**** BEGIN LOGGING AT Fri Feb 09 03:00:00 2007 Feb 09 04:19:03 SpeedEvil: counter is for 23:59:59.999... CST which actually is Monday 13:00 at Taiwan (IIRC) Feb 09 04:35:45 i wonder if they have posted out all the developer phones already so they arrive on the same day as the source gets released or if they will post them on that day Feb 09 04:35:58 in any case, cant wait for all the pics to appear :) Feb 09 04:41:25 I think they post them Monday Feb 09 04:41:49 * aloril remembers some comment about developers working on Sunday, but shipping only Monday Feb 09 04:42:09 though.. Monday at Taiwan is Sunday at USA ;-) Feb 09 04:47:59 Great post on linux audio: http://0pointer.de/blog/projects/foms-lca-recap.html Feb 09 04:51:05 xkr47: http://svn.xiph.org/trunk/speex/libspeex/resample.c Feb 09 05:00:25 CM, hmmmkay Feb 09 05:00:25 :) Feb 09 05:00:47 Read Lennarts post? Interesting I say... :) Feb 09 05:01:10 Lennart is a smart guy, thats for sure Feb 09 05:01:49 I grepped "fixed" and only read a few lines about libsydney and the resampler :D Feb 09 05:01:55 I'll try to look more later Feb 09 05:02:26 however I found a house for sale that I might be interested in buying so my mind is not exactly tuned for deep resampling thoughts atm :) Feb 09 05:03:07 Best of luck :) Feb 09 05:03:24 thx Feb 09 05:03:29 hope you have a great day too :) Feb 09 05:43:56 hmm.. GPRS in order delivery sucks, but I guess it was done because of crappy phones in past Feb 09 05:44:05 why do it in UMTS too? Feb 09 05:44:36 (or I remember wrong on UMTS) Feb 09 07:06:47 counter Feb 09 07:06:47 2 days 22:20:40 (2.931 days) for source for *all* developers and devices for selected developers (2007-02-11); a month, 2 days (30.931 days) for devices for *ANYBODY* for $350 (targeting developers) (preorders: not yet) (2007-03-11); 7 months, 2 days (214.931 days) for mass market (2007-09-11): see topic for more info (182) Feb 09 07:11:36 ... soon for source Feb 09 07:13:27 I'm looking forward to a decent wiki too.. Feb 09 07:13:40 yeah, that too Feb 09 07:14:10 and of course hearing from those > 50 developers Feb 09 07:14:48 And some nice new pics of the hw :) Feb 09 07:17:44 Are there plans for a device emulation project? Something for developers without the hardware? Feb 09 07:18:45 etahhhh: Yes, the simples thing is just to use Xoo: http://projects.o-hand.com/xoo/ Feb 09 07:19:23 No immidate need to emulate the hardware since it will run X and gtk Feb 09 07:20:03 Right. It would be sweet if they could 'award' phones for building something that is ready to run. Feb 09 07:21:40 world won't be same after this (of course you could have build your own linux wearable computer, HMD, twiddler, etc.., but this is first mass market item) Feb 09 07:22:20 etahhhh: That's a nice idea I think. Some kind of competition of "best app" or so Feb 09 07:23:58 Exact, throw a bone to people that aren't in the first wave and don't have 'development' budget. Feb 09 07:25:25 Help out anybody that makes significant contribution to mass adoption/usability Feb 09 07:28:38 Shoot, I'm just trying to figure a way to get my hands one, when I don't have the cash or embedded experience Feb 09 07:29:10 I am sooo ready to ditch my blackberry Feb 09 07:32:42 Hey, if the 'competition' takes off, maybe they would give me one for contributing the idea of giving them to people who contribute ;) Feb 09 07:33:41 umm.. there might be prior art for that idea ;-) Feb 09 07:34:08 If you're in the us I'm sure you can paten it anyway :P Feb 09 07:34:24 s/aten/atent/ Feb 09 07:34:25 CM meant: If you're in the us I'm sure you can patent it anyway :P Feb 09 07:34:32 hmm, if he were in the US he might have money for a developer edition of the phone :) Feb 09 07:34:43 For real. Feb 09 07:35:03 Heh, true. I think $350 is cheap.. Feb 09 07:35:12 for some maybe :( Feb 09 07:35:34 quoting myself: 2007-01-04 16:01:37 and if you develop enough nifty application(s) maybe you get next version for free? who knows .. (I'm not with FIC so of course don't know anything about potential for this ;-) Feb 09 07:35:53 not exactly same idea, but similar ;-) Feb 09 07:37:30 Ditto. It makes a lot of sense reward people who further the cause. Especially if there is a way to do it without the device. Feb 09 07:38:12 actually its quite likely you will get v2 for free if you show yourself being useful with v1 Feb 09 07:38:39 (few intermediate lines and then this): 2007-01-04 16:02:50 absolutely Feb 09 07:38:39 that's a nice idea and a lucrative one too :) Feb 09 07:39:49 Nokia did provide vouchers for 770 developers to get 800 cheaply (or was it for free?): 500 of them IIRC Feb 09 07:40:11 99 euros iirc Feb 09 07:41:01 so_solid_moo: OK, that means its cheap, but still you need to be serious: good idea I think Feb 09 07:41:22 It would also make sense to give them to people lining up bulk purchases, anything over 10 or something like that. Feb 09 07:41:26 * aloril suspects v2 will be significantly more pricey than v1 (at least some versions) Feb 09 07:41:46 v2 will have different versions ? Feb 09 07:41:57 Motivation for voluntary sales force. Feb 09 07:42:19 Eblis: dunno, could have, depends on how you read some Sean mails Feb 09 07:42:29 aha Feb 09 07:42:47 might be only one version or could be 2 versions: with and without keyboard or ..oh well we will see ;-) Feb 09 07:42:57 :) Feb 09 07:47:24 maybe they'll like my mono and portable .net ports to OE :D Feb 09 07:48:14 Eblis: did you send an email to coreteam@ ? Feb 09 07:48:22 yes Feb 09 07:48:39 and openmoko-devel Feb 09 07:48:46 before mickey made the list? Feb 09 07:48:54 it would be nice to have this for v2 (I assume there is no suitable connector for v1 and USB1 is too slow for VGA maybe?): http://www.aeinnovations.com/projects/ver0/ Feb 09 07:48:55 that i don't know Feb 09 07:49:02 i finished porting this week :( Feb 09 07:49:20 hmm. I think mickeyl can add to that list later too..? Feb 09 07:49:26 monday/tuesday Feb 09 07:49:30 dunno - ask him :-) Feb 09 07:50:01 mono was added to OE but portable .net wasn't added yet ... the bugzilla entry is still open. maybe they're waiting a bit before adding a second .net framework app Feb 09 07:50:20 when was the list made ? Feb 09 07:51:47 * aloril thinks it was made about time announcement was made Feb 09 07:53:11 hi Feb 09 08:13:58 i guess our phone is delaying too much Feb 09 08:14:47 have you seen the latest samsung stuff ? Feb 09 08:15:11 http://www.akihabaranews.com/en/en/news-13261-When+the+Korean+GOD+awake%2C+he+gives+us...+the+Samsung+Ultra+Smart+F700.html Feb 09 08:18:23 OMG NO WIFI!!!!! (on the F700) Feb 09 08:18:46 It has that strange qwrty-thing :) Feb 09 08:19:49 don't like that keyboard :( Feb 09 08:21:11 The only benefit I see is the camera and even then I'd buy a real digital camera. Feb 09 08:21:20 rwhitby: haha Feb 09 08:21:39 aloril: :-) Feb 09 08:21:50 * aloril hopes Neo1973 is big success -> more hardware with GPL drivers Feb 09 08:22:07 qwrty ? I wouldn't by a product with no "e" on it ! Typical Samsung :D Feb 09 08:22:24 Sorry, that was my typo. Feb 09 08:22:25 * aloril excepts that F700 to be quite a lot more expensive than Neo1973 Feb 09 08:22:37 j/k ;-) Feb 09 08:22:53 Oh :D Feb 09 08:25:04 if money is not obection: 64GB flash: http://www.cdw.com/shop/products/default.aspx?EDC=959056 Feb 09 08:25:30 s/obection/objection/ Feb 09 08:25:30 aloril meant: if money is not objection: 64GB flash: http://www.cdw.com/shop/products/default.aspx?EDC=959056 Feb 09 08:29:06 * xkr47 just loves that sed support Feb 09 08:30:51 counter Feb 09 08:30:51 2 days 20:56:36 (2.873 days) for source for *all* developers and devices for selected developers (2007-02-11); a month, 2 days (30.873 days) for devices for *ANYBODY* for $350 (targeting developers) (preorders: not yet) (2007-03-11); 7 months, 2 days (214.873 days) for mass market (2007-09-11): see topic for more info (183) Feb 09 08:30:55 woowoo Feb 09 08:33:27 OpenMoko is like an evangelical messianic cult. Feb 09 08:33:52 Evangelical = we must spread the word. Feb 09 08:34:08 Messianic = we wait for the gadget. Feb 09 08:34:25 cult = I WANT MY OPENMOKO NOW!!!! ? Feb 09 08:35:18 You got it. Feb 09 08:35:25 nerd = I need to make a lot of noise on irc on all kinds of niche stuff Feb 09 08:37:09 That's a pricey flash drive Feb 09 08:38:23 :) Feb 09 08:39:08 counter Feb 09 08:39:09 2 days 20:48:18 (2.867 days) for source for *all* developers and devices for selected developers (2007-02-11); a month, 2 days (30.867 days) for devices for *ANYBODY* for $350 (targeting developers) (preorders: not yet) (2007-03-11); 7 months, 2 days (214.867 days) for mass market (2007-09-11): see topic for more info (184) Feb 09 08:39:13 ah Feb 09 08:39:17 almost 3 days ;( Feb 09 08:39:58 hads: indeed, just getting 64x1GB SD cards would be cheaper ;-) Feb 09 08:40:52 True, probably not quite as convinient though :) I wonder how many people would buy such a thing. Feb 09 08:45:47 hads: yes, not as convenient, but I wonder where that 'extra price' comes from Feb 09 08:47:49 topic Feb 09 08:48:25 Indeed Feb 09 08:49:57 30.867 days! less than a month :) Feb 09 08:50:48 is glass half full or half empty? ;-) (less than month or more than month depending on month) Feb 09 08:52:41 aloril: the glass dosen't have wifi or camera Feb 09 08:54:02 my glass has better camera than any phone I know ;-) Feb 09 08:54:15 or rather, will have ... Feb 09 08:54:32 just wish it would support more than basic PTP Feb 09 08:58:03 (probably more useful to use in USB mass storage than in PTP mode) Feb 09 08:58:44 ptp? Feb 09 09:00:51 can we install openmoko on the samsung ultra smart f700?? Feb 09 09:02:23 you just need skill, that's all :) Feb 09 09:05:25 http://en.wikipedia.org/wiki/Picture_Transfer_Protocol Feb 09 09:06:46 looking at libgphoto2-2.3.1/camlibs/ptp2/library.c it seems than Nikon cameras support capture too and Canon capture + review (and other just browsing/copying/deleteing, ie.. I think basically poor man mass storage) Feb 09 09:07:13 aloril: my kodak is PTP Feb 09 09:07:36 counter Feb 09 09:07:43 2 days 20:19:50 (2.847 days) for source for *all* developers and devices for selected developers (2007-02-11); a month, 2 days (30.847 days) for devices for *ANYBODY* for $350 (targeting developers) (preorders: not yet) (2007-03-11); 7 months, 2 days (214.847 days) for mass market (2007-09-11): see topic for more info (185) Feb 09 09:07:43 XorA: yes, most cameras do support PTP Feb 09 09:07:53 aloril: nah, most do mass storage Feb 09 09:07:58 aloril: ptp was stupid Feb 09 09:08:01 XorA: but does your kodak allow you to snap image using PTP? Feb 09 09:08:30 aloril: not under linux, but that could be due to the piss poor implementation Feb 09 09:08:39 XorA: if it allows you to snap picture, then PTP makes sense Feb 09 09:09:14 XorA: could be that or it could be that camera has limited implementation Feb 09 09:09:42 aloril: wouldnt know, gphoto2 tends to crash rather than extract pictures, I used to use a java program to do it Feb 09 09:09:53 aloril: in end its quicker to take out SD card and put in SD reader Feb 09 09:12:00 XorA: indeed, but with Neo1973 you might want to use camera in USB mass storage mode (it's USB1 after all, so speed doesn't 'matter') Feb 09 09:12:11 saves lugging card reader ;-) Feb 09 09:12:31 aloril: put micro SD card in yah camera :-) Feb 09 09:12:40 ptp? never used Feb 09 09:12:40 * aloril hates rebooting Feb 09 09:12:56 aloril: reboot? Feb 09 09:13:04 micro SD is under battery Feb 09 09:13:47 solder wires to battery :-) Feb 09 09:13:53 haha Feb 09 09:14:26 anyway, don't except to do that much stuff with Neo1973, so USB mass storage should be fine Feb 09 09:14:39 (for transmitting images) Feb 09 09:14:46 aloril: yeah should work fine Feb 09 09:15:54 anyway, allows to boast "better camera than any other phone" ;-) (yeah, I guess there are some other phones with USB port or some cameras with bluetooth/wifi and phones with similar support but ...) ;-) Feb 09 09:18:31 neo screen is gonna be gorgeous for photos, 640x480 at nice small size Feb 09 09:19:45 yeah, though for any significant amount you might want to do rescaling in PC Feb 09 09:20:28 VGA takes about 100KB, so 2Gb card should be able to hold 20000 images ;-) (unless you want also high resolution one, then 1/10 or 1/20) Feb 09 09:21:42 the price of flash has been driven down and down Feb 09 09:21:55 * aloril really hopes that OpenMoko + Neo1973 isn't just a really good dream ;-) Feb 09 09:22:00 hey fluffs Feb 09 09:51:24 mickeyl, don't work too late today okay? Feb 09 09:51:24 :) Feb 09 09:52:59 morning Feb 09 09:53:21 morning *checkstime* ;-) Feb 09 09:53:41 aloril: although it feels like a dream so far, I'm actually surprised that no manufacturer has done it before FIC, after well over a decade of cellphones being normal. Feb 09 09:54:11 TRIsoft: ~ugt Feb 09 09:54:14 ~ugt Feb 09 09:54:21 apt ? Feb 09 09:54:23 After all, it's not exactly hard to put bits you don't want touched behind a UART. Feb 09 09:54:26 well, ugt is Universial Greeting Time. Created in #mipslinux, it is a rule that states that whenever somebody enters an IRC channel it is always morning, and it is always late when the person leaves. The local time of any other people in the channel, including the greeter, is irrelevant. http://www.total-knowledge.com/~ilya/mips/ugt.html Feb 09 09:54:31 ~botsnack Feb 09 09:54:31 mickeyl: aw, gee Feb 09 09:55:11 Morgreet: yeah, but I guess they consider carriers as customers :-( Feb 09 09:55:26 mickeyl: :D Feb 09 09:55:26 aloril: true Feb 09 09:56:39 mickeyl: viva? Feb 09 09:57:15 Morgreet: first version of slides finished and sent to supervisor. will now reread my thesis (it's been 4 months since i submitted...) and then work on my share of the OpenMoko presentation and the OE presentation Feb 09 09:57:32 [for FOSDEM, that is] Feb 09 09:57:51 * fluffs prepares a cup of tea Feb 09 09:58:02 fluffs: what kind of tea do you prefer? Feb 09 09:58:04 Super! A relief to send the whole damn thing off. Don't stop reading new research papers in the meantime. Feb 09 09:58:07 * mickeyl is into Roiboos-vanille Feb 09 09:58:34 Morgreet: *nod* i'm 100% sure I will get asked about new developments in meantime. Feb 09 09:58:37 hey mickeyl Feb 09 09:58:40 cheers rwhitby Feb 09 09:58:44 mickeyl: tea-direct standard tea Feb 09 09:59:09 fluffs: ah. is that fair trade? Feb 09 09:59:14 yes Feb 09 09:59:20 cool Feb 09 09:59:34 i should drink more tea and less coffee Feb 09 09:59:34 Bah, now I have to make tea .... Feb 09 09:59:39 btw. Feb 09 09:59:48 * rwhitby is configuring asterisk on an nas100d, in preparation for moko Feb 09 09:59:51 fluffs: i got contradicting infos about S3C2440 Feb 09 09:59:52 * fluffs has some nice arabic coffee Feb 09 09:59:57 mickeyl: hmm? Feb 09 09:59:57 * pH5 is into thyme, currently. damn cold. Feb 09 10:00:02 fluffs: max. freq is that 400 or 533 ? Feb 09 10:00:16 they say 533, never seen a 533 in the wild though Feb 09 10:00:23 all ours are 400 speed grade Feb 09 10:00:55 ah, i see. due to the increased bus speed it should be more than twice as fast as the 2410/200, shouldn't it? Feb 09 10:01:19 yeah Feb 09 10:01:31 though by the time v2 is done, there will be better things available Feb 09 10:01:38 ok, cool. any rough idea about increased power consumption compared to that one? Feb 09 10:01:55 fluffs: i know... rather thinking about a v1 upgrade atm. Feb 09 10:02:01 (just pie in the sky for now) Feb 09 10:02:12 not pin compatible Feb 09 10:02:30 sure, but fairly close Feb 09 10:02:36 mornin' Feb 09 10:02:37 afaict Feb 09 10:03:51 cheers Stephmw Feb 09 10:05:14 is V2 a replacement for V1 or in addition to it? Feb 09 10:05:24 a totally different beast Feb 09 10:05:28 so, addition Feb 09 10:05:35 well Feb 09 10:05:37 or replacement Feb 09 10:05:38 heh Feb 09 10:05:58 i guess the real question is, will V1 stop being produced once V2 is out Feb 09 10:05:59 will there be different versions of v2 ? Feb 09 10:06:21 sorry, no idea on both questions Feb 09 10:06:45 seriously v2 is in early planning stage and the bits that i know i'm not allowed to disclose, so please don't ask me :) Feb 09 10:06:57 fair enough :) Feb 09 10:07:45 v1 will keep me busy for long enough that I don't worry about anything else atm. :) Feb 09 10:07:45 heh Feb 09 10:08:46 mickeyl: I just hope that v1 will be used for atleast 6-12 months Feb 09 10:08:57 hrw: for sure Feb 09 10:09:27 mickeyl: thats enough for me. Feb 09 10:09:34 now week of waiting.... Feb 09 10:10:05 mickeyl: btw - I hope that they are in other version then white? I prefer black phones Feb 09 10:10:27 im thinking people will still be using V1 years after the gsm network is no longer being used. Probably just using it for some sort of tiny computer terminal or something :P Feb 09 10:10:53 there's two versions Feb 09 10:11:07 white/orange [sides], black/gray[sides] Feb 09 10:11:24 i don't think we phase 0 are allowed to chose though ;) Feb 09 10:11:27 mickeyl: what colour are the phase0 phones? Feb 09 10:11:30 * mickeyl has no case at all Feb 09 10:11:34 see above Feb 09 10:11:52 oh, it's just random? Feb 09 10:12:18 Psi_: I do not think so. I do not use my old palmtops Feb 09 10:12:40 true Feb 09 10:13:22 maybe for centamental reasons Feb 09 10:13:40 well, there are quite some things i can imagine for gsm-equipped gadgets at home Feb 09 10:13:43 home automation, for one Feb 09 10:14:10 by that time we will all have pico-cells in our houses anyway ... Feb 09 10:14:17 hehe Feb 09 10:14:19 hah Feb 09 10:14:24 v1 will get fixed to Roomba once I have v2 :-) Feb 09 10:14:46 v1 will become the console for the slug Feb 09 10:15:05 yeah Feb 09 10:15:10 rwhitby: ah yes, that should be a great partnership of devices ;-) Feb 09 10:15:35 the slug can keep the v1 charged through usb Feb 09 10:15:45 and the slug can run from 4 AA batteries. Feb 09 10:15:45 or an info-display in the living room. or an alarm clock. Feb 09 10:16:12 Ot just as a gps :) Feb 09 10:16:15 or Feb 09 10:16:18 Talking of charging, do Li-Ion batteries suffer from being on permanent charge, for non-mobile operation? Feb 09 10:16:19 stick a millimeter turbine in there and be done with it :P Feb 09 10:16:51 Psi_: I saw that one too. but 5-8% efficiency isn't much. Feb 09 10:17:08 or robot control Feb 09 10:17:11 And the hydrogen tank a bit scary Feb 09 10:17:16 people are using ipaqs and zaurus handheld for that atm. Feb 09 10:17:25 Bah, turbines. Zero-point energy please, none of that retro stuff. Feb 09 10:18:32 mickeyl, here is a question about the V1 that i had the other day. Is the low voltage shutdown done in hardware of software? Feb 09 10:20:24 hmm... i think there is both. we