... but if u look http://chicken.wiki.br/Eggs%20Unlimited%203 u see has extensions for most things Mar 04 17:58:48 + Mar 04 19:58:27 does dmcrypt work with openmoko ? Mar 04 20:08:34 elation: Sorry, I have no idea Mar 04 20:48:07 good evening Mar 04 20:48:15 good evening sxpert Mar 04 20:48:20 it appears the documentation to the calypso chip were just leaked Mar 04 20:48:26 see cryptome Mar 04 20:49:29 jaysus I'm starving Mar 04 20:51:14 sxpert: link? Mar 04 20:51:29 http://cryptome.org/ti-calypso1.pdf Mar 04 20:51:37 http://cryptome.org/ti-calypso2.pdf Mar 04 20:53:48 win 15 Mar 04 20:54:06 sxpert: access denied Mar 04 20:54:14 hmm ?? Mar 04 20:54:23 go via the main site Mar 04 20:54:30 http://www.cryptome.org Mar 04 20:54:35 first and second link Mar 04 20:54:57 it's probably checking referrer Mar 04 20:54:58 no, won't work Mar 04 20:55:05 one sec :-) Mar 04 20:55:09 Access Forbidden Mar 04 20:56:11 there, mirrored :-)http://www.sxpert.org/torrents/TI-Calypso%20documentations/ Mar 04 20:56:38 Yay, now how about one for the Hammerhead GPS chip? Mar 04 20:56:48 heh Mar 04 20:57:39 thnx Mar 04 20:58:27 I asked on the mailing list, what does this means by "leaked"? Were these given up freely? Does viewing that put developers at risk? Mar 04 20:58:38 yes Mar 04 20:58:41 well Mar 04 20:59:07 you're vnot covered by any licences that may be granted by the NDA Mar 04 20:59:09 the cat's out the bag now... they can't complain anymore Mar 04 20:59:18 and are violating the copyright of Mar 04 20:59:19 TI Mar 04 20:59:28 sxpert: of course they can Mar 04 20:59:52 well. the "trade secret" ain't secret anymore. too bad for them Mar 04 20:59:52 wow Mar 04 21:00:42 * SpeedEvil wonders what's in it. Mar 04 21:00:58 SpeedEvil: what do you think can happen - worst case? Mar 04 21:01:18 I em: the TI ninjas come and take away your computer. Mar 04 21:01:22 "I got those docs from cryptome"... Mar 04 21:01:46 emdete: practically - they have as much right as any media company to cause problems for your internet connection or whatever. Mar 04 21:01:54 It's no different from distributing an mp3. Mar 04 21:01:56 Strictly Private Mar 04 21:01:59 UNDER NON DISCLOSURE Mar 04 21:02:18 :D Mar 04 21:02:18 funny day Mar 04 21:03:06 this being on cryptome, it's public now... cat ain't going back in the bag Mar 04 21:03:23 As I understand it, copying out the data in altered form, means anyone could look at it legally. Mar 04 21:03:43 (however, you're still in violation of the copyright for making a copy) Mar 04 21:03:49 yeah, well... Mar 04 21:04:00 SpeedEvil: so if i change my program using information from that could that lead to problems? Mar 04 21:04:19 emdete, well... Mar 04 21:04:21 jtag, nice Mar 04 21:04:35 get someone to rewrite the doc, and use said rewrite :-) Mar 04 21:04:52 that will then be a legal doc Mar 04 21:04:57 yeah Mar 04 21:05:19 emdete: in theory, they could cause you problems for copyright infringment. Mar 04 21:05:51 I'm not completely sure what the problems are for users of any GPLd code you may have generated from that copyright infringment. Mar 04 21:06:43 usually they concede defeat Mar 04 21:07:56 anyhoo... this could lead to totally free GSM phone Mar 04 21:08:12 with network debugging features and stuff... Mar 04 21:08:14 sxpert: that would make it illegal to sell in the UK Mar 04 21:08:36 It'd be an actual imprisonable offence to sell it. Mar 04 21:08:41 SpeedEvil, so long as it's sold with the official firmware, you're probably safe. Mar 04 21:09:02 then