**** BEGIN LOGGING AT Fri Feb 02 02:59:59 2007 Feb 02 03:18:54 counter Feb 02 03:18:54 a week, 3 days 02:09:22 (10.09 days) for source for *all* developers and devices for selected developers (2007-02-11); a month, a week, 3 days (38.09 days) for anybody (targeting developers) (preorders: not yet) (2007-03-11); 7 months, a week, 3 days (222.09 days) for mass market (2007-09-11): see topic for more info Feb 02 04:01:56 counter Feb 02 04:01:56 a week, 3 days 01:26:20 (10.060 days) for source for *all* developers and devices for selected developers (2007-02-11); a month, a week, 3 days (38.060 days) for anybody (targeting developers) (preorders: not yet) (2007-03-11); 7 months, a week, 3 days (222.060 days) for mass market (2007-09-11): see topic for more info Feb 02 04:02:06 damn that's annoying. Feb 02 04:02:33 i reckon Feb 02 04:02:57 LoRez: hehe, I changed it slightly again ;-) Feb 02 04:19:51 any openmoko team folks around? Feb 02 04:32:56 *crickets* Feb 02 06:12:38 counter Feb 02 06:12:40 a week, 2 days 23:15:37 (9.969 days) for source for *all* developers and devices for selected developers (2007-02-11); a month, a week, 2 days (37.969 days) for anybody (targeting developers) (preorders: not yet) (2007-03-11); 7 months, a week, 2 days (221.969 days) for mass market (2007-09-11): see topic for more info Feb 02 06:29:13 aloril: That google video of Bob Brodersens lecture was interesting, thanks :) Feb 02 07:42:40 is that TRIsoft from Trisoft ? Feb 02 07:43:43 is that TRIsoft_ from Trisoft ? Feb 02 07:44:21 Yes, it is :-) Feb 02 07:45:08 nice .. how is the Lisa stuff .. is that still alive ? Feb 02 07:45:52 *lookspuzzled* Lisa stuff ? We're the Zaurus guys... Feb 02 07:52:13 oh ... didn you guys some years back also sell Ipaq`s with the strange lisa linux ? Feb 02 07:52:16 sorry then Feb 02 07:52:29 ah right ,,i bought my old zaurus back then from you Feb 02 07:58:10 ah, right. But the Ipaqs came from Lisa Systems in Hamburg. **** BEGIN LOGGING AT Fri Feb 02 08:03:59 2007 Feb 02 08:34:53 "I'm cuirous too, if we're lucky Marcin or Stefan will have a bit of news that will help us out." - why people on openmoko-community think that I have more about openmoko then they? :D Feb 02 08:34:57 morning Feb 02 08:35:14 morning Feb 02 08:35:38 Maybe because you're hanging around here all the time ? Feb 02 08:37:15 25 hits from planet openmoko to my website and it will be top referrer instead of oesf forums Feb 02 08:37:19 ;D Feb 02 08:58:13 hi all Feb 02 09:03:13 hi slava Feb 02 09:03:55 i'm interested in porting my language to the openmoko phone when it becomes available Feb 02 09:04:31 slava: language? Feb 02 09:04:39 http://factorcode.org/ Feb 02 09:06:22 slava: to quote koen: Follow the instructions on http://openembedded.org, MACHINE=ep93xx, DISTRO=generic should build compatible binaries. Feb 02 09:06:45 its not as easy as that :) Feb 02 09:06:46 until we have source code release, it has some additional stuff but is mostly based on openembedded Feb 02 09:06:50 i'm still working on the ARM compiler backend Feb 02 09:06:55 should be done in a few adys Feb 02 09:07:03 mostly i'm interested in writing libraries for accessing the cellular and gps hardware Feb 02 09:07:15 slava: yeah, I saw, but still.. I guess you eventually need above info Feb 02 09:07:16 ah that ep93xx machine ;d Feb 02 09:07:41 ah.. in that case you need to wait about 10 days Feb 02 09:07:52 will your arm9 cpu have an fpu? Feb 02 09:07:52 counter Feb 02 09:07:52 a week, 2 days 20:20:30 (9.848 days) for source for *all* developers and devices for selected developers (2007-02-11); a month, a week, 2 days (37.848 days) for anybody (targeting developers) (preorders: not yet) (2007-03-11); 7 months, a week, 2 days (221.848 days) for mass market (2007-09-11): see topic for more info Feb 02 09:08:18 slava: no: http://www.openmoko.com/press/index.html -> http://www.samsung.com/Products/Semiconductor/MobileSoC/ApplicationProcessor/ARM9Series/S3C2410/S3C2410.htm Feb 02 09:09:06 ok thanks Feb 02 09:10:29 qemu is emulator (don't recall arguments or so, but they have been mentioned in mailing list (I guess searching for qemu at archives should work: http://lists.openmoko.org ) Feb 02 09:10:36 ) Feb 02 09:11:08 i'm running qemu and i also have a gumstix Feb 02 09:11:43 i planned on doing an arm port for a while and started a few days ago, then i came across the openmoko Feb 02 09:14:54 will it support GSM or GPRS only? Feb 02 09:16:44 GSM and GPRS only Feb 02 09:16:55 is GSM the same thing as GPRS (forgive my ignorance)? Feb 02 09:17:23 no Feb 02 09:17:26 GSM voice, data connections, fax, etc..: GSM is reserved connection Feb 02 09:19:05 GPRS is packet protocol (usually IP) Feb 02 09:19:11 oh, ok Feb 02 09:30:44 slava: Hi. I've just