have a power control IC that is doing some things and we also have a kernel thread that's supposed to do a controlled shutdown on low power Feb 09 10:20:43 for more details you need to query mr hardware expert Feb 09 10:20:51 * mickeyl points to LaF0rge Feb 09 10:21:22 bbiab, need to work a bit Feb 09 10:21:28 i just dont want to accidently change that code and have it discharge the Lipo battery below its min volts and cause damage or small explosion :P Feb 09 10:21:43 I'm looking forward to the blogs of hardware peeps who take this thing apart. Lot of questions, like does the microphone have a hardware sound path to the GSM module, and can it be turned on from the GSM side. Feb 09 10:22:30 yes, no Feb 09 10:22:39 k Feb 09 10:22:47 gsm has no control over the audio paths Feb 09 10:22:53 or very little at least Feb 09 10:22:59 couldn't think of a non-invasive usecase for that 2nd one either ;-) Feb 09 10:23:05 one cool i can confirm is that LaF0rge made a phone call yesterday night where he routed an mp3 into gsm :) Feb 09 10:23:11 If there already is an audio path in hardware, all it needs is the ability to turn mike on. Feb 09 10:24:26 Stephmw: oh absolutely. This is about being sure that there are no invasive paths! One of the issues that arose in the CCC conference about closed mobiles. Feb 09 10:24:37 * fluffs is looking forward to reviewing the kernel code Feb 09 10:25:44 mickey|powerpoin: 'please hold on, I will play you some music' mode for annoying telemarketers? Feb 09 10:25:53 Hehehe Feb 09 10:26:11 and then switch to hardcore noice generator Feb 09 10:26:26 We need an official selection of most annoying music to play on hold. Feb 09 10:26:42 anything by the cheeky girls Feb 09 10:28:15 Does anyone know whether Leigh Honeywell or Paul Wouters are among the Phase 0 people? From that CCC conference video, they would make terrific ambassadors for the device, and they said hopeful things about it in their presentation. Feb 09 10:28:41 mickeyl: cool indeed Feb 09 10:31:21 They ended up doing their mobile phone call encryption / VPN stuff using that closed (prolly undisclosed Linux inside) HIPI phone, which was a shame. Feb 09 10:33:30 mickey|powerpoin: use '|slides' ;-) Feb 09 10:33:52 *nod* ;) Feb 09 10:50:19 If HTC still has strong links to FIC, then I guess v2's keyboard will be along these lines: Feb 09 10:50:23 http://www.msmobilenews.com/content/view/1579/ Feb 09 10:50:51 will there be a keyboard? Feb 09 10:51:19 It was rumoured as a possible option.s But hey, even v1 is just a rumour, lol Feb 09 10:51:29 Everything is speculation :P Feb 09 10:53:38 v3 should work also as an emergency floatation device Feb 09 10:53:43 If it doesn't have a keyboard though, then FIC should produce a credit-card size Bluetooth keyboard as an accessory, to keep in shirt pocket. There's always more profit on accessories anyway, so would make business sense. Feb 09 10:54:15 But slide-out would be nicer. Feb 09 10:56:09 The excellent keyboard is the main reason why I still use my very old Nokia Communicator. Feb 09 10:56:19 Morgreet: slide-out would make the device more breakable :( Feb 09 10:58:09 ecraven: don't think so, there's not all that much leverage between the two sections. A clamshell is theoretically even more breakable, but in practice it's robust. So just depends on construction. Feb 09 11:06:09 If v1 development is indeed going to be recouped by sales of same h/w as WinMobiles, then I understand why the external design is really a non-geek one. But if v1 is a success, then I hope v2 h/w is designed with a bit more geek-appeal, beyond openness. With a full Linux inside, a keyboard is extremely powerful. Feb 09 11:06:48 the true measure will be when this phone replaces my Treo 650. Feb 09 11:07:15 you can use keyboard via BT Feb 09 11:07:43 Don't actually know Treo 650 specs. Looking it up. Feb 09 11:07:43 and i want phone as small as possible Feb 09 11:08:53 rwhitby: nice Feb 09 11:09:20 Well if a qwerty keyboard fits on the Treo 650, no reason why it shouldn'd on v2 Feb 09 11:09:24 Morgreet: my treo does bluetooth, ssh, imap/ssl with idle and offline, irc, podcasts, mp3s, blogging, transport schedules, spreadsheets, powerpoint, pdfs, ... Feb 09 11:09:46 oh, and it's phone too :-) Feb 09 11:09:47 rwhitby: hehe, nice Feb 09 11:10:29 at least 80% of my email and irc logs reading is done on the 320x320 treo screen. Feb 09 11:13:14 Well here's the wishlist I've been gathering from people's more serious comments (:-) --> Feb 09 11:13:22 wifi with OLPC mesh 3 accelerometers or 3 chip gyros camera Powered USB 3G or EDGE Axim-style 3D acceleration Hall device compass analog thumbstick many more buttons / qwerty keyboard OLED display with buttons on top credit-card size Bluetooth keyboard (accessory) SAW touchscreen Feb 09 11:14:02 Lemme stick commas between those things with sed Feb 09 11:14:25 toster, coffee machine, alcomat, wings, turbojet engine Feb 09 11:14:37 I said serious :P Feb 09 11:14:40 , kitchen sink Feb 09 11:14:47 madwoota: so emacs? Feb 09 11:14:51 wifi with OLPC mesh, 3 accelerometers or 3 chip gyros, camera, Powered USB, 3G or EDGE, Axim-style 3D acceleration, Hall device compass, analog thumbstick, many more buttons / qwerty keyboard, OLED display with buttons on top, credit-card size Bluetooth keyboard (accessory), SAW touchscreen Feb 09 11:15:23 i dont see the point in using emacs, its far too small, nimble and lightweight Feb 09 11:15:35 Hi Morgreet! Feb 09 11:15:36 Kitchen sink is already in there, since emacs is in OE Feb 09 11:15:46 Hi El :-) Feb 09 11:16:07 So OM/Neo can cook coffee? ;) Feb 09 11:16:16 add firefox and you can do about:kitchensink Feb 09 11:16:22 hehe :) Feb 09 11:16:32 What about xemacs? That was definitely the number 99% cpu sucker at my universiy Feb 09 11:16:40 I'll be adding a BlueSMIRF to my espresso machine for Neo control :P Feb 09 11:17:52 BlueSMIRF? Feb 09 11:17:54 Of course, any espresso app for the Neo must transmit the slurping sounds back to the device, since this is the main attraction of an espresso machine ;-))) Feb 09 11:18:17 nah, the grinding - the smell Feb 09 11:18:18 Morgreet: on serious note, and external battery adaptor would be nice for those actually using the GPS outdoors without chance to recharge it every X hours. would be nice to have power to navigate for couple days Feb 09 11:18:44 Elrond: one of several OEM Bluetooth devices for embedding into hardware. Feb 09 11:18:57 raynet: yes Feb 09 11:19:45 raynet: perhaps we need an OLPC-type yoyo or handcrank charger. Feb 09 11:19:50 Morgreet - Ahh. Is it open? Feb 09 11:20:00 when using my Nokia 6670 + Navicore + Bluetooth GPS the GPS will go for 10-20 hours but Nokia dies in 3 Feb 09 11:20:43 counter Feb 09 11:20:43 2 days 18:06:43 (2.755 days) for source for *all* developers and devices for selected developers (2007-02-11); a month, 2 days (30.755 days) for devices for *ANYBODY* for $350 (targeting developers) (preorders: not yet) (2007-03-11); 7 months, 2 days (214.755 days) for mass market (2007-09-11): see topic for more info (186) Feb 09 11:20:52 raynet: not surprising IMO Feb 09 11:20:55 Do we have any info on battery live? Feb 09 11:21:01 raynet: compare battery capacities and consumption Feb 09 11:21:07 ... for neo v1, that is. Feb 09 11:21:21 Elrond: BlueSMIRF -- http://www.sparkfun.com/commerce/product_info.php?products_id=582 Feb 09 11:22:19 Stephmw: true, which is why external battery would be nice. My Psion Series 7 (with 133MHz ARM) gives me 13 hours of navigation time, but ofcourse it has a bigger battery Feb 09 11:23:21 Morgreet - Is sparkfun.com a good dealer? (A friend ordered there a few days ago) Feb 09 11:24:17 Presumably any external battery pack with the right voltage will charge the Neo fine, so it just needs messing with NiMH cells and USB cables. Feb 09 11:24:29 hoo.. first pics of Nokia's E90 Communicator Feb 09 11:26:42 Elrond: I don't know, Sparkfun are in the US, and I don't even know a distributor here in the UK. Just came across them when thinking up things to do with my Roomba. Feb 09 11:28:22 Morgreet - Ahh, okay. Feb 09 11:34:03 LOL, typical ... "Communicator size/shape as annoying as previous designs". Why are reviewers suck morons always? Each size/shape has its uses. Just because a given one doesn't suit your own use, it doesn't make it bad. Feb 09 11:35:14 Yes, an HGV truck is really annoying when used for shopping downtown. Doh. Feb 09 11:36:13 Morgreet - You prefer some Star "Trek Enterprise"-level thing for shopping downtown, right? ;) Feb 09 11:36:30 s/Star // Feb 09 11:36:30 Elrond meant: Morgreet - You prefer some "Trek Enterprise"-level thing for shopping downtown, right? ;) Feb 09 11:36:38 Nokia needs a new term to describe these decent-keyboard sized mobile devices. Calling them a phone just triggers the moron reviewers into their negativity. Feb 09 11:36:48 *gna* I should not use s/.../.../ here. It always goes wrong. ;) Feb 09 11:37:18 Morgreet: a communicator? Feb 09 11:37:32 radical, i know Feb 09 11:37:35 Hehe, it seems that label didn't work ;-) Feb 09 11:38:00 Elrond: I've not been satisfied with teleporters so far. They always leave some vital piece of me behind. Feb 09 11:38:04 Anyway, I'm out for a few hours, "legacy work" ;) Feb 09 11:38:20 Morgreet - *LOL* Feb 09 11:38:32 cya later El Feb 09 11:38:50 with the n-series i think they talk about multimedia computers Feb 09 11:39:09 somehow i'm guessing that that label won't be a smash hit either Feb 09 11:40:16 Aye Feb 09 11:41:05 but there's clearly a need for more vocabulary on these things Feb 09 11:42:15 communicator wouldn't actually be a bad label per se Feb 09 11:42:32 Yeah. And you just can't foretell which terms will click with people. Eg. "PDA" is a completely hopeless term when you think of it for general use, neither descriptive nor marketting friendly, yet it caught on. Go figure. Feb 09 11:49:32 Does anyone know whether the OLPC's dual-mode LCD display is going to be marketted outside of that project? I would love a very low power HQ monochrome display for ebook reading, overlayed on the normal power-hungry colour LCD. Feb 09 11:49:46 You could read ebooks while the whole rest of device is on standby. Feb 09 11:50:39 the colour on that screen is quite low resolution Feb 09 11:51:12 Matches our eyes --- we have much lower optical resolution for chroma than for luminance. Feb 09 11:52:33 that's a relative, not absolute, thing. Feb 09 11:53:40 That's why we've been happy with extremely smudgy colour res on TVs for many decades. As long as the luminance info is sharp, the eye/brain is perfectly happy with bleeding colours, it moves them into the right place. Feb 09 11:54:00 No, it's absolute. The chroma vs luminance resolutions can be measured. Feb 09 11:54:17 you're misunderstanding. Feb 09 11:54:21 "low resolution" isn't absolute Feb 09 11:54:34 the human eye is capable of vastly higher resolution than the colour on the olpc screen Feb 09 11:54:50 "is absolute".. even Feb 09 11:56:34 Well, resolution is an angular thing, depends on dot pitch and distance from eye. Once you nail those down, resolution at the eye is a very absolute thing. Feb 09 11:57:49 In any case, what matters here is the OLPC display's tech, not its size on that machine, since it would be a different physical display for a phone/ebook-reader. Feb 09 11:58:03 jesus wept. Feb 09 11:58:21 I was just wondering if they were marketting their tech outside OLPC devices. Feb 09 12:03:59 http://en.wikipedia.org/wiki/One_Laptop_per_Child#Display Feb 09 12:04:07 "Negroponte has said at the Technology Review\u2019s Fifth Annual Emerging Technologies Conference that the monochrome display has four times the resolution of the color display." Feb 09 12:05:15 yes, but dude, its monochrome! Feb 09 12:05:19 i cant wait till we start to see color e-paper rollup displays :) Feb 09 12:05:46 or even just color rollup displays of any type Feb 09 12:05:51 Yeah, the monochrome mode is mainly for reading ebooks. Purely reflective in that mode. Feb 09 12:08:46 1200x900 in that mode, one of the reasons why so many people in the west would like to buy XOs at twice or even 3 times the price. Feb 09 12:10:36 Psi_: me too. But I think E-Ink will die, simply because their licensing greed is stopping widespread adoption. They badly need a competitor. Feb 09 12:14:19 OLPC News (2007-01-20): "Kernel: The wireless driver has gone through two rounds of reviews Feb 09 12:14:19 with the upstream kernel networking folks and work continues to get it Feb 09 12:14:19 into the Linux mainline kernel. Marcelo Tosatti also reports that we Feb 09 12:14:23 are down to 5\u20136 interrupts per second (minus a really bad i8042 Feb 09 12:14:23 driver). This is compared with a ~200 per second in a standard Linux Feb 09 12:14:23 desktop. This will have a huge positive effect on our battery life." Feb 09 12:14:37 LOL, article formatting :-( Feb 09 12:15:14 5\u20136 interrupts ? Feb 09 12:15:16 That's "5-6 interrupts per second" Feb 09 12:15:20 ah Feb 09 12:15:26 Yeah, crappy web page, sorry Feb 09 12:15:34 http://laptop.media.mit.edu/laptopnews.nsf/2e76a5a80bc36cbf85256cd700545fa5/482b4202b90899b0852572690063e1bb?OpenDocument Feb 09 12:17:04 Really looking forward to that driver. Hopefully some other device will feature that same ATI wifi/mesh chip, and we can drive it. Feb 09 12:22:48 E-ink is annoying. Feb 09 12:23:01 I was trying to get a sample of film, to do a _cheap_ display with. Feb 09 12:23:02 "Network testing: James Cameron has been an immense aid at early testing of the OLPC system in radio quiet areas (he lives in the Australian outback). Two of our machines have been able to talk to each other over 1.3km apart. Feb 09 12:23:02 " Feb 09 12:23:28 SpeedEvil: they gave you the cold shoulder? Feb 09 12:23:34 Basically an array of electrodes like an electrostatic printer. Feb 09 12:23:43 And you drag the film out over it. Feb 09 12:23:58 And they were basically completely uninterested. Feb 09 12:24:40 Yeah, you're not the first to say it. The main barrier to E-Ink (product) is E-Ink (company). Feb 09 12:24:47 Advantages are that the length is unconstrained, and the film can be easily replaced. Feb 09 12:24:51 And cost. Feb 09 12:25:13 Disadvantage is of course it is literally e-paper, which doesn't change until you move the printer over it. Feb 09 12:26:34 The cost should be near zero once plant costs are recouped, since it's sort of printable by the acre. :-) In practice, E-Ink greed will ensure it never becomes common, at a guess. Feb 09 12:27:29 Cost is far from zero. Feb 09 12:27:43 Driving it like a 'proper' display is quite complex. Feb 09 12:27:54 it'll become common once their patent expires :) Feb 09 12:29:58 Yeah, the drivers are the costly part, not "printable" by current tech. But every few months someone announces progress on printing circuits, so it'll happen one day. Feb 09 12:30:05 Indeed. Feb 09 12:30:32 It's a pity the tech does not lend itself to simple brightly coloured displays. Feb 09 12:30:38 You've gotta have filters. Feb 09 12:31:14 Not as bad as LCD, where at best a reflective display gets 1/6th of the light back. Feb 09 12:31:39 (1/2 lost in polariser, 2/3 lost by light hitting the wrong colour pixel) Feb 09 12:32:12 SpeedEvil: someone'll come out with 5-state particles for the microbeads (black, white, and RGB), and then it'll be a colour tech without filters. Feb 09 12:32:29 That's got real problems. Feb 09 12:32:42 Engineers love problems ;-) Feb 09 12:32:42 The 2-state is simple, it's repel or attract the black bit. Feb 09 12:33:02 And even with that, there is a problem with remenant image. Feb 09 12:33:28 Making a 3-state (you don't want to go more otherwise you get viewing angle problems is much harder. Feb 09 12:36:27 As you can only have the electrostatic field go one way. Feb 09 12:36:42 Adding sensors changes things - but makes it even more complex. Feb 09 12:40:03 I wonder if MEMS people are working on mechanical pixels, sort of like a colour wheel turned inside out, ie. a rotor in each pixel. Lots of possible ways forward with that, but for basic ebook display it should be fairly straightforward, assuming (big assumption) some non-warp drive means of manufacture. Feb 09 12:40:15 Ie. I mean pre-nanotech ;-) Feb 09 12:41:32 Ie. it's macroscale assembly, not nano, and not atomically precise. Feb 09 12:42:48 That may well be true, but without nanotech, it's going to be prohibitively expensive. Feb 09 12:43:22 The ebooks get round that by having really simple bi-coloured (black/white) beads, with differeent charges on the black and white side. Feb 09 12:43:36 Just googling for MEMS + display, there's quite a lot of stuff going on! Feb 09 12:43:44 These are free to move in little bubbles of oil, and when you put + against one side, they go to that side. Feb 09 12:43:57 Yeah - MEMS is largely irrelevant for big displays though. Feb 09 12:44:10 They're going for TB market :P Feb 09 12:44:13 TV* Feb 09 12:44:24 I mean laptop and PDA sized. Feb 09 12:44:31 Projection doesn't work so well at that scale. Feb 09 12:47:53 hello! Is there somebody Feb 09 12:47:58 I'll add "3D display" to the v2 wishlist .... ;-))) Feb 09 12:48:47 Does somebody know, when the neo1973 is gonna be shipped? Feb 09 12:48:55 counter Feb 09 12:48:55 2 days 16:38:30 (2.693 days) for source for *all* developers and devices for selected developers (2007-02-11); a month, 2 days (30.693 days) for devices for *ANYBODY* for $350 (targeting developers) (preorders: not yet) (2007-03-11); 7 months, 2 days (214.693 days) for mass market (2007-09-11): see topic for more info (187) Feb 09 12:53:36 thanks a lot. Go on, cu! Feb 09 12:53:46 /exit Feb 09 12:53:47 cyu Feb 09 12:57:08 http://www.0xdeadbeef.com/weblog/?p=231 Feb 09 12:57:14 OMG FPGA!!!! ;-))) Feb 09 13:01:23 * Morgreet wonders why "just stunning" displays always look poor in photographs. Feb 09 13:01:53 * SpeedEvil waits eagerly to find out that P0 devices really do have camera and wifi, and it was just an early april fool. Feb 09 13:02:10 Hehe Feb 09 13:03:00 Cripes, I hope the concept of "early april fool" never takes off .... It's bad enough just 3 days each side of 1st Apr :-( Feb 09 13:04:09 OTOH, I guess marketting uses the concept of april fool all year 'round ... Feb 09 13:05:16 cu Feb 09 13:06:43 If only they'd do nice surprises too though. Feb 09 13:07:14 'Yes, this feature will be present. Yes it was advertised, and you bought it because of that, no we don't have a timescale' Feb 09 13:07:26 SpeedEvil: I tought the whole point of marketing was to do nice surprises Feb 09 13:08:15 SpeedEvil: especially for the engineers Feb 09 13:08:15 'We did too put a