nothing prevents the owner of hacking said firmware for more features Mar 04 21:09:15 hi all, I want to port openmoko to sun4cdm instead if ficgta01, does any one know how can I do that? Mar 04 21:09:34 Any coder of any such open-source phone software is a mobile phone hacker, and is similarly liable for jail-time. Mar 04 21:09:39 (in the UK) Mar 04 21:10:03 (at least, once the first person abuses the code) Mar 04 21:10:18 furthermore in the UK everyone is presumed terrorist before being proven guilty Mar 04 21:10:40 This legislation was brought in when phone cloning was a big problem. Mar 04 21:10:48 ah well... Mar 04 21:11:00 It's really aimed at guys with soldering irons chipping phones to alter the IMSI/IMEI Mar 04 21:11:07 To get free calls. Mar 04 21:11:33 However, software that lets users do this is equally liable to prosecution. Mar 04 21:11:37 SpeedEvil, hmm... What's the purpose of the SIM then ? Mar 04 21:12:06 isn't the SIM supposed to be not-cloneable and containing the subscription info ? :-) Mar 04 21:12:18 Or rather Mar 04 21:12:33 that was part of it - some phones held the credit info on the Mar 04 21:12:39 phone Mar 04 21:13:06 the legislation was also to cover that. The changing IMEI/IMSI is more if they steal the phone, so it can be unbarred Mar 04 21:14:04 yeah, well.. Mar 04 21:14:19 the thing was obviously badly designed, as usual Mar 04 21:14:41 GSM is quite an old protocol. Mar 04 21:15:14 doesn't excuse the security sloppyness within Mar 04 21:16:01 anyhow, the doc is now in the wild... Mar 04 21:16:37 And I wonder if they're going to find out who it leaked from. I don't want to be in that person's shoes... Mar 04 21:17:22 'my computer got stolen, ...' Mar 04 21:17:26 heh Mar 04 21:17:31 * mjr mostly wonders if it actually is enough to be able to upload new firmware to the chip Mar 04 21:18:12 Even if it is, developing anything to actually _use_ the radio would be a huge task. Mar 04 21:18:19 mjr: download is Mar 04 21:18:53 SpeedEvil, reverse engineering the binary blob is probably doable though, and would allow adding stuff within Mar 04 21:18:59 SpeedEvil, yeah, ynezz, nah Mar 04 21:19:06 If not signed. Mar 04 21:19:16 If not encrypted Mar 04 21:19:18 rather Mar 04 21:19:51 SpeedEvil, can probably be hacked around. docs are dated 2000... Mar 04 21:20:04 it's not like this is the latest Mar 04 21:20:53 good code security in 2000 is good code security now Mar 04 21:20:53 the docs just describe the hw, no word about the software on it... so for me it doesn't help alot. Mar 04 21:21:21 emdete, the software within the binary blob makes use of the hardware Mar 04 21:21:35 to realize the "GSM modem" functionnality Mar 04 21:21:56 sxpert: shure - i am interested in the real behavior of the varius standard and (not documented) non-standard AT commands Mar 04 21:22:19 emdete, that you can gather by disassembling the binary blob Mar 04 21:22:26 sxpert: and: i am not interested to reverse-engenier that. i am intersted in a stable pygsmd ;) Mar 04 21:22:55 one would go with the other I suppose :-) Mar 04 21:23:09 Move pygsmd onto the calypso! Mar 04 21:23:19 the one thing I'd like is network debugging features like certain nokia phones can do Mar 04 21:23:32 SpeedEvil: having a dbus between the 2 arms? :D Mar 04 21:23:56 emdete, using the serial port as communication medium :-) Mar 04 21:24:06 yes! Mar 04 21:24:17 throw away hayes! Mar 04 21:25:15 let me call even using the "public emergency" channels, so that I can still call out when everyone