been reading your site. Interesting. Like a sort of Lua combined with Smalltalk but with a default set of libraries ... Feb 02 09:34:50 mornin' Feb 02 10:09:17 slava: where's the source code to the various images? I've got 'f' compiled, but I'm not going to run it on an unknown binary image. Feb 02 10:09:29 the images are built from the source code in the distribution Feb 02 10:09:37 you need a working factor build new images Feb 02 10:10:15 Does "a working factor" mean a working 'f'? Feb 02 10:10:27 no, it means a f plus bootstrapped factor.image Feb 02 10:11:29 But factor.image* are binaries. I want the source code of one. Can't start with a binary to generate a binary. :-) That's what killed Smalltalk. Feb 02 10:11:54 the factor.image and boot.image.* are generated from source. but the parser/compiler is written in factor Feb 02 10:12:04 now do you build gcc without having gcc? :) Feb 02 10:12:59 Using an earlier gcc, which is a standalone program. Feb 02 10:13:18 well, you need factor to build factor Feb 02 10:13:50 i think java killed smalltalk, not the nature of the implementation. and javac is written in java, you need sun's jdk installed to compile sun's jdk :) Feb 02 10:13:54 The image isn't the equivalent of gcc though. It's the equivalent of all the runtime libraries dumped together, plus the 'f' that you compile independently. Feb 02 10:14:03 'f' isn't part of the image file Feb 02 10:14:19 I know, I've compiled it. What is 'f', conceptually? Feb 02 10:14:36 its a loader, GC, and interpreter Feb 02 10:14:49 when you bootstrap, the compiler runs in the interpreter and compiles code in the boot.image to producce a factor.image Feb 02 10:15:48 OK, I understand that. Unfortunately, it presents a security problem, because the contents of boot.image are unknown in advance. Feb 02 10:16:12 java is crappy Feb 02 10:16:24 Is there a standalone pretty printer at least, so that the contens of the boot.image can be examined? Feb 02 10:16:29 Morgret: the contents of the sources are in a sense unknown too, unless you happen to learn the language and study everything Feb 02 10:16:46 the image format is documented in vm/layout.h; get a hex dump tool and start digging :) Feb 02 10:17:21 a pretty printer would be easy to write, though. Feb 02 10:17:44 Well, your dev approach has cycles. Why not use a DAG approach and make the runtime image a composition of more primitive elements each of which can be produced independently? Feb 02 10:18:00 because that would entail writing large parts in C, such as the parser and core data structures Feb 02 10:18:01 What are you talking about? Feb 02 10:18:03 hi all Feb 02 10:18:11 heard about openmoko on lugradio Feb 02 10:18:21 sounded cool so I thought I'd pop in and say hi Feb 02 10:18:23 Elrond: we're talking about slava's language "factor" Feb 02 10:18:49 Morgret: when one factor image builds another factor image, nothing is 'carried over'. its built up completely from sources Feb 02 10:18:53 cbx33 - Hi. Feb 02 10:18:56 its just that, eg, the parser is written in factor Feb 02 10:19:10 so 'f' cannot parse sources by itself Feb 02 10:19:35 Hmmm. Feb 02 10:19:39 slava: yeah, I'm fine with that. The stumbling block is not being able to see what's inside the pill before you swallow it. Feb 02 10:20:02 the thought of having linux running on a phone so I can use a bash shell...is just awesome ;) Feb 02 10:20:04 when you download a piece of software, do you inspect every line of source code for potential vulnerabilities? ;) Feb 02 10:20:16 slava good point Feb 02 10:20:29 slava: I don't inspect every last piece of code ... BUT I COULD IF I WISHED. Feb 02 10:21:36 slava: your security issue could be addressed simply by providing an image examiner, *not* written in Factor. Feb 02 10:21:47 Right. One of the ideas of (free) opensource. Reviewability. Feb 02 10:22:15 if somebody wishes to write such a tool, they're free to do so. personally i'm more interested in improving the compiler, libraries, GUI, etc. Feb 02 10:22:20 How does one bootstrap factor on a plain box? Feb 02 10:22:32 you use a boot image from the web site Feb 02 10:22:41 slava: for most languages it's not an issue, because their base code is text. Here yours is a binary format, so you have an issue. You ought to find a solution. Feb 02 10:22:42 aka binary? Feb 02 10:22:53 the image is not the 'base code'. Feb 02 10:22:57 the base code is plain text source. Feb 02 10:23:40 I've compiled up f. Now I want another piece of plain text source to start building up a small binary image, compositionally. Apparently, I can't. Feb 02 10:23:49 right, you can't. Feb 02 10:24:13 slava: so, Factor has a problem, for anyone who doesn't dismiss security as a valid concern. Feb 02 10:24:17 Huh. I'm missing