disclaimer on it, it's clearly marked on the bottom, in 1pt salmon on pink text' Feb 09 13:08:48 jannu: nah, the whole point of marketting is to provide employment for the non-productive. Feb 09 13:09:06 Morgreet: so why am i not employed then? Feb 09 13:09:24 jannu: lazy? ;-))))) Feb 09 13:09:33 Morgreet: never! Feb 09 13:09:36 Hehe Feb 09 13:09:47 Morgreet: which is marketing speak for always Feb 09 13:11:53 In a sane world, products would simply get added to a global products tree as they are released, with full machine-searchable descriptions and drill-down data, nothing hidden. And you could find what you want with SQL or grep. :-) Feb 09 13:12:08 No need for a larketting middleman layer. Feb 09 13:12:13 marketting* Feb 09 13:14:11 Marketting is just a distortion -- the products with the best marketting make it, rather than the best products. Feb 09 13:14:28 * CM steals on t from Morgreet Feb 09 13:14:40 one.. Feb 09 13:15:20 Leave my tea alone ;-) Feb 09 13:15:28 Hehe Feb 09 13:16:16 Morgreet: you need to have a lot of people working so you have a lot of consumers with money handy Feb 09 13:16:35 it goes round and round :) Feb 09 13:17:10 Eblis: Just like with communism you needed a lot of bureaucracy and three stamps on each application Feb 09 13:17:42 hehe Feb 09 13:17:57 Eblis: the only difference is that in capitalism the paper most people spend their lives moving around is money Feb 09 13:18:17 Oh no, not the C word ... it stops all rational discussion from our US friends ;-) Feb 09 13:18:46 Like the N word in Europe ;-) Feb 09 13:18:48 getting in debt in the first years of employement and then working untill 60 years old to pay that debt Feb 09 13:19:26 xkr47: As a by-product of the libsydney discussion Jean-Marc coded a fast C resampling library supporting both floating point and fixed point and being licensed under BSD. Feb 09 13:19:29 woo Feb 09 13:20:05 * CM woo's too ;) Feb 09 13:21:14 Eblis: and spending most of the money you make on things you don't need in order to give the money to other people who... Feb 09 13:21:37 yep, an infinite circle :) Feb 09 13:21:54 but we do need the OpenMoko :D Feb 09 13:22:23 Eblis: and the great thing is that people like that keep the society rolling in such a way that smarter people can get by with a lot less work Feb 09 13:23:00 It's a pity anonymous E-cash is illegal generally. Feb 09 13:23:46 I see openmoko as a way to bring some sanity into the mobile phone business Feb 09 13:24:40 LOL, I think you underestimate the current level of insanity :P Feb 09 13:25:18 might be Feb 09 13:25:32 i like the idea behind it, it needed to happen Feb 09 13:25:53 yes Feb 09 13:25:57 I mean, just think about the hardware in samsung f700 Feb 09 13:26:26 There are _beautiful_ phones out there. Feb 09 13:26:34 With wonderful hardware, in tiny packages. Feb 09 13:26:49 That are almost utterly useless, compared to what they could be. Feb 09 13:26:55 exactly. and there's basically no limit to what you could do with those Feb 09 13:27:04 with open specs and open software Feb 09 13:27:48 I had hopes for the Motorola A1200 - but they've done essentially the most annoying thing. Used the linux kernel for its fun stuff, to ease their development, and released the source for it like good boys. Feb 09 13:28:02 But not actually given any way to run programs on the phone. Feb 09 13:28:16 e.g. these four videos by nokia: http://youtube.com/watch?v=beiMw1GkAQ8 http://youtube.com/watch?v=f5iBBEp0Efg http://youtube.com/watch?v=JJ4afZpQAvI http://youtube.com/watch?v=9qIZBo8a_pA Feb 09 13:28:21 It is a pretty, small, featurefull phone. Feb 09 13:28:29 Can't do youtube ATM. Feb 09 13:28:43 most of the stuff in those is completely doable in current hardware Feb 09 13:29:01 the problem is that i'll have to start saving a few good months if i want to buy the phone :( Feb 09 13:29:01 Well the good news is, as long as the FOSS community remains strong on licensing, at some point manufacturers will simply not be able to compete with closed phones, because they'll have to bear the extra cost of software development. So, time is on our side, in due course we will have all the top phones. Feb 09 13:29:29 It's not quite that simple. Feb 09 13:29:43 The networks have a big impact, with subsidies. Feb 09 13:29:52 SpeedEvil: about mesh: imagine earth covered with 1mx1m nodes: so there are about 20 million hops from one side of earth to other side Feb 09 13:30:11 SpeedEvil: all the networks care about is some sort of standard platform Feb 09 13:30:30 SpeedEvil: they'd really prefer to not be tied to certain manufacturers Feb 09 13:30:31 SpeedEvil: yep. there's quite a bit of inertia in the mobile industry Feb 09 13:30:37 SpeedEvil: however, you could form 'channel' line between nodes at opposite sides of earth would communicate with each other and nodes between endpoints just pass traffic Feb 09 13:30:43 And branding, and locking some features out on 'cheap' contents. Feb 09 13:30:49 SpeedEvil: yep Feb 09 13:30:59 SpeedEvil: OK, spread the cost of closed phone development across manufacturers and networks. The point still applies. As time goes on, they will because more and more disadvantaged vs outfits without software costs. Feb 09 13:31:06 SpeedEvil: though the latter is really an american abberration Feb 09 13:31:20 SpeedEvil: so I think you can get 'backbone' type stuff too with suitable configuration Feb 09 13:31:23 aloril: right - however, you're also interfering with adjacent nodes along all the length of that path. Feb 09 13:31:44 robtaylor: it's actually pretty bad in europe as well Feb 09 13:32:04 SpeedEvil: depends on power level, if communication is over 1mm distance, then interfere should not be big ;-) Feb 09 13:32:14 jannu: it is? its ok in UK, afaict Feb 09 13:32:24 Also, you've got the problem that there are natural bottlenecks. Feb 09 13:32:26 especially if its optical or even beyond it Feb 09 13:32:50 robtaylor: in carrier r&d the most important question has always been how the service can be billed like good old fixed line phone calls Feb 09 13:33:04 Consider oceans for example, people in oceans will be rare, and very, very desired paths. Feb 09 13:33:10 Which will be utterly congested. Feb 09 13:33:24 SpeedEvil: well... I assumed whole earth being covered ;-) Feb 09 13:33:34 You really do need overlaid long-haul networks on different frequencies. Feb 09 13:33:48 Unless the links are true point-point. Feb 09 13:33:51 robtaylor: it's not awful here in finland either. it's just 15 years behind the rest of technology Feb 09 13:34:09 don't think so, you can build long-haul type stuff with mesh network too if its configured right Feb 09 13:34:13 robtaylor: actually make that 10 years Feb 09 13:34:38 jannu: but the operatorots don't lock out functionailty, they just dont provide some network services Feb 09 13:34:48 Consider that if you are going through 20000 nodes to talk to someone, then 20000 people will be going through your node too. Feb 09 13:34:52 jannu: thats a bit differet to cingular in the states Feb 09 13:34:56 robtaylor: yep. that's true. Feb 09 13:35:31 ah, but assumed that those 20 million nodes were doing nothing except passing traffic, ie acting as back-haul Feb 09 13:35:32 robtaylor: so we're in the '90s and the u.s. is in the dark age Feb 09 13:36:05 Right - and that means - if they are long haul - that you'll be directly sharing that path with many, many others. Feb 09 13:36:08 Back. Feb 09 13:36:24 e.g. orange/francetelecom have been doing a lot to promote linux on phones Feb 09 13:36:50 what is difference to dedicated long haul? Feb 09 13:37:06 robtaylor: yes, but linux in and of itself doesn't really help Feb 09 13:37:11 s/back-haul/long-haul/ Feb 09 13:37:27 robtaylor: if the hardware specs and the ui code is all closed Feb 09 13:37:29 jannu: with the US in the dark ages, and the UK in 1984, we're in a right mess :-( Feb 09 13:37:47 SpeedEvil: of course I agree that its not feasible for every node to talk with every node, but I think this is problem with special build long haul too Feb 09 13:37:52 At least in the UK, buying SIM-only and sim-free phones is easy. Feb 09 13:38:17 The reason you need a seperate network is to seperate local traffic from long-distance. Feb 09 13:38:29 So you don't get radio interference between the two. Feb 09 13:38:48 We'll have to see what the UK police state does about open phones though. It's OK for now, but I'm not optimistic about the future. Feb 09 13:39:04 isn't that just our primitive antennae technology? Feb 09 13:39:23 without a truly open stack a linux phone is just as useless as a totally closed phone Feb 09 13:39:25 Morgreet: we're all in 1984 ;) Feb 09 13:39:31 aloril: no - there are fundamental limits - if you can't decode a distant signal, it appears as noise. Feb 09 13:39:52 jannu: how do you figure that? If you can use the hardware from your own programs. Feb 09 13:39:59 robtaylor: it's true, that 1984 is definitely consolidating, worldwide. Feb 09 13:40:11 Without any limits for 99.9% of users. Feb 09 13:40:18 SpeedEvil: sure, but there are 2 parts: sending and receiving: both need to be directed, not only receiving one Feb 09 13:40:20 SpeedEvil: yep. but if the whole userspace side is closed source Feb 09 13:40:38 jannu: that's true, but all we need to do is help form a market where its just too costly and TTM is too high if you don't use open components ;) Feb 09 13:40:57 aloril: Directivity of an antenna is at best 1.22* wavelength / diameter of antenna. Feb 09 13:41:00 robtaylor: yep. just like on the pc side Feb 09 13:41:10 So there are hard physical limits. Feb 09 13:42:07 robtaylor: seconded. The best way of fighting restrictive markets is to make the profits from open tech larger. Then greed will do the rest. Feb 09 13:42:18 SpeedEvil: OK, so you just move to higher and higher frequency then ;-) Feb 09 13:42:48 Morgreet: you could put a more positive spin on it, 1984 is only needed when people are completely uncontrollable by the elite.. Feb 09 13:43:16 robtaylor: just thinking what i do with my pc now and compare that to any games console Feb 09 13:43:42 robtaylor: ah, but that then begs the question of who the elite are, and why they want control. It's simpler to just assume that money talks. Feb 09 13:43:50 robtaylor: that's the sort of difference i believe openmoko will have on phones Feb 09 13:43:52 On the UK 'police state' - looking at the mobile phone laws. It's illegal to change IMEI if you don't have the makers permission. It's illegal to sell equipment that can change the IMEI - including phones - if you believe it's being used for criminality. Unfortunately, if you can change IMEI, aboyut the fifth person to buy it is going to try cloning someones phone. There don't seem to be any direct regulations other than changing the IMEI. Feb 09 13:44:24 Morgreet, jannu: nod, FLOSS wins in teh end in the market place cos its more efficient. Feb 09 13:45:03 I can see competition from makers that make their phones more open. Feb 09 13:45:09 Even if not FLOSS. Feb 09 13:45:21 SpeedEvil: there are a lot of legitimate restrictions on mobile phones and IMHO that's where closed source software has an advantage Feb 09 13:45:25 Morgreet: well, the elite are simply the people who poulate the upper regions of our power structures. nothing to controversial there =) Feb 09 13:45:46 But it would have to be a _large_ platform to compete. Feb 09 13:45:58 SpeedEvil: remember that the network is used e.g. for distress calls Feb 09 13:46:24 jannu: i don't see how thats an argument against floss Feb 09 13:46:40 If several phone makers come out in the next couple of months, with a compatible platform that lets programmers access camera/display/keyboard/... then that could take away from momentum. Feb 09 13:46:42 jannu: ok, i do accept the problems with an openly licensed GSM stack, however Feb 09 13:47:16 Especially if it's on millions of phones. Feb 09 13:47:27 SpeedEvil: well, luckily that wont happen.. Feb 09 13:47:34 Probably not. Feb 09 13:47:49 robtaylor: not against floss per se, but one needs to acknowledge the fact that possible dirsuption in mobile networks is a serious problem Feb 09 13:47:50 No controversy because you're speaking to the converted. :-) The man in the street seems to take the media and politics at face value though and sees public/consumer control as the talk of cranks. Feb 09 13:48:03 robtaylor: which will cost lives Feb 09 13:48:24 SpeedEvil: the reason everyone (and i really do mean everyone, can't think of any manufacturer that isn't looking at linux) is looking at linux is that they can't afford to build a whole new platform themselves Feb 09 13:48:46 SpeedEvil: the commodity level is too high Feb 09 13:48:49 Yeah - I know. Especially with ever decreasing spin times. Feb 09 13:48:58 SpeedEvil: yeah :) Feb 09 13:48:58 Aye, money, money, money. Software development is extremely expensive. Feb 09 13:49:20 which is why imho keeping the gsm side closed is actually a pretty good idea Feb 09 13:49:21 jannu: well, thats the problem i was talking about with an open GSM stack Feb 09 13:49:30 robtaylor: exactly Feb 09 13:49:39 jannu: allowing users to modify their gsm stack would be baaad Feb 09 13:50:04 A open source GSM stack, with signed code, and the ability for the maker to allow features to be added might be good. Feb 09 13:50:17 jannu: though there's nothig to say you couldn't do a BSD license GSM stack. but there's really no need for that. The current GSM stacks are very stable Feb 09 13:50:25 robtaylor: obviously the mobile network could be designed to withstand abuse Feb 09 13:50:30 jannu: i guess itgets interesting when we start talking about 4g Feb 09 13:50:40 robtaylor: like the internet basically is Feb 09 13:50:46 The mobile network cannot determine between abusing phones though. Feb 09 13:50:53 robtaylor: true. very interesting Feb 09 13:51:10 Or even a non-abusing phone, and a cloned phone. Feb 09 13:51:18 SpeedEvil: yep. because the current standards don't design for it Feb 09 13:51:27 Not so much. Feb 09 13:51:49 How do you - with two identical transmitters, that don't want to tell you their ID, determine between them? Feb 09 13:52:14 SpeedEvil: well. you don't Feb 09 13:52:46 SpeedEvil: unless you have an id component which can't be spoofed (the current approach) Feb 09 13:52:52 Yeah. Feb 09 13:53:12 SpeedEvil: or rely on secondary identity measures (the internet approach) Feb 09 13:53:50 zero knowledge protocol ? http://en.wikipedia.org/wiki/Zero-knowledge_proof Feb 09 13:53:50 GSM is a component, and its comms stack should be nowhere near any host CPU. We have nothing but contempt for Windows GDI printers in which half the printer logic runs in the host ... well it should be exactly the same for GSM modules. Feb 09 13:53:57 Secondary measures are not implied at the charging time. Feb 09 13:54:09 They use primary measures - what line you are connecting to , ... Feb 09 13:54:48 ~fish lrg Feb 09 13:54:56 SpeedEvil: exactly. the current standard has bee designed with billing as a main priority Feb 09 13:55:00 * apt slaps lrg around with a large trout Feb 09 13:55:06 I see absolutely nothing wrong with putting all the smarts you can in the CPU, as long as it does't impact on battery life. Feb 09 13:55:25 Because spectrum is scarce. Feb 09 13:55:29 counter Feb 09 13:55:29 2 days 15:31:55 (2.647 days) for source for *all* developers and devices for selected developers (2007-02-11); a month, 2 days (30.647 days) for devices for *ANYBODY* for $350 (targeting developers) (preorders: not yet) (2007-03-11); 7 months, 2 days (214.647 days) for mass market (2007-09-11): see topic for more info (188) Feb 09 13:55:32 And relatively costly. Feb 09 13:55:36 SpeedEvil: you've contradicted yourself in that single line. Feb 09 13:55:39 SpeedEvil: i.e. "how do we get their money?" as opposed to the internet's "how do we make communication possible?" Feb 09 13:55:44 Not quite. Feb 09 13:55:52 Morgreet: which line? Feb 09 13:56:00 there could be uses for modified distress call in some situations: for example there might be somebody else closer than official ambulance that could also help Feb 09 13:56:21 jannu: there is a difference between 'the internet' and means of accessing the internet. Those are almost always charged. Feb 09 13:56:26 or maybe in case somebody is stealing you people around could start tracking thief Feb 09 13:56:28 etc.. Feb 09 13:58:08 * SpeedEvil is in your country, stealing your people. Feb 09 13:58:29 however I think both are achievable even with keeping current mode as long as you can talk and have internet connection going same time Feb 09 13:59:06 SpeedEvil: re battery life. An embedded, GSM-specific SoC can have VASTLY less power consumption than any generic host, because it need only come into play when the protocol demands it. The fact that one doesn't want the comms layer compromised makes the whole point moot anyway. It has no business running on the host CPU. Feb 09 13:59:07 SpeedEvil: just like there's a difference between the mobile network and accessing the mobile network Feb 09 14:00:14 Morgreet: http://gizmodo.com/gadgets/gadgets/ultraslim-credit-cardsized-bluetooth-keyboard-235252.php Feb 09 14:00:16 Morgreet: I agree - in this case. A GSM stack it makes sense - unless the host CPU can be woken up and turned off _very_ fast while preserving state. Feb 09 14:00:33 Was that was you talked about this morning? ;) Feb 09 14:00:51 In case the v2 didn't have a slide-out keyboard Feb 09 14:00:52 jannu: but access to that network has to be billable - or otherwise priced in some manner, because in places it is very, very congested - the access points. Feb 09 14:01:22 SpeedEvil: what I'm saying is that the main difference is that due to the desire to design around billing there are practically no advanced services within the mobile network Feb 09 14:01:25 There is lots of (generally, not always) spare capacity around that netowek. Feb 09 14:01:37 Sure, I agree with that. Feb 09 14:01:38 CM: I didn't mention that one today, but yes, that's the well-known Freedom keyboard. Feb 09 14:02:00 Ah, ok. Looks nice Feb 09 14:02:31 SpeedEvil: which is great now, because we only have one global network Feb 09 14:03:17 SpeedEvil: if mobile companies would have gotten it in early '90s we might still have no open gprs internet gateways Feb 09 14:03:48 SpeedEvil: because they'd be promoting their own network Feb 09 14:04:16 SpeedEvil: now, because their network sucks, they have to connect it with the internet Feb 09 14:04:44 It's just you have to have some means of charging or rationing at the comnnection node. Feb 09 14:04:44 Otehrwise you'll get