is trying to call from the convention floor Mar 04 21:26:02 sxpert: excelent use-case! yes, that makes intersting features into the neo Mar 04 21:26:46 also, add encryption after the GSM encoder :-) Mar 04 21:27:06 you don't want to do that really Mar 04 21:27:25 you want a pure data mode - to give you the raw data, over which you run your own FEC Mar 04 21:27:52 ah, that too :-) Mar 04 21:27:58 the GSM voice codecs are designed to accept errors without horrible audio result Mar 04 21:28:07 error free channels are a bit rare. Mar 04 21:28:11 :D the neo could get an interesting device this way Mar 04 21:28:17 ah ok. then... encrypt and then data mode :-) Mar 04 21:28:30 without incurring the data surcharge, obviously Mar 04 21:28:56 and without having to pay 6K to siemens for an encryption module Mar 04 21:29:49 SpeedEvil, alternatively, use a different encoder that allows for 100% data recovery Mar 04 21:30:52 Someone else that read it reports to me that it's basically limited to what hardware is absolutely required to do the hard-realtime. All else is software. Mar 04 21:31:21 So the binary blob is really the core of the implementation. Mar 04 21:31:22 which means we could add anything really :-) Mar 04 21:32:16 Again though, this would violate TIs copyright if you just alter the blob and distribute it. Mar 04 21:32:21 that thing is similar in design to those "softmac" wifi cards then Mar 04 21:32:29 SpeedEvil, right Mar 04 21:32:37 sxpert: basically yes. Mar 04 22:24:36 * dvarnes hopes the doc leak isn't a show stopper for the next round of hardware OM Inc tries to develop Mar 04 22:25:26 doc leak? Mar 04 22:26:46 Blastur, yeppers :-) Mar 04 22:27:41 explain Mar 04 22:27:49 is this common in the mobile handset ODM space, or really unusual? Mar 04 22:28:30 Blastur, documentation for the calypso chipset appeared today on cryptome.org Mar 04 23:02:23 Gary Gygax, RIP. :-( Mar 05 00:33:23 hey, am looking for a beta tester for the new release tangoGPS - anybody with a neo awake? Mar 05 01:22:27 ello :P Mar 05 01:22:40 olleh Mar 05 01:22:55 ?niod uoy woh Mar 05 01:23:19 sleepy Mar 05 01:23:22 heheh Mar 05 01:23:26 :o Mar 05 01:23:34 can't write THAT backwards! Mar 05 01:24:10 you have a neo? Mar 05 01:24:25 nah :( Mar 05 01:24:29 waiting for freerunner Mar 05 01:24:37 you? Mar 05 01:24:40 yep Mar 05 01:24:54 nice Mar 05 01:24:57 am developing software Mar 05 01:25:02 tangogps.org Mar 05 01:25:06 850mhz working yet? ;) Mar 05 01:25:14 am in europe Mar 05 01:25:36 nice :) Mar 05 01:25:39 yeah, ok :P Mar 05 01:25:43 bad question haha Mar 05 01:25:51 actually I'm quite happy with the neo1973 Mar 05 01:25:59 (nice about tangogps, i mean ;)) Mar 05 01:26:19 hmm, i'm awaiting a few upgrades which will make me happy, heheh Mar 05 01:26:20 new release is due tomorrow Mar 05 01:26:28 of tango? Mar 05 01:26:31 yes Mar 05 01:26:42 goodie, but really not in a condition to test that now Mar 05 01:27:05 mjr - but you do have a neo? Mar 05 01:27:32 it is a lot more polished now and a couple of nice new features Mar 05 01:27:35 gamin, just sent you an overlong email..... Mar 05 01:27:51 hmm... Mar 05 01:28:04 yes I do Mar 05 01:28:11 so the GTA02 will not be quad-band? :( Mar 05 01:28:17 it will not, sadly Mar 05 01:28:26 uriahheep, 850 version though for you Mar 05 01:29:06 its a hemisphere phone, not a world phone :-) Mar 05 01:29:13 ah, ok, cool then :) Mar 05 01:29:32 ...how will the 850 version work? will it simply be a modded version of the current hardware? Mar 05 01:29:57 ...because, well, if possible, it would be cool to have some kind of switch so that it can either be 850 or 900... Mar 05 01:30:12 not gonna happen Mar 05 01:30:18 you would just open the phone up, change a jumper, and then it would be a 900 phone instead of 850 ;) Mar 05 01:30:24 hmm, too bad... Mar 05 01:30:35 jamewill_: thanks for the mail, just read it... will answer tomorrow. do you have a neo already? Mar 05 01:30:36 perhaps that's something that i could do to the phone myself or something, heh Mar 05 01:31:14 gamin, nope.... waiting patiently... Mar 05 01:32:10 ok, 2 more hours and this damned fedora 8 dvd is finished downloading :-/ Mar 05 01:32:18 the things i do for customers, i tell ya Mar 05 01:39:16 gamin, annyway, does there happen to be a method of preloading maps a bit better in the upcoming version? Mar 05 01:40:22 mjr: no, not in the next one. however, there are some scripts floating around for doing so. Mar 05 01:41:08 yeah well, I did do one myself too, just noted how insanely much there is to download for say, a city Mar 05 01:41:34 can you mail me yours? Mar 05 01:41:37 hmm, why not use google maps? Mar 05 01:41:49 uriahheep: because osm is open? Mar 05 01:41:50 well... that would only be good when connected to the net i guess Mar 05 01:41:55 gamin: ah, ok :P Mar 05 01:42:13 actually you can use googlemaps Mar 05 01:42:26 at least the maps and the terrain. Mar 05 01:42:35 the sat images don't work Mar 05 01:42:41 with tangogps you mean? Mar 05 01:42:44 yes Mar 05 01:42:49 cool :) Mar 05 01:43:19 why wouldn't the satellite pics work? Mar 05 01:43:52 they use a different coding scheme and I haven't yet found any other open server using it. Mar 05 01:44:01 hmm, ok Mar 05 01:44:06 gamin, it's very quick and dirty, but sure, address? Mar 05 01:44:11 not so nice legal ground, even if maaemo-mapper is doing it Mar 05 01:44:19 could you just use webkit and just run the google maps js api though? Mar 05 01:44:32 marcus dot bauer at gmail dot com Mar 05 01:45:17 uriahheep: webkit is just overkill and way to slow and crashes too often (as of now) Mar 05 01:45:25 ah, ok ;) Mar 05 01:45:41 I like links, has a great graphical mode too. Mar 05 01:46:00 ...so, how ready for production do you guys think the software available for openmoko right now really is in general? Mar 05 01:46:06 unfortunately it stopped working here the other day. Mar 05 01:46:10 hmmm Mar 05 01:46:16 links <3 Mar 05 01:46:17 heheh Mar 05 01:46:35 uriahheep: I use it as my daily phone, but I'm tough in taking ;) Mar 05 01:46:37 it isn't that great for xhtml+css3, though ;) Mar 05 01:46:47 gamin: heheh :P Mar 05 01:47:03 gamin: what do you use for music? Mar 05 01:47:11 or do you not use it for that? :P Mar 05 01:47:34 currently I don't use it for music, but madplay (commandline) works great. Mar 05 01:48:11 the mediaplayer suffers from pulseaudio which in turn suffers from missing floating point and some other stuff Mar 05 01:48:11 hmmm Mar 05 01:48:35 but lennart is aware of this Mar 05 01:48:42 gamin: what about for listening to an entire album? how long does it take for madplay to load the next song? Mar 05 01:48:56 no delay at all Mar 05 01:48:59 hmm, nice Mar 05 01:49:05 what about for flac, though? ;D Mar 05 01:49:26 ...and, well, does the wifi support WPA right now? ;P Mar 05 01:49:31 * mjr goes to crash now Mar 05 01:49:39 gnight mjr Mar 05 01:49:51 * gamin goes to sleep too Mar 05 01:49:56 aw :( Mar 05 01:49:59 gnight :P **** ENDING LOGGING AT Thu Mar 06 12:37:06 2008