something here. Feb 02 10:24:35 gcc has the same problem. Feb 02 10:24:37 so does the linux kernel Feb 02 10:24:40 slava: which is a pity, because I think Factor sounds terrific. Feb 02 10:25:18 slava - gcc can be compiled using another c compiler. Feb 02 10:25:30 Elrond: yep, good point Feb 02 10:25:31 anybody can implement a new factor compiler Feb 02 10:25:51 gcc comes in source only. ;) Feb 02 10:26:09 the boot image doesn't even have machine code. Feb 02 10:26:22 its just a serialized object graph of the source code in core/, generated by core/tools/image.factor Feb 02 10:26:47 So there is some sort of assembler for factor? Feb 02 10:27:05 Which converts "bytecode" to machine code? Feb 02 10:27:14 when you bootstrap, the compiler compiles all library definitions to machine code and saves a factor.image Feb 02 10:27:18 slava: anyway - to get software to openmoko it has to be buildable in a way that if I start from clean dir then I fetch/unpack/patch/configure/compile all dependencies, toolchain and finally package. Feb 02 10:27:20 the compilation step is optional Feb 02 10:27:40 so how do you plan on building gcc? Feb 02 10:28:13 Using a cross compiler. :) Feb 02 10:28:23 Yeah, I understand how it hangs together. Slava, it's cool. Unfortunately, I can't run it, for the reasons I gave, which is a pity (for me) because I'd like to. The unknown binary image needs to be composable from smaller visible elements first, or at the very least convertible into a viewable form. Feb 02 10:28:47 i'm not pushing for factor to be included with the openmoko distribution, anyway. Feb 02 10:28:53 i'm just interested in putting some time into porting it over. Feb 02 10:29:03 slava: factor just need the trust path. How did you bootstrapped it from the beginnings? Feb 02 10:29:13 s/ed// Feb 02 10:29:14 dottedmag meant: slava: factor just ne the trust path. How did you bootstrapped it from the beginnings? Feb 02 10:29:22 it used to be implemented in java, and the java implementation would run entirely from source Feb 02 10:29:26 this was back in 2004 Feb 02 10:29:32 slava - Can you give me a short (three sentence) intro on factor? Feb 02 10:29:41 the first image for the native implementation was generated by the java implementation Feb 02 10:29:44 factor.sf.net Feb 02 10:29:52 ahh. Feb 02 10:30:22 slava: well, it will enormously helpful if you can restore this implementation which can build at least subset of factor enough to compile full image. Feb 02 10:30:54 the java implementation is too out of date now. i haven't worked on it for 3 years Feb 02 10:31:24 I love the idea behind it. It's just blocked by a security/trust problem. You've done a good job building up a bootstrap path, but you need a deconstruct path too. Security is a concern these days. Feb 02 10:32:01 squeak smalltalk and various lisps such as sbcl follow a similar approach Feb 02 10:32:31 I know, that's why I chucked out Smalltalk here. I loved the language. Feb 02 10:32:34 i understand security is a concern; are you worried that I could hypothetically be inserting backdoors into the boot image? Feb 02 10:32:47 or that the boot images on the web site are comporomised? Feb 02 10:33:02 slava: not worried about you personally. But one needs to be consistent. Feb 02 10:34:51 I especially love Factor's philosophy of "If you can't understand the stack, your code segment is too long". Feb 02 10:35:01 Yep, web site compromises weren't too uncommon in the old days. ;) Feb 02 10:37:51 i guess you're ruling out anything other than gcc, and a handful of scripting languages Feb 02 10:37:55 for your own use Feb 02 10:39:22 Just an image walker would probably be enough to provide the trust, as long as the image structure is defined and can't contain hidden inserts. I don't think recreating an alternative implementation is viable. Feb 02 10:39:51 Right. Feb 02 10:40:43 I still haven't really gotten the runtime model... Feb 02 10:41:09 its something i'll keep in mind for the future. i don't have a lot of time or resources right now. Feb 02 10:43:28 Elrond: the part written in C is an executable named 'f'; when you run it, it loads an image into memory and relocates it to its final address. this becomes the heap Feb 02 10:43:58 the heap holds data and code and is managed by a garbage collector and can be saved back to disk at any time Feb 02 10:44:11 you can also generate a new image from scratch, instead of taking a snapshot of the current heap Feb 02 10:44:29 where 'from scratch' means running a program inside your image which loads all the sources and serializes them in the correct format Feb 02 10:45:06 Ahhh. Feb 02 10:45:20 So how do you port to a new CPU? Feb 02 10:45:50 the C part has an interpreter, which is trivial to port. Feb 02 10:46:00 to