people wandering around makdoing P2P stuff, and emergency calls not getting throughl. Feb 09 14:05:01 (wifi went down for a mo) Feb 09 14:05:42 SpeedEvil: yep. charging for connection is more or less the same on both networks Feb 09 14:06:25 Logical lowest-cost routing (from the network providers point of view) would be good. Feb 09 14:06:36 SpeedEvil: but you also need to have restrictions so that the emergency bandwidth can't be used for arbitrary stuff Feb 09 14:07:32 * SpeedEvil leaves to do other stuff. Feb 09 14:07:55 Wow, big split. Feb 09 14:08:06 cya Speed Feb 09 14:15:23 Morgreet: I should have linked to the mobility side instead Feb 09 14:15:31 Looks like a good keyboard :) Feb 09 14:16:43 I'll buy some sort of CC-side keyboard when I get the Neo. I've got too used to the Nokia Communicator, lack of a keyboard is just slumming it for me. Feb 09 14:16:52 CC-size* Feb 09 14:17:16 I'd like to try something like hexinput first Feb 09 14:17:29 Or I'll just get something like that small bt keyboard Feb 09 14:17:54 hexinput? Feb 09 14:18:18 SpeedEvil: http://www.strout.net/info/ideas/hexinput.html Feb 09 14:19:09 Don't know if it's good or so, but I like the hex-shape ;) Feb 09 14:19:16 CM: i kinda like the idea Feb 09 14:19:17 Ah. you mean divide up into 16 Interesting. Feb 09 14:19:43 It's like a mix between stroke and tap Feb 09 14:20:18 And that quong layout looks pretty nice, just missing åäö for me Feb 09 14:20:26 I've been thking of doing a real handwriting recognition engine Feb 09 14:21:04 I think i've got an idea that works for cursive and printed, and can use printed to learn your cursive Feb 09 14:21:16 I've considered something like http://easygestures.mozdev.org/ Feb 09 14:21:31 This is stroke based menuing system basically. Feb 09 14:21:43 * aloril was thinking Nokia communicator (or some future model of it) as exiting phone, but now it sounds boring compared to completely open phone ;-) Feb 09 14:21:45 hold down a button, and you get a radial menu. Feb 09 14:21:48 SpeedEvil: doesnt openmoko have that already? Feb 09 14:22:03 Nice :) Feb 09 14:22:03 Some mix of handwriting reconition and T9 should work pretty good Feb 09 14:22:16 CM: T9 is patented to the eyeballs Feb 09 14:22:25 stroke in that direction - even without waiting for the display, and you get a choice. Feb 09 14:22:43 2 layers or so gets you a keyboard for every letter in two strokes with hints Feb 09 14:22:51 i would have so gotten the nokia e90 if neo didnt appear on the scene Feb 09 14:22:53 now i'm torn Feb 09 14:22:58 Bah.. Stinking patents.. Let's just call it "heuristic dictionary spelling" instead then Feb 09 14:23:09 CM: the idea would be that as an engine learns your handwriting, it sends it back to a central server (probably via tor, for anonymity) Feb 09 14:23:21 robtaylor: Ah, cool Feb 09 14:23:22 Ah - tor over GPRS. Feb 09 14:23:25 CM: so each new release would have a better engine Feb 09 14:23:30 You _want_ this thing to fail, don't you. Feb 09 14:23:34 :) Feb 09 14:23:40 CM: the hard bit is foiling poisonings and outlying results Feb 09 14:24:16 SpeedEvil: have you seen dasher? Feb 09 14:24:16 SpeedEvil: well, it doesn;t really matter when stuff is transmitted back, and the user could of course switch it off Feb 09 14:24:18 Why would you want the engine not on the device? Feb 09 14:24:32 * SpeedEvil is not understanding. Feb 09 14:24:56 SpeedEvil: no, the engine would be on the device, its just things it learned get sent back for the global learning set Feb 09 14:25:19 Robot101: no. I've not looked much in the touchscreen data entry set. Feb 09 14:25:41 dasher's cool. just a pain with the amount of screen real estateit takes Feb 09 14:25:41 Robot101: I think dasher look a bit hysterical with all those colors rushing by Feb 09 14:25:59 robtaylor, yeah, thank CM for that :) Feb 09 14:26:00 ROB1963: in the assumption that there will be commonality between people? Feb 09 14:26:01 SpeedEvil: right, but you reminded me of it with your radial input thing Feb 09 14:26:06 or well Feb 09 14:26:20 also the author of libsydney obviously Feb 09 14:26:25 heh Feb 09 14:26:44 I don't see a problem with simply adding a 'share info' button to the UI, and encouraging users to use it. Feb 09 14:26:46 Much simpler. Feb 09 14:26:52 * CM bows to Lennart and Jean-Marc Feb 09 14:27:22 There is no need to define a central repository for learned strokes. This is non-realtime data, and anyway, centralized services suck in general, for a huge pile of reasons. Feb 09 14:27:24 robtaylor: Or maybe a simple way to "Submit my statistics" Feb 09 14:27:31 SpeedEvil: well, yes, it doent matter *when* its sent back, that can be dependant on lots of things Feb 09 14:27:45 SpeedEvil: but yes, the assumption is there's commonality. which there is Feb 09 14:28:01 I'd prefer to send back all of it as a batch when I'm on a bt pan or usb lan Feb 09 14:28:03 I'm unsure how much 1000 peoples statistics will help over 20 though. Feb 09 14:28:21 its not statistics, its the actual learned cursive strokes Feb 09 14:28:34 robtaylor: What languages btw? Or should I say characters? Feb 09 14:29:02 it'd be a HMM and a TDNN Feb 09 14:29:33 This is for proper handwriting recognition? Feb 09 14:29:47 What's the display again, 60*40mm? Feb 09 14:30:11 CM: well, i think you could design it pretty language agnostic. I've seens similar methods used for arabic. but for pictograms, i dunno. maybe the TDNN would be enough Feb 09 14:30:17 If you're going to build up a knowledge base for recognizing cursive strokes, might as well do it for full handwriting, far more useful. Feb 09 14:30:43 robtaylor: True. Would be a cool project though :) Feb 09 14:31:08 SpeedEvil: well you can do the same as on the messagepad. but yeah handwritings not gonna be huge on a small device, i'm thinking more N800/tablets for real usecases Feb 09 14:31:09 Training users on optimum input systems is cool, but first we need something anyone can pickup and use. Feb 09 14:31:27 for which handwriting is a plus. Feb 09 14:32:00 Yay, imagine, a future where someone/something can actually recognize your doctor's handwriting ... Feb 09 14:32:15 As long as it doesn't do autocorrect :) Feb 09 14:32:15 SpeedEvil: Dasher example: http://www.inference.phy.cam.ac.uk/dasher/images/newdasher.gif Feb 09 14:32:19 Morgreet: i think *that* might be wishful thinking ;) Feb 09 14:32:22 with 2D paper you have only end result that you read, however with touch screen you have also writing timing data so I think traditional handwriting will probably not be optimal input method Feb 09 14:32:44 I think it's actually code, encrypted so that only the pharmacist can decrypt it :P Feb 09 14:32:47 Timing data, and no need to evenly space. Feb 09 14:32:53 which adds. Feb 09 14:33:13 and with multitouch even multiple fingers Feb 09 14:33:34 aloril: you are right there. Handwriting may become obsoleted eventually Feb 09 14:34:28 * SpeedEvil is now blind, after looking at the dasher example. Please excuse typing errors. Feb 09 14:35:06 dasher looks cool and really really weird Feb 09 14:35:16 it is really weird Feb 09 14:35:31 aloril: I have tried to think of more efficient input methods, but not had luck yet Feb 09 14:35:55 With a decent camera, eye-tracker input with dasher would be very cool. Feb 09 14:36:00 i though maybe a two-d dasher (with the letters circuling your pointer) would be cool Feb 09 14:36:02 Speech is about it. Feb 09 14:36:09 speech sucks Feb 09 14:36:15 In some ways. Feb 09 14:36:16 i dont want people snooping on me when i write mails Feb 09 14:36:17 i'd rather have palm graffiti Feb 09 14:36:18 but it requires you to pick up and reset your pen too frequently Feb 09 14:36:30 Graffiti is quite handy. Feb 09 14:36:39 * robtaylor votes dor DNI Feb 09 14:36:46 *for Feb 09 14:36:55 * SpeedEvil votes for robtaylor getting the alpha. Feb 09 14:37:01 heh Feb 09 14:37:14 * CM adds another vote Feb 09 14:38:00 it woudn't be that dangerous ;) Feb 09 14:38:15 bluetooth in your head! Feb 09 14:38:17 robtaylor: A tunnel dasher would be crazy btw.. Feb 09 14:38:42 No need to say speed out loud, it can be subvocalized, most of the cues are still in there. Feb 09 14:38:43 robtaylor: primitive idea: http://lists.openmoko.org/pipermail/community/2007-January/002101.html Feb 09 14:38:44 Everyone looking over your shoulder would think you were playing some game Feb 09 14:38:47 speech* Feb 09 14:39:41 And a throat mike attached to your collar will pick it up. Feb 09 14:40:03 What about running Hildon on the Neo? http://blogs.gnome.org/view/lucasr/2007/02/09/0 Feb 09 14:40:34 CM: that is such a cool idea! Feb 09 14:40:52 CM: ooh, now i want to code one up Feb 09 14:41:29 robtaylor: I'd be happy to test it, or help in any way I can :) Feb 09 14:42:19 aloril: i've been trying to pursude one of my freinds who works at Plastic Logic that they really should do some work on flexible multi-input displays.. Feb 09 14:43:10 gloves + (multi)touch might add further potential Feb 09 14:43:17 (+ camera) Feb 09 14:43:53 aloril: in not really conviced by that, it'd restrict the privacy of your communication Feb 09 14:45:10 + EEG Feb 09 14:45:26 sing-dasher. lower tones = down :) Feb 09 14:45:40 Or whistle-input. Feb 09 14:45:54 For maximisign annoyance to others. Feb 09 14:46:34 with gloves and HMD you could even keep neo in pocket and still input/output Feb 09 14:47:20 aloril: or a bluetooth twiddler Feb 09 14:47:39 * aloril nods to robtaylor Feb 09 14:47:49 or just wires in fingers ;-) Feb 09 14:48:00 EEG is in the works btw. :-) Feb 09 14:48:08 Morgreet: you might want to add this to v2 wishlist: http://www.aeinnovations.com/projects/ver0/ Feb 09 14:48:18 Looking Feb 09 14:50:25 Hehe. With that recent US attempt at legislating against crossing the road while playing an iPod, I can just imagine what they'd make of displays built into glasses. :P But yeah, it has to come. Feb 09 14:51:06 Morgreet: didn't see multitouch in wishlist either (or I missed it somehow) Feb 09 14:51:41 aloril: you're right, enough people have mentioned multi-touch for it to be on there. Feb 09 14:52:13 Done :P Feb 09 14:53:03 Morgreet: 2-3 batteries: one internal (as in v1), 1-2 AA's that you can replace when run down: -> really long runtime (in heavy usage from hours to days) Feb 09 14:53:20 though I guess that will happen as external accessory in v1 too Feb 09 14:53:50 * SpeedEvil tried dasher. Interesting. For the moment - xvkbd + thumbstick is faster. Feb 09 14:53:51 That's more along the lines of my modular phone suggestion, but I didn't add that because nobody else wanted it. Feb 09 14:54:31 wated is different from not believing to be manufacturable with no compromise. Feb 09 14:54:54 aloril: do we know how the battery is attached physically to the Neo yet? Eg. can a larger battery be attached in the same place directly? Feb 09 14:55:20 Morgreet: larger battery is not enough, I don't want it *that* big ;-) Feb 09 14:55:29 aloril: hehehe Feb 09 14:55:48 12V heavy duty lead-acid, huh .... Feb 09 14:55:52 just swappable battery while its using 100% of CPU ;-) Feb 09 14:56:53 Morgreet: hehe, I guess what would give hmm.. if current is 3h heavy usage, then it might be close to 1000 hours Feb 09 14:57:34 aloril: just plugging an external pack into the USB is not enough? Feb 09 14:58:28 Morgreet: yeah, I guess need to settle for that, but having it internal is more convenient (and at the same time would give powered hub too) Feb 09 14:59:09 might be possible even on v1 Feb 09 15:00:55 aloril: I think the problem with "interesting" h/w design is FIC's ambivalence about the target audience. Yes, we want these things, but as long as they want Mom+Dad as a target, and/or as long as they want to sell the same h/w in the WinMobile market, then we have a problem ... they're just not going to produce what techs really want. Feb 09 15:01:46 Morgreet: yeah, so provide swappable case where you could select a bit bigger case to fit 2 extra AA's and extra USB ports ;-) Feb 09 15:02:13 aloril: good idea! It may be as simple as just a case swap! I like that. Feb 09 15:03:01 I guess they'd have to not populate the motherboard with components for some of their batches, for cost reasons, but shouldn't be a problem. Feb 09 15:03:19 aloril, :) Feb 09 15:05:23 Multiple case options are important anyway just for style/colour marketting reasons. It's not a huge leap from there to make the cases actually physically different. Feb 09 15:05:38 * SpeedEvil wants PCI slots! Feb 09 15:05:53 PCIe, please. 64-lane ;-) Feb 09 15:07:20 there is actually a new standard for pcie over cables Feb 09 15:07:24 that would be a nice thing Feb 09 15:08:01 Oh, remote extension to PCIe? Not seen that, but makes sense. Feb 09 15:08:09 i definitely want that in my next notebook Feb 09 15:08:09 Have you seen the ATI R600 card? It's huge.. http://www.vr-zone.com/?i=4622 Feb 09 15:08:21 or even a bit bigger case and put extra USB port and a bit empty space inside case Feb 09 15:08:21 gimme 4*4 lane connectors Feb 09 15:08:35 12.4" = 31.5 cm Feb 09 15:09:00 aloril: oh, so *THAT's* the lunchbox version that FIC was talking about ..... ;-))))) Feb 09 15:09:32 * Morgreet gets sandwiches ready ... Feb 09 15:09:43 wifi for v1 without outside items Feb 09 15:10:04 or drill small hole, but small camera ;-) Feb 09 15:10:11 or memory stick or .. Feb 09 15:10:46 CM: that's gotta be a PS/GIMP fake Feb 09 15:11:03 I don't think so.. It's supposed to be huge.. Feb 09 15:11:08 April 1's come early ... Feb 09 15:11:39 :) Feb 09 15:11:41 270W? Feb 09 15:11:46 thats gotta be a joke Feb 09 15:12:05 aloril: what are you talking about? Feb 09 15:13:16 If that's a real graphics card, nVidia will be laughing themselves silly. After all, they went through the same ridicule with their "leaf blower". Feb 09 15:13:21 loufoque: optional bigger case for Neo1973 which would include 2 AA's to charge it + extra powered USB ports (powered by those AA's) (and maybe even one inside extra USB port and a bit empty space) Feb 09 15:14:04 loufoque: modification to board should not be big: current USB connector -> to this new case -> new 2-4 USB connectors Feb 09 15:14:15 aloril, someone should start a "neo1973 hacks" page :) Feb 09 15:14:30 aloril: I saw you talking about wifi Feb 09 15:14:36 success stories in modding the cute little phone Feb 09 15:14:37 actually, with 3D printer it could be done by somebody else even... Feb 09 15:14:58 loufoque: well.. you could use that extra *inside* USB port + empty space to put wifi USB stick there ;-) Feb 09 15:15:15 loufoque: in this hypothetical bigger case Feb 09 15:15:50 i'd totally buy that Feb 09 15:15:58 I like the idea of optional cases. It's just plastic. Hardware hackers can then do their thing inside them. Feb 09 15:15:59 there should be a lego or mechano case where you could add new features by plugs & bolts :DD Feb 09 15:15:59 make it 6mm thicker Feb 09 15:16:17 so that a AAA battery fits Feb 09 15:16:53 doo Feb 09 15:16:58 those are nearly 10mm Feb 09 15:18:05 hmm.. replace internal battery with smaller one and put AAA's on top of that? (or is that really feasible) Feb 09 15:18:31 lol Feb 09 15:18:39 replace battery with external battery in backpack Feb 09 15:19:21 * xkr47 thinks a second device in the backpack with usb ports, bluetooth connectivity and aa batteries would be nice Feb 09 15:19:52 hmm.. that could work Feb 09 15:19:55 USB over bluetooth Feb 09 15:20:18 is there such a thing ? Feb 09 15:20:21 ah... but it doesn't charge neo Feb 09 15:20:25 true Feb 09 15:20:34 * aloril wants days of runtime for Neo without rebooting it Feb 09 15:20:40 Why focussing on those standard cylindrical batt sizes? Why not just carry standard size mobile batteries as your "rechargeable batteries"? Far better form factor, and greater power density. Feb 09 15:20:49 aloril, then hook up a car battery to it :D Feb 09 15:21:08 xkr47: that would give month runtime, don't think I need that much ;-) Feb 09 15:21:09 Morgreet, but how do you charge those ? Feb 09 15:21:30 aloril, probably the battery would run empty just as fast without anything connected :) Feb 09 15:21:34 xkr47: same as ordinary batteries ... in an external charger :P Feb 09 15:21:58 which you get where? Feb 09 15:22:13 maybe he means like camera batteries which you can buy chargers for Feb 09 15:22:38 buz: think "cameras", not mobiles. Li-Ion powerpacks with external chargers are everywhere. Feb 09 15:22:50 mhh those have flat powerpacks? Feb 09 15:23:07 my camera has a more cubic thingy Feb 09 15:23:41 run it on a feul cell Feb 09 15:23:43 buz: that would depend on the camera. I've only got a DSLR, so the battery is large. Feb 09 15:24:39 Morgreet: AA's are more standard (charging, batteries) -> good price/energy ratio Feb 09 15:25:23 aloril: price yes, but you don't get Li-Ion AAA's so you don't get Li-Ion power density. Feb 09 15:25:39 NiMH is comparatively poor. Feb 09 15:27:11 yeah, maybe its worth to invest on those, anyway need enough big capacitor to hot swap even with 100% CPU usage (or some small extra battery) Feb 09 15:27:21 And Li-Ion has the additional advantage that it doubles as your campfire ... at least if you buy from Sony. ;-) Feb 09 15:29:07 NiMH is everywhere though, as are chargers. So unless weight is a problem, the best solution is probably to build up your own large external powerpack out of NiMH cells. Feb 09 15:29:53 Morgreet: yes, but that means external wire and I would like to avoid that Feb 09 15:30:29 an the other hand it doesn't need to be connected all the time, just enough to charge Feb 09 15:30:32 Of course, you'd get detained and interrogated several times a day if you wore a power belt of NiMH cells .......... :-( Feb 09 15:31:13 heheheh Feb 09 15:32:20 isn't the Neo1973 big enough? Feb 09 15:32:37 it's actually bigger than the iphone, that must have a great power consumption Feb 09 15:33:12 Nah, aloril wants to fit Battersea power station, and I need somewhere to carry my sandwiches ;-) Feb 09 15:34:55 we need nucular batteries Feb 09 15:35:33 I'm actually on the lookout for another gadget, with a screen the size of a paperback novel. For some reason, there is no such thing, outside of picture frames and portable DVD players. Feb 09 15:35:50 thereis Feb 09 15:35:54 its called subnotebook Feb 09 15:36:08 nucular. i love that pentagon brogue :) Feb 09 15:36:42 Subnotebooks would be fine if they had rotateable screens to work as a tablet. Without that though, they don't fit my bill. Feb 09 15:37:13 ive seen a 10 tablet Feb 09 15:37:16 inch Feb 09 15:38:00 Expand the Nokia 770 or 800 by a factor of 2.5 each way, and that would be what I want. Feb 09 15:38:27 A Simpad? Feb 09 