get the interpreter running, i emit a boot image with the correct word size and endianness. Feb 02 10:46:07 then i port the compiler Feb 02 10:46:41 slave: You're porting the compiler from its binary. Feb 02 10:46:55 So the boot images aren't really CPU dependent, but only word size / endianness dependent? Feb 02 10:47:12 Elrond: right. Feb 02 10:47:32 the files on the web site have cpu extensions because the relevant compiler backend is pre-loaded Feb 02 10:47:39 but one can do ./f boot.image.x86 -no-compile on an ARM, for instance Feb 02 10:48:14 So you really could put a boot.image.32bit-little-endian up? Feb 02 10:48:21 yes Feb 02 10:48:38 i could also create minimal boot images without any libraries other than the bare minimum to load other libraries. Feb 02 10:48:40 And the c runtime has an interpreter for that? Feb 02 10:48:42 this is very easy and i've done it before, just for fun Feb 02 10:48:58 Elrond: yes Feb 02 10:49:10 Ahh. Feb 02 10:49:25 It reminds me a bit of the labview runtime model. ;) Feb 02 10:49:48 slava: given such a barebones image, how would it be extended? I ask, because it it can then you have a means of growing your image compositionally, instead of used an opaque lump. Feb 02 10:50:04 you can load new code into a factor instance at any time. Feb 02 10:50:22 i'm doing this as we speak, testing bits of the ARM compiler, changing the source, and reloading it, without shutting down and restarting the environment. Feb 02 10:51:23 So there could be a minimal image, which only has the parser in it, and enough to parse/extend itself? Feb 02 10:51:25 slava: there's your solution then -- defactor everything into individual uncoupled images, write an image walker to display what' inside of each. Then let images grow by composition of only those parts needed. Feb 02 10:51:30 does anyone have the possibility to test Mono on an arm cpu ? Feb 02 10:51:47 Elrond: yes Feb 02 10:51:54 i've built such an image, its around 500kb Feb 02 10:52:20 That, and a image-pretty-printer would make Morgret happy. :) Feb 02 10:53:39 One could then compare the pretty printed with the shipped sources and say "Okay, there are no easter eggs in the image. I'm happy." Feb 02 10:53:55 Hehe. Well I love the idea behind the language. Just got a security concern here, and it's a concern that everybody else *should* have too, although of course that's for them to decide. Feb 02 10:54:21 Elrond: dumping the image won't give a byte for byte reproduction of the source Feb 02 10:54:36 it is possible to reconstruct source from the image -- you can use the 'see' function inside factor to do that Feb 02 10:54:43 slava - Yes. But a structural. The rest is Morgret's problem. ;) Feb 02 10:54:45 however you don't get the same whitespace, etc Feb 02 10:54:55 and comments are lost Feb 02 10:55:14 Yeah, yeah. Feb 02 10:55:37 How hard would it be to write see as a standalone program in *any* other language? Feb 02 10:55:48 s/see/"see"/ Feb 02 10:55:50 Elrond meant: How hard would it be to write "see" as a standalone program in *any* other language? Feb 02 10:55:57 'see' isn't an image dumper, its a function which takes a function as as parameter and shows its source Feb 02 10:56:02 but to answer your question, not very hard Feb 02 10:56:27 slava: how come that comments would be lost? The image contains the full source as I understand it, but fully tokenized. This must contain the comments too, surely. Feb 02 10:56:38 no it doesn't contain the full source. Feb 02 10:56:46 Well, that tool could be extended to recursively walk all functions, that are referenced from the main entry function. Feb 02 10:56:49 slava: so where is the source? Feb 02 10:56:54 the source is in text files Feb 02 10:57:28 Morgret - The images on the web site are pre-compiled. ;) Feb 02 10:57:30 So ... we *do* have source for the image? Feb 02 10:57:35 Yes. Feb 02 10:57:40 yes, but you need to run an image to produce a new image. Feb 02 10:57:43 But you can't compile them on a bare system. Feb 02 10:58:04 Unless you write a compiler in some other language. Feb 02 10:58:25 So you have two options: Feb 02 10:58:36 - Either write a pretty printer and compare with shipped sources. Feb 02 10:58:45 - Or write another compiler. Feb 02 11:02:00 It needs addressing, slava. You see, if this ever became popular, and virus/malware/spyware/crapware/crime writers ever discovered it ... it would be absolutely end of story. No means of verifying what one has running locally. Feb 02 11:03:06 Morgret - Well, except for the boot-strapping issue, you can verify all sources. Feb 02 11:03:09 have any crapware writers discovered squeak and managed to inject malicious code into the base system? Feb 02 11:03:28 slava: you know that's a logical fallacy :-) Feb 02 