15:38:51 Or just x2 might be enough, haven't measured it exactly. Feb 09 15:40:37 I've never seen a Simpad on sale, but yeah, that sort of thing. I've got a tablet PC (Tosh Tecra M4), but it's entirely the wrong form factor. Feb 09 15:40:43 hey all Feb 09 15:41:25 fwiw, i doubt wearing a powerbelt of nimhs would get you stopped at an airport Feb 09 15:42:09 i do a lot of travel with an apc universal external battery wired up in a scott evest with charging cables to all the pockets Feb 09 15:42:17 fwiw: you're probably right. They would more likely just shoot you on the spot. Feb 09 15:42:38 and the only place i've ever been stopped was when i entered disneyworld in 2005 Feb 09 15:44:23 robtaylor: did you manage to have a look at that pixops patch? Feb 09 15:52:39 Just noticed, unlike the Tecra M4, the Tecra M5 isn't a tablet/covertible. :-( Glad I bought the M4 when it was available. Feb 09 15:54:28 http://uk.computers.toshiba-europe.com/cgi-bin/ToshibaCSG/jsp/productPage.do?service=UK&PRODUCT_ID=104798&tab=3 Feb 09 15:54:58 Fantastically specc'd product for low price. Hilden could run on it nicely, with the pen-driven touch screen. Feb 09 15:55:25 I'll prolly run OM on it at some point, with pen input. Feb 09 15:55:26 counter Feb 09 15:55:27 2 days 13:31:58 (2.564 days) for source for *all* developers and devices for selected developers (2007-02-11); a month, 2 days (30.564 days) for devices for *ANYBODY* for $350 (targeting developers) (preorders: not yet) (2007-03-11); 7 months, 2 days (214.564 days) for mass market (2007-09-11): see topic for more info (189) Feb 09 15:56:31 Last 24 hour countdown will be funny :-) Feb 09 15:59:25 hehhe, someone will issue a counter every minute :) Feb 09 16:11:26 counter Feb 09 16:11:26 2 days 13:15:58 (2.553 days) for source for *all* developers and devices for selected developers (2007-02-11); a month, 2 days (30.553 days) for devices for *ANYBODY* for $350 (targeting developers) (preorders: not yet) (2007-03-11); 7 months, 2 days (214.553 days) for mass market (2007-09-11): see topic for more info (190) Feb 09 16:28:25 i wish you all a nice WE, good night, ugt Feb 09 16:31:03 WE? That was yesterday. Feb 09 16:31:04 Oh Feb 09 16:46:57 At this stage in the counter, its timezone is starting to become relevant. :-) Presumably it's either UTC or Taiwan's? Feb 09 16:47:58 Actually, no, it's simply stating time left, not a time of day. Cool! Feb 09 16:48:53 and the time left is a very easy conversion to swatch internet time Feb 09 16:51:28 Actually, it does appear to be sync'd to Swatch time, ie. UTC+1 Feb 09 16:51:48 Ah no it doesn't. Feb 09 16:51:55 counter Feb 09 16:51:55 2 days 12:35:29 (2.525 days) for source for *all* developers and devices for selected developers (2007-02-11); a month, 2 days (30.525 days) for devices for *ANYBODY* for $350 (targeting developers) (preorders: not yet) (2007-03-11); 7 months, 2 days (214.525 days) for mass market (2007-09-11): see topic for more info (191) Feb 09 16:52:20 midnight GMT-4? Feb 09 16:52:39 er, GMT-6 Feb 09 16:52:43 That's what it looked like to me, except it would be midday not midnight Feb 09 16:53:07 midday in taiwan Feb 09 16:53:20 But there's a 30-minute discrepancy Feb 09 16:53:34 no, 2pm in taiwan Feb 09 16:54:25 LOL, not there isn't a discrepancy, the PC clock is shafted on this box, and NTP broke some time back, lol. Feb 09 16:56:08 That's better, I'm back in sync with the planet now ;-) Feb 09 16:56:13 counter Feb 09 16:56:13 2 days 12:31:11 (2.522 days) for source for *all* developers and devices for selected developers (2007-02-11); a month, 2 days (30.522 days) for devices for *ANYBODY* for $350 (targeting developers) (preorders: not yet) (2007-03-11); 7 months, 2 days (214.522 days) for mass market (2007-09-11): see topic for more info (192) Feb 09 17:07:43 LOL, half a magazine was inside the house and half outside the letterbox. The half outside was a solid block of ice. Feb 09 17:09:43 *lol* Feb 09 17:19:22 I wonder if OM will spawn multiple distros for Neo ... ones with different desktops, different app sets, different levels of cuteness (very handy for Neos that we give as presents!), etc Feb 09 17:20:36 re Feb 09 17:20:54 Morgreet: rather it will be base system + feeds with misc software Feb 09 17:21:46 hrw|tv: yeah, that's the corporate plan mentioned by Sean, but of course the community is free to go further. Feb 09 17:27:07 Does anyone have any details on the syncml functionality that openmoko is going to have? Feb 09 17:27:58 like is the work already done? :) Feb 09 17:28:57 bipolar: "Because OpenMoKo consists exclusively of open-source software, the Neo1973 will ship with a limited feature set, including a dialer (image at left), unified SyncML-enabled email/text messaging client, phonebook, (image at right), and media player, according to Moss-Pultz." Feb 09 17:29:34 http://www.linuxdevices.com/news/NS2986976174.html Feb 09 17:30:32 hmmm Feb 09 17:31:02 that links to http://linuxdevices.com/news/NS7056478804.html Feb 09 17:31:23 which talks about the addressbook and calendar also syncing, but it's vauge Feb 09 17:33:12 countdown Feb 09 17:33:20 counter Feb 09 17:33:20 2 days 11:54:04 (2.496 days) for source for *all* developers and devices for selected developers (2007-02-11); a month, 2 days (30.496 days) for devices for *ANYBODY* for $350 (targeting developers) (preorders: not yet) (2007-03-11); 7 months, 2 days (214.496 days) for mass market (2007-09-11): see topic for more info (193) Feb 09 17:33:23 there we go. Feb 09 17:33:53 Morgreet: I guess I can find out more when the source is realeased :) Feb 09 17:34:40 bipolar: well I'd guess that Sean's statement referred to the public release date anyway, ie. in September. Feb 09 17:35:02 ahh Feb 09 17:35:29 There's a lot of work already being done in related areas though, like EDS and dbus. Feb 09 17:35:53 For all I know, it might all be working. We're all speculating :P Feb 09 17:36:44 Here at work we us Citadel for our email/groupware, and I have Funimbol set up to sync palms with it via syncml. Feb 09 17:37:02 it's not finished yet, but it's really close. Feb 09 17:37:10 cu Feb 09 17:37:17 hrw|gone: later Feb 09 17:45:38 What's the difference between a "feature phone" and a "smartphone"? Feb 09 17:45:46 price Feb 09 17:46:00 and with nokia devices the os Feb 09 17:46:35 nokia feature phones run nokia's internal os and you can't install any apps on them Feb 09 17:46:54 apart from java sh.. i mean software Feb 09 17:47:09 That's the definition of "smartphone"? The ability to load external apps? Feb 09 17:47:09 nokia smartphones run symbian and series60 Feb 09 17:48:07 the (undelivered) promise of extensibility i think Feb 09 17:48:51 i think samsung and motorola call some of their models smartphones as well Feb 09 17:49:06 Oh well, we're certainly about to see it delivered .... bigtime :-) Feb 09 17:49:07 even though there's no 3rd party software for them Feb 09 17:50:00 not that the application catalog on series60 is much better Feb 09 17:50:12 Neo may not be leading edge on h/w, but it's sure as hell going to be "smart". Feb 09 17:51:19 and imho being leading edge on hardware isn't really that important to begin with Feb 09 17:52:11 when we have a software and services ecosystem that actually can fully take advantage on things like 3g, then you might want to be on leading edge Feb 09 17:52:56 right now people buy the leading edge hardware mainly as a show-off Feb 09 17:53:17 not as something that actually makes the device more useful Feb 09 17:53:21 That's true enough. And anyway, we can compensate somewhat for lack of the latest and greatest by using external h/w. Not as convenient, but we're techs, we can cope. Feb 09 17:53:56 techs can cope with almost anything Feb 09 17:54:18 the hard problem imho is how we can get this platform into the mainstream Feb 09 17:54:54 jannu: we don't yet know if that's an issue or not. FIC is very ambivalent about it. Feb 09 17:54:54 and make it all usable for normal people. in a way that it actually makes their lives easier Feb 09 17:55:48 3G would be a lot more useful with just a little larger CPU. Feb 09 17:56:07 10x10cm? Feb 09 17:56:13 would that be a big enough cpu? Feb 09 17:56:13 :) Feb 09 17:56:26 Morgreet: well imho regardless of the target segment for the neo1973, i'd say it's really important to get the openmoko platform into mainstream phones Feb 09 17:57:04 Morgreet: because only then the platform will start attracting bigger developer masses Feb 09 17:57:19 Morgreet: and that's when everyone benefits. big time Feb 09 17:57:52 SpeedEvil: a more powerful CPU would probably be the single most important change, methinks, simply because this phone will succeed if its software succeeds ... and that needs power. FOSS is better at adding features than reducing resource footprint. Feb 09 17:58:02 SpeedEvil: the problem with 3g isn't just processing power. it's the lack of actually useful services Feb 09 17:58:04 Y&eah. Feb 09 17:58:22 CPU and memory Feb 09 17:58:33 SpeedEvil: after all, 3g is just bandwidth Feb 09 17:58:37 Memory, well... IMO, 128M is quite adequate. Feb 09 17:58:42 nothing more, nothing less Feb 09 17:58:46 True, but for stuff like video decompression. Feb 09 17:58:51 And compression in real time. Feb 09 17:58:52 jannu: less latency Feb 09 18:00:12 koen: yes. usually. unless there's lots of higher priority traffic (i.e. carrier controlled stuff) Feb 09 18:00:28 jannu: I don't think it needs to encourage larger developer masses. The entire principle of OM and OE is very close to "all FOSS software works" (within reason). There are already hundreds of thousands of developers working on OM ... indirectly. ;-)) Feb 09 18:00:49 Very little FOSS stuff works well. Feb 09 18:01:00 Most of it is geared to the desktop. Feb 09 18:01:00 Morgreet: yep. but you've got the same problem there as with mobile internet Feb 09 18:01:10 A 2.8" screen at arms length is very, very different. Feb 09 18:01:23 Morgreet: the UIs designed for desktop use will not work on a phone Feb 09 18:01:42 They work if you hold the phone 20cm away from your eyes, and use a pointer. Feb 09 18:01:45 But... Feb 09 18:01:47 Morgreet: and even on the desktop side the UIs aren't that great Feb 09 18:02:12 SpeedEvil: true. But FOSS peeps aren't generally good at optimizing, because they tend to develop for desktops. If jannu had said "attract more optimizers", I'd have agreed ;-) Feb 09 18:02:53 but for backend stuff the existing stuff (like dbus, eds, telepathy) will do wonders Feb 09 18:03:31 We could really do with prizes to reduce the amount of memory stuff uses. For example a copy of firefox, with a specified list of URLs clicked - say 1000 or so. Now, you get a dollar for every K you can take that down. Feb 09 18:03:51 Morgreet: i wouldn't necessarily agree with that Feb 09 18:03:53 jannu: I bet it's judged to be sluggish, unless a huge amount of filing and honing is done. Feb 09 18:04:13 Morgreet: the biggest bloatware i've seen has always been commercial Feb 09 18:04:34 jannu: that's true :P Feb 09 18:04:49 Hello Feb 09 18:05:11 Who should I speak with about the 2/11 deadline for development kits on the Neo1976..? Feb 09 18:05:22 nobody. Feb 09 18:05:32 Wait till preorder is announced, and buy one. Feb 09 18:05:33 nab: What's your question? Feb 09 18:05:46 oops. Feb 09 18:05:48 My bad. Feb 09 18:05:48 of course the difference here is that commercial phone software has been developed for phones from the get go Feb 09 18:05:50 It's one of the things I hate most, commercial bloatware. It's also why I give projects like OpenOffice a thumbs down. I like small independent and highly specific apps, not integrated suites. Feb 09 18:06:07 cjb: http://www.linux-iscsi.org/index.php/ISCSI/GSM Feb 09 18:06:22 I just need the phone and a method to access the bandband processer with kernel-level sockets. Feb 09 18:06:23 :P Feb 09 18:06:47 Actually, $$ is not a concern, I just want to get my hands on one as soon as possible. Feb 09 18:06:54 Morgreet: I still can't understand why the openoffice guys can't make a library out of the MS office file format converters Feb 09 18:07:04 nab: Where are you based, out of interest? Feb 09 18:07:12 cjb: Bay Area, CA. Feb 09 18:07:26 nab: you can't. There is only a modem connection to the GSM module. If I understood what you asked. Feb 09 18:07:35 nab: There were only fifty or so of the development phones, so most people are waiting another month for the release. Feb 09 18:07:35 UART Feb 09 18:07:37 cjb: The project will be undertaken with one of the quad bandband flat-rate data providers in these parts. Feb 09 18:07:57 cjb: Ahh, I see. Feb 09 18:07:59 nab: The GSM module is proprietary, and communicated with via AT commands. Feb 09 18:08:03 jannu: abiword can do conversion from CMDLINE Feb 09 18:08:06 (or something like that.) Feb 09 18:08:23 How are userspace applications that use sockets communicate with the UART..? Feb 09 18:08:23 the openoffice guys are doing a converter toolkit Feb 09 18:08:38 nab: it looks like a serial port. Feb 09 18:08:45 SpeedEvil: Yes, a UART.. Feb 09 18:08:46 You do serial port stuff to it, and send it AT commands. Feb 09 18:09:00 koen: does it use the oo code to do it, or another implementation? Feb 09 18:09:09 it uses abiword code Feb 09 18:09:10 My question is how are other applications that communicate with a endpoint using TCP communicating with the bandband processor. Feb 09 18:09:21 so it doesn't require 6GB ram and a 100GHz cpu Feb 09 18:09:23 what do you mean by bandband? Feb 09 18:09:33 er, baseband. Feb 09 18:09:42 koen: so i'm guessing the compatibility isn't quite as good? Feb 09 18:09:55 SpeedEvil: The GSM processor that is communicating via a serial interface with the application processor. Feb 09 18:10:04 I continue to be unimpressed with OOo Feb 09 18:10:05 while abiword impresses me Feb 09 18:10:07 Right - google GPRS. Feb 09 18:10:15 both used libwv iirc Feb 09 18:10:34 koen: me too. but the ms office compatibility is a must for me Feb 09 18:10:38 SpeedEvil: huh? Feb 09 18:10:41 That's the only probably supported way of doing IP over the air through the mobile. Feb 09 18:10:44 I know what the General Packet Radio Service is. Feb 09 18:10:47 koen: libwv is for wordperfect Feb 09 18:10:49 jajanyou want oo then Feb 09 18:10:55 sorry, jannu Feb 09 18:10:59 It'll appear something like a PPP interface I imagine. Feb 09 18:11:20 buz: i think want is the wrong word Feb 09 18:11:29 SpeedEvil: I am not asking about the wire protocol, I am asking about the API that applications will be using. Feb 09 18:11:33 need/whatever Feb 09 18:11:36 buz: being forced to use is a bit closer Feb 09 18:11:47 It'll look like any other network interface. Feb 09 18:11:48 nab: all the ETSI GSM TS 07.07 commands are available, via the AT comamnd set Feb 09 18:11:49 SpeedEvil: That is the answer I am looking for. Feb 09 18:11:50 nab: probably dbus or linking against libgsmd Feb 09 18:11:57 koen: Yikes Feb 09 18:12:17 koen: I am dealing with a kernel module here that uses sock_[send,recv]msg().. Feb 09 18:12:24 nab: there will be source available at 2/11 for all developers - there will be your answer i think Feb 09 18:12:24 the points of libgsmd is to hide the AT layer Feb 09 18:12:43 s/st/on/ Feb 09 18:12:46 koen: Hmmmm Feb 09 18:12:46 err Feb 09 18:13:00 * fluffs gets on with submitting his latest round of 24xx updates upstream Feb 09 18:13:12 so_solid_moo: you wouldn't have a link or other info on that toolkit? Feb 09 18:13:16 nab: so your application works on every phone with libgsmd Feb 09 18:13:21 koen: Not quite Feb 09 18:13:25 koen: http://www.linux-iscsi.org/index.php/Core-iscsi_Interopt Feb 09 18:13:30 jannu: http://odftoolkit.openoffice.org/ Feb 09 18:13:44 The project is a kernel module that implements the initiator side of RFC-3720. Feb 09 18:13:51 nab: how's that related to AT commands? Feb 09 18:13:58 koen: Its not. Feb 09 18:13:58 AT SCSI? Feb 09 18:14:39 iirc gprs shows itself as a ppp interface Feb 09 18:14:41 so_solid_moo: wow. hopefuly that'll take off so I can finally get ms office stuff working in koffice Feb 09 18:14:41 The problem is that if userspace applications link against a userspace library to hide the UART communication, what are kernel-level applications that use IP supposed to do..? :P Feb 09 18:14:57 but with some neat logic to stop it going through the kernel 2 or 3 times Feb 09 18:15:10 why do kernel modules want to use the uart? Feb 09 18:15:17 nab: I don't get it. You quoted us RFC-3720, and you say it's not relevant? Feb 09 18:15:24 especially if they don't care about AT commands Feb 09 18:15:28 nono.. Feb 09 18:15:30 jannu: with current state of koffice, i'd rather see them get the thing stable first Feb 09 18:15:50 Ok, an iSCSI Initiator is a stack that accepts SCSI CDBs and data from the Host OS'es SCSI subsystem. Feb 09 18:16:06 Encodes the SCSI data into iSCSI packets and sends it off to it's target endpoint. Feb 09 18:16:15 buz: that wouldn't hurt either Feb 09 18:16:36 buz: although kword hasn't crashed on me for months now Feb 09 18:16:49 and then maybe, just maybe, make it open ODT files somewhat sanely Feb 09 18:16:52 And due to the fact that the SCSI CDBs and data are coming down from a subsystem that resides in the kernel, all of the existing implementations are in kernel space, or at least all the ones that I have seen in my 6 years of doing iSCSI. Feb 09 18:17:37 So, if I understand correctly, and the only method that currently exists for applications to communication with the serial UART that is connected to the GPRS module is via a userspace library, I have a problem. Feb 09 18:17:52 nab: you don't understand Feb 09 18:17:59 you want to iscsi over gprs? Feb 09 18:18:03 or at least are ignoring most of the stuff we are saying Feb 09 18:18:06 buz: That is what I am doing. Feb 09 18:18:11 IT SHOWS UP AS A PPP INTERFACE Feb 09 18:18:16 omfg Feb 09 18:18:20 koen: I understand that piece. Feb 09 18:18:30 ok Feb 09 18:18:39 but WHY? Feb 09 18:18:39 then why are you still talking about uarts? Feb 09 18:19:18 buz: This is what I have been doing for a number of years. :P Feb 09 18:19:28 Putting iSCSI in places that people say it cannot be put. :P Feb 09 18:19:40 Interesting hobby ;) Feb 09 18:19:46 so next thing is iSCSI over pigeons i imagine Feb 09 18:19:51 it's like we are saying "ethernet interface" and you keep raving about "cat5 access" Feb 09 18:20:15 koen: Where I am getting confused is the userspace library that hides things.. Feb 09 18:20:27 PPP is never hidden by anything Feb 09 18:20:41 only abstracted further Feb 09 18:20:59 koen: So when I hear that linking against a userspace library