11:03:35 Morgret - So slava has a little point in pointing us at gcc: How did you install the box you're sitting front of? ;o) Feb 02 11:04:31 Morgret: yeah, but its 6:30am and i got sucked into a long, drawn out discussion on irc. and my development is proceeding slowly because qemu isn't a speed demon. :) Feb 02 11:04:34 takes a long time to run tests Feb 02 11:05:12 ;) Feb 02 11:05:21 counter Feb 02 11:05:21 a week, 2 days 18:22:52 (9.766 days) for source for *all* developers and devices for selected developers (2007-02-11); a month, a week, 2 days (37.766 days) for anybody (targeting developers) (preorders: not yet) (2007-03-11); 7 months, a week, 2 days (221.766 days) for mass market (2007-09-11): see topic for more info Feb 02 11:08:26 i need to reflash my gumstix. qemu is far too slow Feb 02 13:03:23 can one use udp to send packets to GRPS phones? Feb 02 13:03:52 or is that prevented by nat? Feb 02 13:06:50 buz: it's operator-specific Feb 02 13:07:12 buz: Orange-France, for instance, allowed such stuff in 2005... whether that's still the case though... Feb 02 13:07:30 im wondering how much traffic jabber on grps would ause Feb 02 13:07:30 to be just online 16h a day Feb 02 13:07:42 messages would obviously be on top of that Feb 02 13:09:09 i guess it would depend on the timeout of a tcp connection first and foremost Feb 02 13:12:06 http://www.freesoft.org/CIE/RFC/1122/114.htm Feb 02 13:12:16 mhh does that mean a sane NAT server would hold open TCP for at least 2 hours? Feb 02 13:13:11 maybe i should go dig for my good old tcp ip illustrated Feb 02 13:22:47 buz - Straight TCP should have NO timeout. Feb 02 13:23:32 TCP sessions don't break *EVER* on principle, but a timeout can be specified in socket options. Unfortunately quite a while back networking libraries and/or TCP stacks started making non-inifinite timeout the default socket option for TCP, as a result of which TCP is nowadays flakey by design, for the simple pragmatic reason of self-cleanup of old sessions. Feb 02 13:23:33 Of course, many people enable keep-alive, or something alike in the application layer. Feb 02 13:23:50 Isn't the max timeout 10 hours? Feb 02 13:23:58 Only in the socket options Feb 02 13:24:16 Ah - I'm probably thinking of max RTT Feb 02 13:24:30 TCP used to be 100% reliable ... no breakage ever, regardless of how long you were at max retransmission interval. Feb 02 13:24:55 I mean, it fails in terms of bandwidth use if the link you are on exceeds 10 hours RTT Feb 02 13:24:58 But that was back in the old days, before the socket options thing. Feb 02 13:25:27 Because it can't back off far enough adn will always retransmit packets. Feb 02 13:25:37 Yeah, but only as a socket option. Feb 02 13:26:30 I mean the RTT counter only goes up to 10 hours - it's impossible in tcp/ip to have a longer one. Feb 02 13:26:35 Cleanup of effectively dead streams used to be considered an application problem. In practice it no longer is, because long-dead streams will self-break. It's bad on principle, but good in practice. Feb 02 13:26:57 This is only generally a problem for ip-via-mail, or space apps though. Feb 02 13:28:43 buz: FWIW, current phone IM clients for multi-protocols tend to use a proxy host Feb 02 13:28:48 It's also a problem for covert communications, using very low bandwidth covert channels. Feb 02 13:28:56 buz: specifically to reduce traffic Feb 02 13:29:39 It'd be nice if there was some way to authenticate packets. A router for a host will only accept packets with a given signature, which is revokable, and many signatures per host are possible. (say 256) Feb 02 13:29:52 So I'm guessing the spooks don't use TCP anyone. It's actually a good concept to have a fully reliable, non-breaking stream capability, decoupled from data rate. Feb 02 13:30:00 Yeah. Feb 02 13:30:18 Or just tweak the rtt algorithm on stuff you're using to alter the RTT algorithm. Feb 02 13:30:41 I did that once - to limit backoff to 1 second, when I was having trouble with packet loss. Feb 02 13:31:54 Well you can always set your comms app socket options to 0, and TCP will remain retrying forever. But you can't use default comms apps for it anymore, as they'll use default socket settings. Feb 02 13:33:04 After all, TCP only really exists at the two endpoints. For everyone else in between, they're just IP datagrams, nothing to break :P Feb 02 13:33:36 Unless the boxes in between are "smart" ;o) Feb 02 13:34:01 Yeah, they could send a reset back. Feb 02 13:34:55 It's one of the problems with stateful inspection in firewalls ... effectively they encode networking *policy*. It's not particularly good, just pragmatic. Feb 02 13:35:28 Right. Feb 02 13:36:45 SpeedEvil, your revokable packet signature exists as part of IPSec. Feb 02 