to get socket-level access to the GSM module, I start to think that kernel-level socket access will require additional work. Feb 09 18:21:13 buz: rerouted would be a better description for what harald has created Feb 09 18:21:26 ifconfig gprs0 Feb 09 18:21:26 So let me see if I got this right.. Feb 09 18:21:32 route add gprs0 Feb 09 18:21:34 job done Feb 09 18:21:35 nab: "socket level access to the gsm module" doesn't exist Feb 09 18:21:39 * buz is still in total disbelief Feb 09 18:21:43 There's nothing inherently wrong with iSCSI over a mobile phone, in fact it's nice lateral thinking as long as the application is some kind of odd slow one. But I don't actually see the problem. It's just IP. Feb 09 18:21:48 nab: it's like "socket level access to cat5" Feb 09 18:21:53 that's gotta be the by far weirdest (and that's saying something) i've read in this channel so far Feb 09 18:22:02 koen: ..? Feb 09 18:22:15 Ok, lets try this again. Feb 09 18:22:22 nab: "ppp interface" is all the information you need for your iscsi toy Feb 09 18:22:42 isnt iSCSI running on top of IP annnyhway? Feb 09 18:22:43 koen: So all outgoing packets will be rounted to said interface, yes? Feb 09 18:22:50 yes. Feb 09 18:22:52 buz: Most definately. Feb 09 18:23:01 all gprs traffic flows through the ppp interface Feb 09 18:23:01 then why do you even care how the interface is implemented Feb 09 18:23:02 I don't think nab means what he's saying. Or at least, doesn't really expect it to work that way. Doing that stuff in kernel in a mobile phone would be daft. Feb 09 18:23:09 it will talk IP for like totally sure Feb 09 18:23:58 Morgreet: I am getting alot of different and conflicting answers right now. Feb 09 18:24:01 :P Feb 09 18:24:14 Ignore totally the phone hardware. Feb 09 18:24:19 It does not matter to you. Feb 09 18:24:21 think of it as linux Feb 09 18:24:23 You have an interface. Feb 09 18:24:25 And I am too excited to finally move iSCSI packets to wait until monday to look for myself. Feb 09 18:24:35 And forget kernel modules too :-z0 Feb 09 18:24:42 It looks like an ethernet, ppp, or ... interface. Feb 09 18:24:49 Yes, understood. Feb 09 18:24:51 When it's up, you can route tcp/ip packets over it. Feb 09 18:24:55 Of course. Feb 09 18:25:44 When it's down, you can bring it up if in range of a cell, for a fee. Feb 09 18:25:52 nab: for making phonecalls, libgsmd abstracts a lot away Feb 09 18:25:53 Hehehe Feb 09 18:26:06 nab: you're not interested in phonecalls, but in packet data Feb 09 18:26:09 telcos will just love iscsi over gprs Feb 09 18:26:13 koen: Ok, libgsmd is where I was getting confused. Feb 09 18:26:28 buz: Bringing it terabytes over a flatrate provider. :P Feb 09 18:26:28 that's for dialer apps to talk to Feb 09 18:26:36 koen: Ok, thank you. Feb 09 18:26:38 or apps that wants to read contacts from SIM Feb 09 18:26:42 you must have a lot of time if you talk about TB Feb 09 18:26:43 The confusion has now gone away. Feb 09 18:27:00 buz: Not really, iSCSI is what I have been doing for a living for along time now. Feb 09 18:27:23 buz: But usually it involves breaking records for speed. Feb 09 18:27:34 at like 5kb/s a TB would mean a looooong time Feb 09 18:27:44 So now the linux-iscsi.org team wants to break records for squeezing TBs into a mobile. :P Feb 09 18:28:05 you want one with HSDPA then Feb 09 18:28:15 buz: The device I personally want is a access to my media library from my PS3. Feb 09 18:28:15 :P Feb 09 18:28:32 well, audio library actually. Feb 09 18:28:51 And 5k-10k on GSM with an intelligent player is just enough for audio. Feb 09 18:28:51 nab: does the openmoko give you something more useful than you'd get from, say, a Nokia 770 paired with a 3G phone, then? Feb 09 18:29:04 nab: I guess I don't see why you think the openmoko is special for this. Feb 09 18:29:16 cjb: Already did that Feb 09 18:29:17 It doesn't have to be used for transporting terabytes though. It's just SCSI after all, so conceptually you could grab a photo from your scanner with it. Feb 09 18:29:18 http://www.linux-iscsi.org/index.php/Core-iSCSI/Nokia_770 Feb 09 18:29:26 cjb: Plus it means I have to have two devices in my pocket Feb 09 18:29:29 I want one device. Feb 09 18:29:47 nab: Okay, just checking. Feb 09 18:30:00 The neo will be several dozen times slower than a 3G capable phone. Feb 09 18:30:05 cjb: I have already spent some time hacking on a Rokr E2, but an open stack is what would make the team's life much easier. Feb 09 18:30:26 SpeedEvil: iSCSI has already maxed out 802.11b bandwith on a NintendoDS> Feb 09 18:30:31 WIth single digit CPU load. Feb 09 18:30:40 Well, yes. Feb 09 18:30:48 The SoC that the neo uses will be more than enough power to max out GPRS bandwith. Feb 09 18:30:48 It's quite irrelevant though. Feb 09 18:31:03 For iSCSI/GSM it is quit relivant. Feb 09 18:31:06 Maxing out GPRS can almost be done by a very fast typist. Feb 09 18:31:33 SpeedEvil: You would be amazed of the misconceptions about iSCSI that I have had to disspell over the years. Feb 09 18:31:39 10 monkeys with a typewriter could do it Feb 09 18:31:44 My apologies if I tried to dispell one by mistake. Feb 09 18:32:20 my congratulations to you, for the weirdest project in a long time Feb 09 18:32:22 iScsi is just SCSI - over a 5K link, why. Feb 09 18:32:37 heck original scsi had what like 10mb/s? Feb 09 18:32:48 nab: it could be very cool, given sensible targets. But kernel modules would be a really bad idea. Feb 09 18:33:21 Morgreet: On the initiator side, since the initiator stack has to be speaking with the OS'es SCSI subsystem, there is not much choice. Feb 09 18:34:41 Also, I think alot of the questions that you folks may have on the transport side are answered iSCSI/GSM wiki entry. Feb 09 18:35:02 But I am more than happy to answer and other questions about the how and why. Feb 09 18:35:37 The why is obvious. Feb 09 18:35:54 I have a bunch of music sitting on my playstation3 that is acting as an iSCSI target. Feb 09 18:36:08 I want to be able to access my music wherever I go on my mobile Feb 09 18:36:27 Here in sunny CA, there are flat-rate data providers. Feb 09 18:36:41 In general, you really don't want to do that. Feb 09 18:36:52 You'll be reading filesystem metadata over GPRS. Feb 09 18:36:54 That's insane. Feb 09 18:37:25 SpeedEvil: For daily use the filesystems are mounted R/O to save on that particular overhead. Feb 09 18:38:02 And you do know that it'll take 10-20 minutes to download your typical 3 minute MP3? Feb 09 18:38:17 SpeedEvil: Depends on your provider. :P Feb 09 18:38:25 nab: no it doesnt Feb 09 18:38:32 This is not a 3G phone, it's GPRS. Feb 09 18:38:35 ~10k is what I am looking for. Feb 09 18:38:37 nab: okay Feb 09 18:38:53 it wont do 10k Feb 09 18:38:56 And here in the Bay that is about what we see. Feb 09 18:39:06 that depends on the phone Feb 09 18:39:54 buz: And the Neo is limited to 5k..? Feb 09 18:40:12 as a rule of thumb 5k doesnt sound so bad Feb 09 18:41:11 Ok, obviously one of the research items for this project has been to determine what the bandwith requirements will be to actually be functional. Feb 09 18:41:32 I don't think that the problem here is iSCSI, but using iSCSI for access to storage, which is rather badly mismatched to GPRS. But if you can map SCSI commands at the far end to something else with some lateral thinking, it could be interesting. Feb 09 18:41:49 So after researching what the various providers in my area claim, against what the actual customers see, the provider that I chose for the project on a provider sold phone was ~10k. Feb 09 18:42:00 So I am hearing now that with another quad-band GSM phone I will not see that..? Feb 09 18:42:43 Against, please understand that I am new to developing on mobile devices. :-) Feb 09 18:43:13 Morgreet: Actually, iSCSI is the best protocol for moving storage packets over a link that expects to drop/restart many many times. Feb 09 18:43:49 Morgreet: The biggest advantage of iSCSI is the fact that it supports multiple communication paths within the nexus, as well as true active-active recovery when communication paths go away. Feb 09 18:44:00 (My team was the first to implement this back in 2003) Feb 09 18:44:02 :P Feb 09 18:44:40 You can read more about that probject here btw: http://commsdesign.com/printableArticle/?articleID=172300547 Feb 09 18:44:45 For those who are interested. :_) Feb 09 18:45:32 Morgreet: Compared with any other NAS/SAN transport, iSCSI is best suited for moving packets over a slow radio link. Feb 09 18:46:07 Actually, iSCSI has already moved packets over an even slower link. Feb 09 18:46:37 http://www.linux-iscsi.org/index.php/ISCSI/CDMA Feb 09 18:47:53 Researchers at the University of Seoul beat the linux-iscsi.org team on CDMA (due to lack of hardware), but I think that Linux/iSCSI's iSCSI/GSM will be the first useable software package. Feb 09 18:48:13 But again as mentioned above, this all goes back to the pipe. Feb 09 18:48:15 brb Feb 09 18:56:39 I just had a very funny image of doing a sort of multicast p2p over iSCSI by treating a bunch of peers as a RAID array ...... Feb 09 18:57:50 Download cooperation - if in BT range, and the user next to you has a cheaper GPRS rate, buy some transfer. Feb 09 18:57:54 mhh run it over satelite ;) Feb 09 18:58:20 SpeedEvil: hahaha, nice. And evil :P Feb 09 18:58:59 Hmmm, takes us into mesh territory. Feb 09 19:00:18 or dont buy it Feb 09 19:00:26 claim to just share it and sharing should be free after all Feb 09 19:00:57 Morgreet: press release happened few minutes after 23:59:59 CST, thus counter also counts to that time and according to mickeyl its more likely right with that than with 00:00:00 UTC Feb 09 19:01:46 Morgreet: Interesting Feb 09 19:01:59 That would definately be tough with the limited resources on the mobile though.. Feb 09 19:02:21 Typically when it comes to a very low-powered device, putting as much recovery and multipath logic into the transport as possible is the way to go. Feb 09 19:02:53 It saves on administrative overhead of having to automate or yikes, setup, outernexus multipath software that is dependent upon the OS running on the applications processor. Feb 09 19:03:18 The nice thing about iSCSI is that it has the internexus multipathing as a optional feature in the standard. Feb 09 19:04:07 So when communication paths that may be running over GSM + 802.11b fail, the active paths retry unacknowledged packets, and explictly reassign acknowledged onces. Feb 09 19:04:10 s/onces/ones Feb 09 19:04:46 One of the pieces that the linux-iscsi.org team has been working on is making all of this happen transparently so the user. Feb 09 19:05:28 ie: there is always be multiple connection login attempts fired off into the background, once one successed the nexus is established with a single communication path, say over GSM. Feb 09 19:07:26 Once the device comes within access of an 802.11 hotspot, depending upon if the device supports running GSM/802.11 at the same time, a second communication path is opened to the target node endpoint over 802.11. Feb 09 19:07:39 Without having to stop or pause the existing nexus. Feb 09 19:08:24 Once the device leaves 802.11 range, the communication paths fails, and the GSM communication path completes it's outstanding failed commands transparently to the SCSI subsystem (and applications) above. Feb 09 19:09:15 This is defined as ErrorRecoveryLevel=2 or Connection Recovery within the RFC, and is a *KEY* requirement to making iSCSI/GSM useable on a day-to-day basis. Feb 09 19:09:51 This is the primary reason why things like NFS and SAMBA are not practical weak radio link. Feb 09 19:10:42 From what you say, the iSCSI model sounds like it would be the storage method of preference for meshes. Feb 09 19:11:48 Morgreet: I can't say I have put alot of thought into that setup, but the standard has some features that make it very flexable from that standpoint. :-) Feb 09 19:13:19 Ok, wew. Feb 09 19:13:31 So I will be heading up the transport side of iSCSI/GSM. Feb 09 19:13:39 I hope the work isn't being tied to wifi+GSM. It's relevant to absolutely paths at all. Feb 09 19:13:55 any* paths Feb 09 19:13:57 There are also two UI projects underway on the Nokia770 for creating a adminstirative UI for an iSCSI initiator. Feb 09 19:14:17 Morgreet: This is all based on IP, so no. :-) Feb 09 19:14:31 Also, the project will be supporting both iSCSI/TCP and iSCSI/SCTP Feb 09 19:14:39 Which are currently in stable release. Feb 09 19:15:24 (still need to catch up): nab: about 50 developers were selected some time ago, but it might still be possible to get on list: see http://www.openmoko.com ; however it might not be possible to get into that list anymore, I don't know which is the case Feb 09 19:15:45 There is one ncurses based UI, and another GTK based interface that fits into the Hildon API for Maemo. Feb 09 19:15:57 aloril2: Ah, thank you for the info! Feb 09 19:16:02 nab: that's what got you off on the wrong foot here. Instead of talking about IP, you started talking about GSM ;-) Feb 09 19:16:14 Morgreet: :P Feb 09 19:16:40 counter Feb 09 19:16:40 2 days 10:10:43 (2.424 days) for source for *all* developers and devices for selected developers (2007-02-11); a month, 2 days (30.424 days) for devices for *ANYBODY* for $350 (targeting developers) (preorders: not yet) (2007-03-11); 7 months, 2 days (214.424 days) for mass market (2007-09-11): see topic for more info (194) Feb 09 19:17:19 Morgreet: So with the Maemo GTK based UI, I see alot of symmetry between the GUI project for the 770 and a UI for OpenMoko. Feb 09 19:18:01 nab: undoubtedly. We're just more open, in the sense of having fewer closed modules, but the base is the same. Feb 09 19:21:23 And our UI is pure Gtk, with just some new widgets for look'n'feel purposes. Feb 09 19:21:59 Both of the UI pieces are still in the early phases, so I just want to make sure that the developers are aware of what is going on with OpenMoko. Feb 09 19:22:00 :-) Feb 09 19:22:01 But of course pretty much anything else will run too, given a spot of hacking. Feb 09 19:22:40 Morgreet: I will be doing a developer release for the stack and userspace tools bits basically within a few days of getting my hands on the device. Feb 09 19:23:00 Well even we're not sure .... but it will all become stunningly clear in 2.4 days' time ;-)))) Feb 09 19:23:23 :P Feb 09 19:27:54 Morgreet: I am very excited to finally have a vehicle to pull off a project like iSCSI/GSM. :-) Feb 09 19:28:23 transparant migration from gprs to usbnet would be neat Feb 09 19:28:25 But really, I just want to have my GB of music in my pocket on my own storage. Feb 09 19:28:37 koen *g* Feb 09 19:29:09 damnit that the specter micro-sd-wlan card didnt hit the market yet Feb 09 19:29:18 woglinde: it wont work in neo anyhow Feb 09 19:29:21 doesnt fit Feb 09 19:29:30 and below the battery, reception would be awful Feb 09 19:29:39 Is that a GB of storage in your pocket, or are you just glad to see me? ;-) Feb 09 19:29:55 buz hm on the website stands micro-sd slot Feb 09 19:29:58 i dont think the girl would be happy upon seeing the size of 1G in these days Feb 09 19:30:00 Morgreet: Also, having the 2nd generation phone from FIC with 802.11b will be great too. Feb 09 19:30:04 woglinde: look at the size of it Feb 09 19:30:10 it's NOT standard microsd Feb 09 19:30:14 has overlength Feb 09 19:30:16 buz: hehe Feb 09 19:30:17 buz hm we will see next week Feb 09 19:30:24 hm Feb 09 19:30:24 I have not been able to get my hands on a GSM/802.11 Linux phone yet. Feb 09 19:30:24 woglinde: i mean the spectec thing Feb 09 19:30:35 nab: i'm SOOO waiting for that one too Feb 09 19:30:43 there's the GN1 Feb 09 19:30:46 but that's a bad joke Feb 09 19:32:10 Or rather, a GSM/802.11 Linux phone that you are only limited by your imagination, and not software licensing. ;P Feb 09 19:32:39 nab: doesn't iSCSI use TCP? Don't all the crappy default socket timeout options give you a headache for your work? Or do you turn all timeouts off to make TCP non-breaking? Feb 09 19:33:37 Wifi too, but even more so on GPRS. Feb 09 19:33:41 Morgreet: iSCSI is not strictly tied to TCP, although there where quite a few TCP considerations put into the final RFC to deal with some of the pieces that TCP lacks, mainly the lack of framing. Feb 09 19:34:31 Morgreet: Also, TCP_NODELAY is enabled for all iSCSI traffic. Feb 09 19:35:25 The initial research that will be published will also contain iSCSI/SCTP numbers as well. Feb 09 19:35:49 Although I don't personally know of any research of using SCTP over slow links. Feb 09 19:35:51 So we will have to see. Feb 09 19:35:53 So you use the O/S's default socket options, and when TCP breaks because of RF things, you just reconnect, or what? Feb 09 19:36:24 The RFC defines a couple of different options. Feb 09 19:36:46 Keep in mind that in the iSCSI Session (aka the nexus) between an initiator and target, you can have multiple TCP communication paths, or iSCSI connections Feb 09 19:37:13 In ERL != 2, when one iSCSI connection fails, lets say due to a TCP reset, all of the other connections have to be shutdown and restarted. Feb 09 19:37:47 In ERL == 2, when one iSCSI connection fails, the other communication path can retry and/or reassign the outstanding commands from the failed connection. Feb 09 19:38:05 Morgreet: So short answer, yes, it just reconnects Feb 09 19:38:34 Long answer, depending on the implementation, it will have to restart all paths, or continue moving cdbs and data. Feb 09 19:38:53 Well I sure hope TCP timeout isn't directly mapped to nexus failure, or you'll have more reconnection traffic than data traffic. Feb 09 19:39:10 The overhead for a restart is minimal, but the problem arises when you are dealing with multiple communication paths across multiple subnets. Feb 09 19:39:55 The recovery logic when all of the communication paths are forced to fail is pretty simple because the stack won't be able to determine which of the network portals (the IP endpoints) are still up. Feb 09 19:40:15 Morgreet: I fully expect there to be iSCSI and TCP tuning required. Feb 09 19:40:17 :-) Feb 09 19:41:24 nab: hmmm, well it not "tuning" in the conventional sense. RF isn't so much about speed as about making sense out of yoyoing connections. Feb 09 19:41:33 My group had done lots of iSCSI WAN testing, espically with folks in europe and the middle east, so I expect most of the issues to have already been encountered in one from or another. Feb 09 19:41:55 Morgreet: Agreed. Feb 09 19:43:08 The transport must be able to handle not only dropped connections, but recovery scenarios that involve things like application layer CRC failures, out