13:37:59 * SpeedEvil stabs people that steal his ideas before he has them. Feb 02 13:38:02 :) Feb 02 13:38:34 * nnpiggy runs away screaming! Feb 02 13:38:35 :-) Feb 02 13:38:40 nnpiggy - Do you have any experience with ipsec? Is it still a nighmare to setup? Feb 02 13:38:46 Interestingly, socket timeouts will raise there head again as soon as the Internet starts getting connected to the planets ;-) Feb 02 13:39:03 their* Feb 02 13:39:11 Morgret - *biggrin Feb 02 13:39:11 * Feb 02 13:39:21 ssh will be fun. ;o) Feb 02 13:39:43 http://www.ipnsig.org/ Feb 02 13:39:44 Yeah - the moon isn't so bad. Feb 02 13:39:56 Yes, IPSec is still a pain to set up. Feb 02 13:40:58 And ipnsig is not a joke or purely speculative activity... Feb 02 13:41:00 Neo2050 ... GSM-9.5G using asteroid-belt GPS. Feb 02 13:41:32 Morgret - *lol* ;) Feb 02 13:41:40 Neo2010 ;o) Feb 02 13:42:04 Well you never know, this might be the start of something big ;-) Feb 02 13:44:31 counter Feb 02 13:44:31 a week, 2 days 15:43:41 (9.655 days) for source for *all* developers and devices for selected developers (2007-02-11); a month, a week, 2 days (37.655 days) for anybody (targeting developers) (preorders: not yet) (2007-03-11); 7 months, a week, 2 days (221.655 days) for mass market (2007-09-11): see topic for more info Feb 02 13:45:04 less then 10 days and we'll have agpsd in our hands. ;) Feb 02 13:46:19 i'll have to wait a little longer.. i was a bad dev and blew all my money for the month already Feb 02 13:47:17 hm.. or i could just download the framework and see if the gps data can be simulated. Feb 02 13:47:40 You can surely just strap a standard GPS in, or use a simulator. Feb 02 13:47:46 BTDT. Feb 02 13:47:51 leventhal - For real hw, people have to wait 10 days + shipping anyways. ;) Feb 02 13:47:54 simulator it is Feb 02 13:48:12 + shipping + dancing up and down time. Feb 02 13:48:31 and looking at the software Feb 02 13:48:37 objdump --disassemble-all. ;o) Feb 02 13:48:44 i just hope there will a non orange theme Feb 02 13:48:52 you get the software when it's shipped. Feb 02 13:48:57 So before you get it - prolly Feb 02 13:49:34 buz - *duck* Create the theme yourself. ;o) Feb 02 13:57:15 The standard theme won't matter until September. It's only then that it needs to sell on the basis of looks. Feb 02 13:57:48 I bet there will be 100 themes before then ;-) Feb 02 13:58:02 slava: you need to get factor included in Fedora, Debian, etc.: then its in same situation as gcc ;-) (no need for images, just use system suplied factor compiler) Feb 02 13:59:07 aloril2 - hehe ;) Feb 02 13:59:14 aloril2: yep, trust by association is also valid trust. However, that doesn't make the problem of hacks go away. If your image is compromised, you need some way of cleaning it without running it. Feb 02 13:59:19 Morgret - Only 100? ;) Feb 02 13:59:50 Elrond: hehe, OK, as many these as there are people ;-) Feb 02 13:59:59 s/these/themes/ Feb 02 13:59:59 Morgret meant: Elrond: hehe, OK, as many themes as there are people ;-) Feb 02 14:02:35 Hi I was just on the openmoko website but it doesnt have alot of usefull info... what exactly is openmoko? Feb 02 14:03:25 It's the same thing as having a compromised root volume. The first thing you do is turn off the power switch, no waiting, no clean shutdown, just BANG, off. Then you bring up another box and mount the old root elsewhere, and start poking. Feb 02 14:04:10 Tantalum: Did you check http://openmoko.com/press/index.html ? Feb 02 14:05:44 cool I'v heard about that, I remember the pictures. Feb 02 14:07:01 if your / is compromised and you're not into investigating it, you simply nuke it booting froma livecd Feb 02 14:07:16 I tried google but there doesnt seem to be much in the way of documentation other then the spec sheet ;-( Feb 02 14:08:38 Tantalum: Look at http://lists.openmoko.org/pipermail/announce/2007-January/000000.html Feb 02 14:09:58 I need a login for the openmoko wiki?! Feb 02 14:10:21 counter Feb 02 14:10:21 a week, 2 days 15:17:53 (9.637 days) for source for *all* developers and devices for selected developers (2007-02-11); a month, a week, 2 days (37.637 days) for anybody (targeting developers) (preorders: not yet) (2007-03-11); 7 months, a week, 2 days (221.637 days) for mass market (2007-09-11): see topic for more info Feb 02 14:10:27 There is currently very little info out there. Feb 02 14:10:40 Other than the spec sheet, and that derived from it. Feb 02 14:10:41 Tantalum: The wiki and more will open in about a week :) Feb 02 14:10:46 The official wiki is not open Feb 02 14:10:50 yet Feb 02 14:11:09 i wished release would be delayed until 19th Feb 02 14:11:13 thats when my last exam takes place Feb 02 14:11:16 grmbl Feb 02 14:11:18 Hehe Feb 02 14:11:49 ok... does openmoko