of order acknowledgements, just to name a few. Feb 09 19:43:29 I'm looking forward to seeing what you can get working. It's certainly not the conventional use of iSCSI, but novelty is good. Feb 09 19:43:56 Luckly, the initiator stack that will be used has been stable early 2005 with the WIFI and WAN scenarios, so I am confident that no additional coding will be required. Feb 09 19:44:09 Morgreet: I have been hearing that for years. :P Feb 09 19:44:19 nab: hehe Feb 09 19:44:30 http://www.linux-iscsi.org/index.php/Playstation2/iSCSI Feb 09 19:44:31 :p Feb 09 19:45:06 I always figure the more that you hear that something cannot be done, or that it is unreasonable or unpracticle, the more you know you are on the right track. :P Feb 09 19:45:08 The FSM for state tracking in a multipathed nexus over > 1 RF links must be mind boggling. Feb 09 19:45:24 Morgreet: Not really Feb 09 19:45:32 ps2 with iscsi? wtf? Feb 09 19:45:33 Since the link is slow, there are not a whole lot of outstanding CDbs. Feb 09 19:45:39 http://www.linux-iscsi.org/index.php/Playstation3/iSCSI Feb 09 19:45:40 :P Feb 09 19:45:48 Yes, the linux-iscsi.org has done alot over the years. :P Feb 09 19:45:51 woglinde: welcome to a parallel universe :-) Feb 09 19:46:13 i wish they made a few more linux/ps2 disks :P Feb 09 19:46:15 Morgreet: The more outstanding CDBs on the failed connection, the more complex it gets. Feb 09 19:46:39 Morgreet: But if we are talking about a GSM link, I imagine we will only have 1 outstanding CDB per communication at a time. Feb 09 19:46:39 If even that. Feb 09 19:47:20 Minus the NOPs that are sent periodiaclly to help determine communication path failures as early as possible. Feb 09 19:48:17 Morgreet: Luckly RFC-3720 defines the statemachine for the scenario you mentioned, which definately helps alot toward talking interoptable implementations. Feb 09 19:49:56 wifi? Feb 09 19:49:56 When version 1 was designed there was no sufficiently low-power WiFi chip available which has an open driver. You can attach (battery) powered USB hub to Neo1973 and then use supported WiFi USB stick. For more information see http://www.linuxtogo.org/gowiki/OpenMoko/InternetAccess Feb 09 19:50:59 aloril hm thats nen option Feb 09 19:51:23 this post by sean about hardware change confuses me Feb 09 19:51:36 not wanting to commit not to add entirely new subsystems is weird Feb 09 19:53:26 because if Phase2 neo gets wifi, i'd wait Feb 09 19:55:34 Are you saying that the wifi driver situation is changing rapidly because of OLPC? Feb 09 19:55:57 morgeet I dont think so Feb 09 19:56:54 that nexperia reference design looks very interestin Feb 09 19:57:27 the s3c2410 is arm926 too, right? Feb 09 19:57:58 buz hm we have to look at samsungs website Feb 09 19:58:04 http://www.linuxdevices.com/news/NS8417344656.html Feb 09 19:58:15 I'm all for FIC making the best open hardware at any given time, without waiting for marketting phases or releases. But waiting "just in case" makes no sense at all, life is too short. Feb 09 19:58:28 looks like it can only do WVGA though Feb 09 19:58:45 i'm against waiting just in case Feb 09 19:58:52 i'm however fine with waiting for a reason Feb 09 19:59:33 technically this is more fighting with the voice in my head requesting to justify such a purchase Feb 09 20:03:05 I'm interested in VirtualLogix's virtualization stuff in that design. It has potential for bad as well as good though --- eg. bad would be if they use virt to close off access to proprietary subsystems running on the same SoC. Feb 09 20:05:21 buz: arm920t Feb 09 20:06:06 that might make for a cheap 3G neo ;) Feb 09 20:06:18 Yes, but not an open one. Feb 09 20:06:39 that we dont know for sure Feb 09 20:06:44 I can just see it happening. "Reduced bill of materials" is the selling point. But an "open phone" based on that approach does give you a host CPU to own, it gives you part of one. Feb 09 20:06:53 hm I think otherwise you cant hold the price Feb 09 20:06:58 with a better soc Feb 09 20:06:59 doesn't* give you ... Feb 09 20:07:48 if the performance is good enough, i dont i'd care Feb 09 20:07:48 buz: I do, or I wouldn't be interested in the openness of the Neo. Feb 09 20:08:00 well if it comes with proper drivers it would be open too Feb 09 20:08:00 Just out of curiousity, what is the CPU frequency scale that is supported in gen1..? Feb 09 20:08:27 266 Feb 09 20:08:27 s/scale/scaling Feb 09 20:08:27 the virtualization thing is basically "hardware" Feb 09 20:08:27 Morgreet: That is full power..? Feb 09 20:08:56 That's all, yes. This is an S3C2410 Feb 09 20:10:53 buz: you'd get sources only for your virtual guest O/S, not for the hypervisor. It would be no more open than the PS3 is open when it runs Linux. No access to the graphics for example, it's blocked by Sony's hypervisor. Feb 09 20:11:24 the plan is to get a dirt cheap v1 to get the software into shape and have more feature-rich (and hence more expensive) models later on Feb 09 20:11:25 Morgreet: Plus it disallows any ATAPI CDBs related to DVD-CSS or ACCS. Feb 09 20:11:29 Boooooo-Hisss! Feb 09 20:11:34 :-) Feb 09 20:11:36 nab: yeah Feb 09 20:11:45 that's what you get for buying sony Feb 09 20:11:52 having v1 cost €700 isn't going to get you a lot developers nor users Feb 09 20:12:10 i'd pay that much for my $dreamphone Feb 09 20:12:11 Virtualization is *great* when you're in charge of the hypervisor. It sucks rocks when you're a mere guest O/S. Feb 09 20:12:19 350 euro is expensive too Feb 09 20:12:25 buz than wait for the road phone Feb 09 20:12:32 and pay 1200 euros Feb 09 20:12:32 that's pure vapor ware Feb 09 20:12:36 no Feb 09 20:12:42 I have touched it Feb 09 20:12:42 i wont pay 1200EUR Feb 09 20:12:57 i might pay that amount for a new laptop Feb 09 20:13:02 but not a phone Feb 09 20:13:19 woglinde: so when is road planning on actually delivering those things? Feb 09 20:13:36 they sure are nifty, but the pricing is way out Feb 09 20:14:57 buz dont know, dont know why they struggeling with the software Feb 09 20:15:00 I think the current neo1973 is already too costly Feb 09 20:15:04 so long now Feb 09 20:15:19 loufouge it is a good price Feb 09 20:15:25 the price is ok Feb 09 20:15:34 tho i'd rather pay 20% more for wifi and a cam Feb 09 20:15:59 I dont need a cam Feb 09 20:16:08 but wifi Feb 09 20:16:16 a working cam would be useful Feb 09 20:16:22 and wifi reaaaaallly useful Feb 09 20:16:23 Yes, Wifi is key for me as well. Feb 09 20:16:31 Do you have any idea how much say, a Rokr or a SE W800i would cost without being subsidized/tied to a contract? $350 for a phone with those specs is very cheap. Feb 09 20:16:47 Once the device is within range of the hotspot, you can watch video from your device. Feb 09 20:16:55 sun look at ebay Feb 09 20:17:05 Morgreet: why does Sony restrict access to the hardware from linux? Feb 09 20:17:17 loufoque: because Sony is evil Feb 09 20:17:18 loufoque: Because they have a media arm of their company Feb 09 20:17:55 hm because sony is a large company and the right hand dont know waht the left hand do with linux Feb 09 20:17:58 And since they are selling those boxes at a ~250 USD loss, the can't afford to give hacks a very easy way to turn their hardware against their software interests. Feb 09 20:18:02 But anyways. Feb 09 20:18:02 :P Feb 09 20:18:13 loufoque: nab is 100% right. Sony the equipment maker is no more. They are purely a media-delivery company now, a fact that even the Japanese minister for IT has criticised. Feb 09 20:18:14 The Rokr E2 starts at $180 and it's not nearly as powerful. Feb 09 20:18:29 SuN: I picked up mine on ebay for ~220 Feb 09 20:18:34 a SE W800i roughly 250E Feb 09 20:18:38 new without contract Feb 09 20:19:52 unsubsidized as well? Feb 09 20:19:55 yes Feb 09 20:19:55 Morgreet: I thought they wanted to see various individuals participate in content creation for the PS3. If they restrict access, those individuals are rather limited. Feb 09 20:20:00 Morgreet: The problem for Playstation3/iSCSI is that all of the storage devices have been successfuly exported out over iSCSI. Feb 09 20:20:05 if it's without contract, it's unsubsized Feb 09 20:20:14 buz: not necessarily Feb 09 20:20:18 In that light, I don't think $350 for something with a screen that's 4x as sharp, GPS AND an open platform is unfair. Feb 09 20:20:22 around here it usually is Feb 09 20:20:23 Except, that under GuestOS only DVD-Rs can be successfully played under the hypervisor. Feb 09 20:20:23 buz: could be surplus subsidized phones Feb 09 20:20:29 loufoque: yes, that's what Kutaragi said, once upon a time. But he was either lying, or he was overruled. Feb 09 20:20:32 Any actual pressed DVDs that you own, cannot. Feb 09 20:20:37 Go figure that one out?! Feb 09 20:20:43 Feb 09 20:20:44 :P Feb 09 20:20:47 The subscription is what gets them the subsidy, not the phone. Feb 09 20:21:13 it's without signing a contract, i.e. you can buy it without selling your soul Feb 09 20:21:23 i dont care if someone is stupid enough to subsidize that Feb 09 20:21:42 be my guest Feb 09 20:21:45 It's ridiculous to have an operating system on a console without graphics hardware acceleration Feb 09 20:21:54 loufoque: Yes. Feb 09 20:22:09 nab: so what you mean is, Sony's protections will soon enough come to naut. Feb 09 20:22:18 loufoque: But most people who run Linux on the PS3 arent really interested in the graphics hardware I don't think. Feb 09 20:22:35 They want a cheap way to have access to the CBA. Feb 09 20:22:42 Morgreet: I can't really say. Feb 09 20:23:02 I just want to be able to play all of the media that I own on all of the devices I own from the PS3. Feb 09 20:23:04 personally i think the ps3 itself is somewhat ridiculous Feb 09 20:23:09 And I can't do that right now.. Grrrr Feb 09 20:23:09 nab: Two different camps. All the games people wanted access to the graphics. All the computing people wanted access to the Cell. Feb 09 20:23:19 Morgreet: Fair enough. :P Feb 09 20:23:31 Hell, even the PS2 Linux Kit had access to the graphics hardware. Feb 09 20:23:46 I really have no idea why SCEI prevented that in the PS3 launch. Feb 09 20:23:49 nab: if you just want to experiment the Cell, better buy the real thing. Feb 09 20:23:58 PS2 was from before Sony's death as a hardware manufacturer. Feb 09 20:24:01 loufoque: The real thing is much more expensive. Feb 09 20:24:11 does ibm actually ship cell now? Feb 09 20:24:19 loufoque: The next cheapest thing is a development board from Terasoft. Feb 09 20:24:22 buzL only on blades Feb 09 20:24:40 right so the case costs 10times what a ps3 costs alone Feb 09 20:24:46 Yep Feb 09 20:24:53 buz: Yes, also Mercury Systems is gearing up to ship a 4x 4U Cell rackmount very soon. Feb 09 20:25:08 They are the first vendor aside from Sony and IBM who will be shipping CBA based hardware. Feb 09 20:25:10 nab: yeah, but they're a military supplier. Feb 09 20:25:37 when i can get a 8 core 1u operon box i think i prefer that one ;) Feb 09 20:25:40 Morgreet, i think terrasoft is going to ship the mercury stuff too, but i'm not certain (i used to work @ TS) Feb 09 20:26:01 jebba: interesting Feb 09 20:26:14 jebba: That sounds about right. Feb 09 20:26:18 :-) Feb 09 20:26:41 they work with mercury quite often, iirc (i haven't worked there for years, but they are still on my radar... ;) Feb 09 20:26:41 loufoque: So away, the PS3 is by far the easiest and cheapest way to get your hands dirty with the SPEs. Feb 09 20:27:04 Well, unless of course you count the IBM Alphaworks Cell Simulator. Feb 09 20:27:06 buz: you can get 8-core Opterons from Sun right now, quad dual cores in the Sun Fire 4x00 series. Feb 09 20:27:10 But lets take that off the shelf for now. Feb 09 20:27:22 i know Feb 09 20:27:25 and for a while Feb 09 20:27:32 soon 16x Feb 09 20:27:58 prsonally i'm more interesting in a quad core notebook though Feb 09 20:28:01 And damn good prices from Sun too. Pity that they have no interest in selling, even when money is waved under their nose. Hopeless sales division. Feb 09 20:28:24 the fire2100 is a steal Feb 09 20:28:43 for a long while it was cheaper than even hell's P4 crapola Feb 09 20:29:08 people actually bought P4's for performance? Feb 09 20:29:10 buz: woop Feb 09 20:29:30 koen: people buying servers from dell obviously would Feb 09 20:29:32 koen: Actually, some of the later models preformed quite well when they where put into a decent chipset. Feb 09 20:29:41 Yep. X2100 is the lowest price 1U thing around, and the X2100 M2 is the cheapest full virt 1U box with Pacifica. I want one, but I'll be darned if I can get any them to sell without all manner of third-party reseller barriers. Feb 09 20:29:50 nab: compared to what? Feb 09 20:29:54 IBM's xseries with the layer model netbursts are actually extraordinary fast. Feb 09 20:30:10 nab: the IPC factor was still ridicilously low Feb 09 20:30:15 the xseries is decent Feb 09 20:30:15 koen: Comparted to a single northbridge for cache coherency traffic, memory traffic and IO traffic. Feb 09 20:30:25 but x2100 was much more bang for the money Feb 09 20:30:30 OMG 10GHz!!!! with the performance of a 2GHz p3 Feb 09 20:30:34 buz: What is interesting is that the xseries is more expesive than the pseries. Feb 09 20:30:43 pseries is ppc or what? Feb 09 20:30:44 s/expesive/expensive Feb 09 20:30:49 Feb 09 20:30:54 ppc used to be way over 5K usd Feb 09 20:30:58 buz: power, not powerpc Feb 09 20:30:59 The p505 have I/O bandwith out the ying-yang Feb 09 20:31:09 power5 is a beast Feb 09 20:31:28 next zseries will be power too Feb 09 20:31:32 Software only iSCSI has gone 1200 MB/sec on two p505s.. Feb 09 20:31:40 Crazy I/O bandwith Feb 09 20:31:42 so ibm has one line for all series i p x and z Feb 09 20:31:48 Perhaps I should look at IBM then. Maybe they don't have a sales force that is completely out to lunch like Sun. Feb 09 20:32:42 Morgreet: honestly, I would beg, borrow or steal to get my hands on p505s Feb 09 20:32:51 5200EUR for a lowend p505 doesnt seem cheaper than xseries to me Feb 09 20:33:16 I had access to a bunch of pseries machine through a partner, and it was very hard to let them go. Feb 09 20:33:16 Aye, that's pricey. Feb 09 20:33:25 buz: They are. Feb 09 20:33:36 i only care for the very cheapest price Feb 09 20:33:39 buz: That is the 2x socket 4x core module I beleve. Feb 09 20:34:18 Those also ship with PCI-X v2.0 266mhz if anyone cares. Feb 09 20:34:19 :P Feb 09 20:34:29 I have old power, power2, powerpc at the univ. They're not any fun anymore. Feb 09 20:34:41 IBM and Newisys was the only vendor to actually implement PCI-X v2.0. Feb 09 20:35:04 long live pci-e Feb 09 20:35:13 buz: here-here Feb 09 20:35:27 pci-x has always been an evil spec Feb 09 20:36:22 buz: I am waiting for 10 Gb/sec with iWARP running 16x PCIe. Feb 09 20:36:34 wtf is iwarp Feb 09 20:36:43 That is what will propel IP storage into the same level as FC. Feb 09 20:36:45 apple meets startrek Feb 09 20:37:10 buz: The primary of iWARP (Internet Wide Area RDMA Protocol) is DDP (Direct Data Placement) Feb 09 20:37:11 can you run 10gige on copper now? Feb 09 20:37:30 or is that to remain fibre only Feb 09 20:37:40 ie: enough metadata in IP packets to the receiving end can deremine the final offset into the application buffer. Feb 09 20:37:56 ie: Gets rid of the reassembly -> application buffer copy when using sockets. Feb 09 20:38:11 buz: Some switch vendors are shipping Copper 10Gb/sec now. Feb 09 20:38:19 I think one NIC vendor is as well. Feb 09 20:38:31 on cat7? Feb 09 20:38:32 But it is using the Infiband cabling, not Cat6. Feb 09 20:38:38 Not just yet. Feb 09 20:38:40 iieew Feb 09 20:38:48 i rather use fibre than infiniband stuff Feb 09 20:38:48 iieew indeed. Feb 09 20:39:04 What's infiniband really? Feb 09 20:39:16 ultra low latency network Feb 09 20:39:20 buz: I keep hearing Q3 2007 for Cat6 + 10 Gb/sec. Feb 09 20:39:27 schweet Feb 09 20:39:41 if i wasnt so lazy i'd run cat6 here Feb 09 20:39:44 That should definately reduce the BOM for 10 Gb/sec hardware significantly. Feb 09 20:39:50 buz - Ahh, that's why they keep talking about it for clustering stuff. Feb 09 20:40:06 yup clustering is the main use for it Feb 09 20:40:13 as it is for myrinet Feb 09 20:40:13 And DLing porn. Feb 09 20:40:18 (is that still around?) Feb 09 20:40:36 Elrond: It is typically used for shared memory clusters, ie: so a single application on a single machine can access all of the memory from all of the machines in the cluster. Feb 09 20:40:46 i wonder if those PCI-E cabling spec could be used for clusters Feb 09 20:40:57 think a 32way 16x PCI-E crossbar Feb 09 20:41:00 And of course the biggest benefit is that the application does not have to be modified. Feb 09 20:41:12 each linking 4*4 core opterons Feb 09 20:41:15 buz: What are the distance limitations on PCI-e cabling..? Feb 09 20:41:19 not sure Feb 09 20:41:26 only read about it on ars yesterday or so Feb 09 20:41:30 buz: I know that Newisys is working on a 64-socket operation setup. Feb 09 20:41:44 that's gonna cost like $SERIOUS Feb 09 20:41:55 4way is sort of affordable Feb 09 20:41:55 If this is using a single machine or multiple machines with a shared memory interconnect, I have no idea. Feb 09 20:42:17 buz: I imagine that you can purchase a certain number of sockets at a time and scale up with your requirements. Feb 09 20:42:18 nab, buz - I also read it on ibm's gpfs stuff. (I hope, they make gpfs once FOSS...) Feb 09 20:42:37 i rather thinker myself ;) Feb 09 20:42:47 not that i actually have any use at all for such stuff Feb 09 20:43:33 Fun stuff.:P Feb 09 20:43:48 But lets scale it back from 64 sockets to one 266mhz SoC. ;P Feb 09 20:44:06 neo beowulf Feb 09 20:44:07 hehehehe Feb 09 20:44:09 over usb1.1 Feb 09 20:44:20 over BT! Feb 09 20:44:46 grps really Feb 09 20:44:50 Someone has to do a "Neowulf" cluster just for the sake of it.. Feb 09 20:44:56 CM i agree Feb 09 20:45:03 just to get on slashdot once more Feb 09 20:45:03 Wow.. Feb 09 20:45:04 So when we meet up... our neos can start simulating half the world. ;o) Feb 09 20:45:08 And you think that iSCSI/GSM is crazy.. Feb 09 20:45:16 i wonder if emulated neos would work Feb 09 20:45:23 nab: i had two beers in the meantime ;) Feb 09 20:45:32 Oh, that sounds good Feb 09 20:45:33 suddenly iscsi over gsm doesnt seem to be that carzy anymore Feb 09 20:45:39 Oh, don't worry about 266MHz SoC. We've figured out that 266MHz SoC + FPGA + ponies == 10GHz octocore. Feb 09 20:45:46 <-- getting food now. Feb 09 20:46:07 10 transistor octocore or what Feb 09 20:46:41 Probably, "what" Feb 09 20:46:54 LOL Feb 09 20:46:58 i need to remember that one ;) Feb 09 20:47:26 I think the mobile beowulf is in line