work in the us? Feb 02 14:12:12 should Feb 02 14:12:16 with GSM provider anyway Feb 02 14:13:13 damn... Getting GSM here is a pain Feb 02 14:13:25 tmobile, cingular Feb 02 14:13:35 what's a pain is pay-as-you-go Feb 02 14:13:47 no prepaid data afaict Feb 02 14:14:54 yeah but where I live me and every one else uses verizon..... I never use my out of network minuets ;-) Feb 02 14:15:04 In the UK - it's not too bad - contract phones - many of - come with a meg of data a month, and 3 pounds ($6) / Mb Feb 02 14:15:32 3G is a whole nother ballpark of course. Feb 02 14:15:48 yeah I am sticking w verizon but getting prepay for devel Feb 02 14:15:52 But AIUI, all the providers require you to have a 3G capable phone. Feb 02 14:16:16 SE: Yeah, I'm jealous.. GPRS rates in Canada are a ripoff Feb 02 14:16:52 Tantalum: OpenMoko is OE based Linux distribution for mobile phones that initially runs on Neo1973 phone Feb 02 14:17:09 * Sketch has unlimited data (GPRS or 3G) for US$20/mo with cingular Feb 02 14:17:13 For 3G - IIRC you cang et a gigabyte a month for 40 quid. Feb 02 14:17:25 Fido dropped the unlimited GPRS because $50 was too cheap Feb 02 14:17:27 though i'm theoretically not supposed to tether to another device with that ;) Feb 02 14:17:32 Now it's $50 for 50mb and 0.03/kb after that Feb 02 14:17:44 Rogers has their $100 for 100mb Feb 02 14:18:14 ouch, 7.50 UKP for 2Gig here Feb 02 14:18:17 in my area coverage is good enough that I could switch from cdma to gsm Feb 02 14:18:17 * Sketch notes that that is $20 on top of the regular phone plan Feb 02 14:18:22 OpenMoko will be released in 10 days, you can order Neo1979 in 40 days Feb 02 14:18:27 but I am used to 3g :( Feb 02 14:18:32 I'm clinging to my hiptop solely for the data plan Feb 02 14:18:39 It's very closed and limited though Feb 02 14:18:41 i think tmobile is the cheapest for data in the US (dunno if they support 3G though) Feb 02 14:19:40 I dont think they do Feb 02 14:20:17 Oh wtf.. fido dropped all of their data plans except the 2mb for $5 plan Feb 02 14:21:13 Which is idential to the rogers plan Feb 02 14:21:29 Thanks every one for the info..... I'll probably be back when the phone is released and if I get one Feb 02 14:23:01 tmobile has edge at least Feb 02 14:24:17 t-mobile and cingular both have edge Feb 02 14:24:22 in many areas Feb 02 14:24:57 neo doesnt have edge Feb 02 14:25:29 2.5-not-edge Feb 02 14:27:19 cingular also has 3G Feb 02 14:27:39 (in some areas, of course) Feb 02 14:55:44 SpeedEvil: why would you need external GPS? Just route chip communication both to binary agpsd and your own agpsd and compre outputs Feb 02 14:56:29 aloril2 - All of that gets down to reverse engeneering. ;) Feb 02 14:59:01 Elrond: yup, but no need for external GPS for it Feb 02 15:00:50 Might not need to reverse engineer. We'll just have to wait and see whether Sean can open enough of it up. I think he *wants* to, from reading some of his visionary statements about openness. Perhaps he can reach agreement with Infineon/GL to release just the basics of the positional data formats. Feb 02 15:07:01 somehow i doubt it Feb 02 15:07:11 from what i understand, most of the work is done in software with GL chips Feb 02 15:07:24 and i dont think they will want to give their crown jewels away Feb 02 15:08:33 aloril: I was assuming that the context was for devs without physical devices. Feb 02 15:09:40 The crown jewels however isn't involved in the output of the hardware. Feb 02 15:09:57 That's probably pretty simple and relatively irrelevant. Feb 02 15:10:21 SpeedEvil: agree Feb 02 15:11:43 It's the converting, filtering, ... Feb 02 15:12:42 I bet that's true for the interfaces to nVidia's and ATI's graphics cards too, and that opening those interfaces would have no commercial impact whatsoever. They just want the control. In this GPS case it's even more-so. Feb 02 15:13:56 I think the thinking is that in the graphics case, the worry is that someone will take the detailed hardware specs and clone. Feb 02 15:14:10 Probably also there are horrible IP risks. Feb 02 15:18:03 having worked in a gfx chip manufacturer the number 1 reason for no open specs is they expose bugs in the chips, your competitor then does "demos" that expose those bugs Feb 02 15:18:03 anyone who has capability of actually cloning the hardware will probably have no problem doing so without the hardware interface specs Feb 02 15:18:23 but work perfectly on their chips Feb 02 15:18:51 PHB's who are involved in buying depts dont know technicall details, they can just tell perfectly rendered boobs from ones with flaws Feb 02 15:25:04 SpeedEvil: agree about crown jewel, but if you know chip protocol then you can feed it to existing GPL implementation Feb 02 15:25:24 Yeah - then tweak