with running PoE over wifi, and the Tivo that can fast-forward live tee-vee. Feb 09 20:47:36 Something to do with the flux capacitor I think. Feb 09 20:47:41 LOL Feb 09 20:47:46 not really Feb 09 20:47:46 lol Feb 09 20:47:53 wifi and pink probably feel left out now. But only 2 days to go, they'll survive the ordeal ... Feb 09 20:47:59 the mobile beowulf is actually possible right now Feb 09 20:48:17 thining of it, so is PoE over wifi Feb 09 20:48:26 Hahahaha Feb 09 20:48:26 only living next to that wifi is not really feasible Feb 09 20:48:36 Well with enough wattage ... Feb 09 20:48:55 microwave ovens run in the 2.4ghz band right Feb 09 20:48:59 yep Feb 09 20:49:05 Morgreet: PoE+Wifi sounds more useful has a riot control mechinism.. Feb 09 20:49:22 :-) Feb 09 20:49:30 i think there was news about something like that two weeks ago or so Feb 09 20:49:32 nab: Oh I thin they're already deploying it as such Feb 09 20:49:32 s/has/as Feb 09 20:49:49 non lethal but very painful Feb 09 20:50:09 (and somehow illegal by geneva convention, apparently outright killing people is ok though) Feb 09 20:50:30 It sounds ideal for testing on the politician who wishes to authorize it. Feb 09 20:52:23 Wow, it is almost 1:30PST and I have not done a single bit of work. Feb 09 20:52:34 ok folks, catch you in a bit. :P Feb 09 20:52:45 * nab dives into his buglist Feb 09 20:53:07 hm its 12.30 pm here Feb 09 20:53:12 args Feb 09 20:53:14 10.30 Feb 09 20:53:30 cya nab Feb 09 20:53:53 Thanks for the information gents, I am looking forward to working with OpenMoko! Feb 09 20:54:11 Cheers Morgreet! Feb 09 20:54:17 "work" is something you don't enjoy. The only time techies work is when they're doing the paperwork. Feb 09 20:54:30 :P Feb 09 20:54:55 Morgreet: Work == Coding that does not involve the most interesting idea on the most 'shiny' new piece of hardware. Feb 09 20:55:19 i pray for you it's not on ms platform Feb 09 20:55:29 hehe Feb 09 20:55:30 because that's not work Feb 09 20:55:34 that's punishment Feb 09 20:55:36 Of course not. Feb 09 20:56:07 My day job is iSCSI related. ;P Feb 09 20:56:14 er, storage related. Feb 09 20:56:31 so do you pronounce iscsi as i-sexy? Feb 09 20:56:41 I-Scuzzy Feb 09 20:56:43 hehe Feb 09 20:56:47 Vrrroooommm Feb 09 20:56:59 will MACHINE=ep93xx and `bitbake gpe-image` build everything needed for bootstrapping the phone and such too? (i'm just a little impatient as it churns along...) Feb 09 20:57:40 it will package packages that are more or less 'compatible' with the phase 0 software Feb 09 20:57:48 s/will/will build/ Feb 09 20:57:48 koen meant: it will build package packages that are more or less 'compatible' with the phase 0 software Feb 09 20:58:22 i guess what i don't get is where the actual boot-code is then, if it's not part of OE Feb 09 20:58:35 MACHINE=smdk2440 is a whole lot closer to the phone Feb 09 20:58:40 * XorA|gone whistles innocently Feb 09 20:58:52 fic-gta01 even closer :)P Feb 09 20:58:58 s:P:: Feb 09 20:59:07 koen: shush, were not allowed to say that for 48 hours Feb 09 20:59:13 Development by successive approximation ... ;-) Feb 09 20:59:28 XorA|gone: I already mailed the .conf to the devel list last week :) Feb 09 20:59:50 XorA|gone: it's OK, we didn't hear it. Feb 09 21:00:26 uh, so, um, in sum, i'm kind of building a bunch of stuff that i'll have to start over with again, so these are kind of wasted cycles for now... (and I should just go back to being patient ;) Feb 09 21:01:02 jebba: just join the rest of us in random jabber to mask our impatience :-) Feb 09 21:01:27 ;) Feb 09 21:01:41 actually it's more of distraction from the fact that i have exams next week ; Feb 09 21:01:46 counter Feb 09 21:01:46 2 days 08:25:37 (2.351 days) for source for *all* developers and devices for selected developers (2007-02-11); a month, 2 days (30.351 days) for devices for *ANYBODY* for $350 (targeting developers) (preorders: not yet) (2007-03-11); 7 months, 2 days (214.351 days) for mass market (2007-09-11): see topic for more info (195) Feb 09 21:02:30 Mid-term? Feb 09 21:02:47 end term Feb 09 21:02:52 term eneded today Feb 09 21:04:27 swiss university courses in general have no mid term Feb 09 21:06:16 are there any swiss in here, btw? Feb 09 21:07:56 i'm wondering if one could save shipping witha mass order Feb 09 21:08:57 buz hm the second round will be shiped from europe too I read in a forum Feb 09 21:09:16 so shipping will not be the expensive Feb 09 21:10:04 mhh is second round phase2? Feb 09 21:10:08 i dont wanna wait that long Feb 09 21:10:13 heh, forums Feb 09 21:10:29 "bunch of clueless people too stupid to use email" Feb 09 21:11:13 koen that are people who use social group sides as facebook xing or studivz Feb 09 21:11:36 oh, we have a forum. Does this mean community ml is useful again? Feb 09 21:12:09 stefan_schmidt: better than before Feb 09 21:12:42 good Feb 09 21:12:59 stefan_schmidt: openmoko still doesn't have a forum Feb 09 21:13:19 it was in the heise.de froum Feb 09 21:13:39 koen: So we have booth no forum and calm community ml. Even better. :) Feb 09 21:14:01 stefan_schmidt: http://www.angstrom-distribution.org/unstable/images/20070207/a780/Angstrom-gpe-image-test-20070207-a780.rootfs.tar.bz2 should boot into gpe and keep the phone alive Feb 09 21:14:04 woglinde: I stopped reading heise forum 3 years ago. No time for bullshit. :) Feb 09 21:14:56 stefan hm sometimes its fun Feb 09 21:15:09 but the poster sounded reasonable we will se Feb 09 21:15:10 compared to heise, slashdot is for intellectuals Feb 09 21:15:11 e Feb 09 21:15:36 koen: Ok. I stopped all my work until university tests are finished. Cold and learning. :( Feb 09 21:15:53 Just hanging around here for some minutes before sleep. Feb 09 21:16:18 stefan which courses? Feb 09 21:17:14 woglinde: Nebenfach Psychologie and Wahscheinlichkeitstheorie (Informatik). Feb 09 21:18:22 isnt it math? Feb 09 21:18:40 I will write next week about np problems Feb 09 21:18:44 *sigh* Feb 09 21:19:02 woglinde: Yes, one of the math courses I need. Feb 09 21:21:00 mhh statistics is actually one of the more useful topics Feb 09 21:21:14 statistics is of rapidly increasing importance in most fields Feb 09 21:21:24 (not that i believed that when i had to learn it ;) Feb 09 21:21:39 hm becoming a guru in spss? Feb 09 21:21:47 yikes spss Feb 09 21:22:00 talk about bad softare Feb 09 21:22:08 dont know Feb 09 21:22:19 I only install it for people who need it Feb 09 21:22:34 i think it's mostly used in social sciences Feb 09 21:22:41 not sure why anyone would want to use it thoug Feb 09 21:24:09 These social guys use it to get a statistic-on-button-press Feb 09 21:24:58 even excel can do that Feb 09 21:24:59 spss and stata Feb 09 21:25:03 and with a much saner ui Feb 09 21:25:29 hm he spss brings his own java 1.3 vm Feb 09 21:25:41 but isn't psych also a social science topic? SCNR Feb 09 21:26:01 buz yes Feb 09 21:28:38 buz: Yes, and they also use it. *sigh* Feb 09 21:29:14 hihi Feb 09 21:45:40 | *** LD_LIBRARY_PATH shouldn't contain the current directory when Feb 09 21:45:40 | *** building glibc Feb 09 21:51:44 hey what carrier will openmoko be available for Feb 09 21:52:06 cpfr: all Feb 09 21:52:06 Any GSM, in theory. Feb 09 21:52:16 verizon is not gsm, right Feb 09 21:52:37 verizon is cdma Feb 09 21:54:43 http://www.gsmworld.com/roaming/gsminfo/cou_us.shtml Feb 09 21:54:46 For GSM carriers Feb 09 21:55:40 T-Mobile and Cingular (ATT) are the two large national carriers in the US. Feb 09 21:56:54 tiz easy, just get a gsm->cdma converter :-D Feb 09 22:00:42 | sed -e 's/gsm/cdma/g' Feb 09 22:00:51 There ya go, problem solved. Next. Feb 09 22:13:12 counter Feb 09 22:13:12 2 days 07:14:11 (2.302 days) for source for *all* developers and devices for selected developers (2007-02-11); a month, 2 days (30.302 days) for devices for *ANYBODY* for $350 (targeting developers) (preorders: not yet) (2007-03-11); 7 months, 2 days (214.302 days) for mass market (2007-09-11): see topic for more info (196) Feb 09 22:14:01 Hmm, so monday morning in the office, I will see the sources and the whole bugzilla ;o) Feb 09 22:14:51 No you won't, because OM will be Slashdotted ;-) Feb 09 22:15:01 *lol* Right. Feb 09 22:15:50 Not to worry, the Slashdot kiddies will lose interest soon. Feb 09 22:16:23 Why do you think? Feb 09 22:16:33 ... or do you mean, they will find a new attraction soon? Feb 09 22:17:00 Yep, as soon as the next article on the iPhone comes along ... Feb 09 22:17:41 hihi, yeah, okay. Feb 09 22:17:58 So at midday, I will be able to look at bugzilla ;) Feb 09 22:17:59 i didn't realise a thing until just a few days ago, and i'm still uncertain if it's true or not Feb 09 22:18:05 will OM have multi-touch or not ? Feb 09 22:18:22 i read an article that said it will but i don't remember reading about multi-touch on the official announcement Feb 09 22:18:31 Elrond - You mean s/OM/neo/. single-touch is the current answer. Feb 09 22:18:49 Huh. Feb 09 22:19:02 Eblis - You mean s/OM/neo/. single-touch is the current answer. Feb 09 22:19:15 ok, i'll search for that article again Feb 09 22:19:28 Initial v1 hardware doesn't seem to have multitouch, as it's a simple resistive 2-axis input to an ADC, we believe. But TS hardware could change at any point. Feb 09 22:19:36 i also got the impression that OM will have single touch but that article mentioned multi-touch Feb 09 22:20:14 and i thought: Wow :) Feb 09 22:20:29 multi touch might be possible in software, using trickery. Feb 09 22:20:36 Yeah? Feb 09 22:20:48 Morgreet - If we get the ADC output. ;) Feb 09 22:21:09 ... and people give us a few ADC samplings between toucu-1 and touch-2... it _might_ be possible. Feb 09 22:21:56 It could be possible if we reverse the polarity applied to the overlay, because then the ADC converts using a different 2D origin. Feb 09 22:22:10 But that's unlikely to be possible. Feb 09 22:22:32 Shouldn't the second touch change the ADC again? Feb 09 22:24:08 I guess that depends on the ADC conversion method, I guess it might work with successive approximation, but not sure how. Haven't thought about it really. Feb 09 22:24:20 The one-touch screen, that I payed ages ago in some store, used to put the cursor in between my two fingers. ;) Feb 09 22:24:25 For a single touch display, being touched again, you have to assume quite a lot about the size/shape of the touching points. Feb 09 22:24:33 Interesting Feb 09 22:24:47 Yeah - you have to assume that a touch, which instantly moves, is in fact another touch. Feb 09 22:24:50 touch is resistive, so the point will be in the middle of the tow Feb 09 22:24:53 two Feb 09 22:24:55 s/payed/played/ Feb 09 22:24:56 Elrond meant: The one-touch screen, that I played ages ago in some store, used to put the cursor in between my two fingers. ;) Feb 09 22:25:23 And so, if you get a touch, and it moves 100 pixels north instantly, it's probable that the user touched the screen again 200 pixels north Feb 09 22:25:43 SpeedEvil - Yeah, exactly. Feb 09 22:25:52 no Feb 09 22:26:06 not with this technology Feb 09 22:26:17 If the display has another couple of electrodes - and you can measure the difference in resistance between top-left and bottom-right, and bottom left and top right, then you can get a bounding box. Feb 09 22:26:44 Yes, with this technology - a simple resistive touchscreen, with two electrodes Feb 09 22:26:56 It does not give you the location of the touch. Feb 09 22:27:04 It gives you the average of all touch locations. Feb 09 22:27:26 Sort-of, with 'average' being complex mathematically. Feb 09 22:27:28 nope, you cant do multi touch as will average the x,y based on x1,y1 and x2,y2 Feb 09 22:27:44 Yes, however, in the instant that the second touch happens. Feb 09 22:27:57 and when do you know this happens ? Feb 09 22:27:57 The touch point moves to the average position. Feb 09 22:28:13 It moves _fast_. Feb 09 22:28:30 From this instant movement, you can guess that the new touch happened on the other side of the average position. Feb 09 22:28:35 http://gizmodo.com/gadgets/smartphones/fics-linuxbased-smartphone-213016.php Feb 09 22:29:04 Assuming it samples at around 1Khz, rather than 20Hz, which might make it problematic. Feb 09 22:29:11 SpeedEvil: you should write this code and let us see, it's not easy second guesing the user Feb 09 22:29:28 Oh, maybe the key is noting that the sudden change in position occurs without the initial touch disappearing. Feb 09 22:29:42 The user does not have fingers which can move halfway across the screen in half a millisecond. Feb 09 22:29:53 Or if they do, we need to report that to someone :) Feb 09 22:30:11 Morgreet - You just get a _quick_ hop of the coords. Without a release between. Feb 09 22:30:29 fwiw, TI have patented this form of touch controller. Feb 09 22:30:31 It's not second guessing the user. It's trying to guess the users intent, past the mask of poor hardware. Feb 09 22:30:37 Sigh. Feb 09 22:30:50 * SpeedEvil burns all patent examiners at the stake. Feb 09 22:31:01 Ok, so we lose einstein, but it's worth it. Feb 09 22:31:02 Elrond: indeed, that could be the trick. Because if there was release first, then it's not multitouch, so you then have to assume that it's just another 1-finger touch. Feb 09 22:31:28 So yeah, dynamic might allow it. But that requires a lot of CPU! Feb 09 22:31:34 And if it's configurable on or off, it's hardly objectionable to the user if it works better for 99% of users. Feb 09 22:31:34 there are no cdma gsm convetors Feb 09 22:31:48 did anyone look at the link i pasted ? :) Feb 09 22:31:53 I think that was a joke cpfr :) Feb 09 22:31:59 SpeedEvil: it is, have you even written any touch drivers, it's not that easy as you suggest :( Feb 09 22:32:05 i know Feb 09 22:32:38 lrg: no - it's horribly harder. But you can probably get _some_ information out of a multiple touch - even if you can't get the full coordinates. Feb 09 22:32:59 lrg: I think we're just trying thought experiments to see if it's *possible* at this point, rather than easy ;-) Feb 09 22:33:19 For example, move finger across the screen, to select something, then tap another finger on the screen to actio n it. Feb 09 22:33:32 "possible at all". Assuming we don't need the cpu for anything else, etc. Feb 09 22:33:32 Think maybe a scrollbar at the bottom. Feb 09 22:33:59 you move your finger to move the scrollbar, and when something is highlighted, tap with another finger on the display. Feb 09 22:34:14 hihi. Feb 09 22:34:35 It depends - you might get a nice stream of interrupts from the ADC. Feb 09 22:34:42 In which case it's not too bad. Feb 09 22:34:57 Considering that you only need to do this when the screen is touched. Feb 09 22:35:09 http://gizmodo.com/gadgets/smartphones/fics-linuxbased-smartphone- Feb 09 22:35:09 213016.php Feb 09 22:35:11 gah Feb 09 22:35:26 SpeedEvil: how do you know when the screen is touch the second time ? Feb 09 22:35:31 Hey guys, multitouch isn't a unique magic solition just because Jobs said so ... you're really just using one finger to set a mode. And modes can be set in other ways too. Feb 09 22:36:01 Because the coordinate you get instantaneously jumps from one touch position to anotehr, then back again. Feb 09 22:36:14 SpeedEvil: you dont get a second irq Feb 09 22:36:25 You don't know the hardware. Feb 09 22:36:32 SpeedEvil: and how do you tell this isnt analog noise ? Feb 09 22:36:41 It could be that you can set it up so that you get an interrupt whenever it changes. Feb 09 22:36:49 Or it could be that you have to poll. Feb 09 22:37:03 Analog noise generally doesn't move the touch point significantly. Feb 09 22:37:12 (based on trying to draw on touchpads) Feb 09 22:37:29 SpeedEvil: it does (based on driver development) Feb 09 22:37:44 And the new position will keep for a few sampings at least. That will say "this was not noise". Feb 09 22:37:51 Yeah. Feb 09 22:38:13 There _is_ systematic noise that occurs in very worn touchpads that will screw this. Feb 09 22:38:24 if it jumps like mad, we have to assume "child touching mums neo" ;o) Feb 09 22:38:24 But... V2 is a cure for that. Feb 09 22:38:31 it's probably best to wait and see :) Feb 09 22:38:43 but you still are assuming the first finger is still being pressed Feb 09 22:38:57 Yes. Feb 09 22:39:17 Ok - I just did some tests, with my ancient handspring visor. Feb 09 22:39:20 And a drawing app. Feb 09 22:39:56 fwiw, neo touch is single finger only Feb 09 22:40:12 so we don't have to think about it atm ;) Feb 09 22:40:15 lrg - What will it do, when two fingers are on it? Feb 09 22:40:19 It works as expected - draw a line up the screen with a finger, then tap off to the left of that line, you get a new line between the new touch and the old line, which then goes back to the original line when you remove it. Feb 09 22:40:34 Elrond: middle point Feb 09 22:40:40 Sigh. Feb 09 22:40:57 lrg - So we have a jump. ;) All we need. :) Feb 09 22:41:17 Yes, this is _guessing_! Feb 09 22:41:39 If there are more than two touches and we assume two, we're screwed. Feb 09 22:41:48 But the point is not to get an accurate second touch. Feb 09 22:41:48 SpeedEvil - Right. Feb 09 22:42:00 The point is to be able to click while dragging. Feb 09 22:42:03 (sort-of) Feb 09 22:42:18 SpeedEvil - But we're scrwed anyway. :) Feb 09 22:42:25 Being able to select something without removing finger, and trying to replace in same place. Feb 09 22:43:56 As I said: When the thing hops like mad, we have to assume (small-child-got-mums-neo)-mode. ;o) Feb 09 22:44:38 Is the neo small-child safe? ;o) Feb 09 22:44:39 Morgreet: i think you were talking about marketing ealier today :) http://www.youtube.com/watch?v=Go_VtqtxCHY Feb 09 22:47:04 Eblis: I can't see it, it's flagged and I don't have a login. :-( Feb 09 22:47:35 create an id ? (www.pookmail.com :) ) Feb 09 22:48:38 Eblis: can't be asked to make a login for every damn site that wants registration ;-) Feb 09 22:48:53 bugmenot.com ? Feb 09 22:48:54 :) Feb 09 22:48:58 Hehe Feb 09 22:50:51 *lol* There's a website for any crap, right. :) Feb 09 23:10:11 night all Feb 09 23:10:28 night too Feb 10 00:08:43 Night people. Feb 10 00:08:51 WAve. Feb 10 00:09:13 *waves* Feb 10 00:11:34 count Feb 10 00:11:45 Where's that countdown bot? Feb 10 00:11:54 counter Feb 10 00:11:55 2 days 05:15:27 (2.219 days) for source for *all* developers and devices for selected developers (2007-02-11); a month, 2 days (30.219 days) for devices for *ANYBODY* for $350 (targeting developers) (preorders: not yet) (2007-03-11); 7 months, 2 days (214.219 days) for mass market (2007-09-11): see topic for more info (197) Feb 10 00:12:01 That's more like it! **** ENDING LOGGING AT Sat Feb 10 02:59:57 2007