that. Feb 02 15:25:36 * SpeedEvil sees if he's had a response on the ITAR issue. Feb 02 15:25:55 Nope. Feb 02 15:32:58 Well if the competition can find your bugs for you, that's a bonus :-) Feb 02 15:34:01 hi Feb 02 15:34:13 Not if they are hardware, and would cost you tens of millions to fix. Feb 02 15:34:30 And they can be exposed by competitors in shipped product. Feb 02 15:34:47 Nah, it all goes through your driver ... easy to fix. Feb 02 15:35:08 SpeedEvil: bundled agpsd can have those ITAR limitations regardless whether proprietary binary is used or one compiled from GPL sources. Neither ase changes availability of GPL sources ;-) Feb 02 15:35:19 And so it should be. Covering up bugs is a mark of incompetence, not engineering. Feb 02 15:35:43 ITAR limitations? Feb 02 15:35:50 Basically: Feb 02 15:36:03 GPS that can report speeds over 1000 knots and 60000 feet are a munition. Feb 02 15:36:10 As defined in the ITAR regulations. Feb 02 15:36:21 SpeedEvil: can the Hammerhead? Feb 02 15:36:24 You need a special licences to import or export them. Feb 02 15:36:32 ah Feb 02 15:36:36 The hammerhead basically can - however, it's half of the story. Feb 02 15:36:47 The hammerhead outputs uncooked data. Feb 02 15:36:56 The gpsd cooks it and produces a position. Feb 02 15:37:24 It's pretty clear that if you sell the device with an unlimited gpsd that it is in fact subject to the ITAR regs - which means basically it can't be sold. Feb 02 15:38:00 Well if the Hammerhead is a munition and is being distributed to hundreds of thousands of consumers worldwide, then Infineon/GL are idiots. Feb 02 15:38:00 It's as I understand it not been tested in law if an easy to alter source gpsd that has the limits imposed is allowed. Feb 02 15:38:06 No. Feb 02 15:38:32 It's only a munition when it's combined into hardware + software that reports positions over the limits. Feb 02 15:38:57 And it's the manufacturer of the units problems, not infinieons. Feb 02 15:39:16 Well in that case the Hammerhead isn't a munion when it's standalone. Simply logic. So, let's have the documentation. ;-) Feb 02 15:39:57 No, it's not. However, it's possible a court could rule - pushed by apple - that an open source gpsd makes the neo a munition. Feb 02 15:39:58 s/munion/munition/ Feb 02 15:39:58 Morgret meant: Well in that case the Hammerhead isn't a munition when it's standalone. Simply logic. So, let's have the documentation. ;-) Feb 02 15:40:44 SpeedEvil: I think you aren't using enough layers of speculation. How about another 10? That would do it .... ;-) Feb 02 15:41:04 The ITAR regs are pretty clear. They are still enforced. Feb 02 15:41:05 If ... if ... if ... if ... Feb 02 15:41:29 It is totally clear that if you sell the neo with an unlimited GPS that it is a munition. Yes, this is insane. Feb 02 15:41:47 And by the ITAR, you just said the Hammerhead alone isn't a munition. So they don't apply. Everything else is speculation about the future. Feb 02 15:41:49 SpeedEvil: well, the chipmaker will also have an interest in not having all of their sold gps chips become munition Feb 02 15:42:05 The hammerhead is part of a hardware+software package that makes up the GPS. Feb 02 15:42:30 It is not a munition unless both parts of that package report positions/speeds outside the limits. Feb 02 15:42:48 The hammerhead can - but when tied to software that imposes the limits, it's fine. Feb 02 15:43:14 The hammerhead doesn't output a position, it outputs relative speeds of satellites in view. Feb 02 15:43:31 The position is generated by the gpsd Feb 02 15:44:15 sounds like you would be fine if you distribute it with a gpsd that obeys teh limits? if someone replaces the software to get around the limits, that's their own problem? ;) Feb 02 15:44:26 Probably. Feb 02 15:44:40 if it's already being distributed by other vendors with different software, that software could surely be replaced by someone who knows enough... Feb 02 15:45:10 Yes - however essentially all GPSs with flashable firmware are in that boat too. Feb 02 15:45:28 well, just because something is binary-only doesn't mean it's secret, so that problem exists already Feb 02 15:45:38 it's already replacable Feb 02 15:45:43 SpeedEvil: I don't know why you're trying to put a damper on GPS in OpenMoko, but it's not helping. If you think there is an ITAR issue, please take it up with Sean, and let's see if he says "You are forbidden from using the Hammerhead directly." Without that, it's just negative speculation. Feb 02 15:45:47 for anyone smart enough to reverse engineer the gpsd Feb 02 15:45:47 I think it's pretty safe to say you won't get a court to rule that a closed binary+hardware violates ITAR Feb 02 15:46:47 I am not. I just don't want any trivial way of taking the thing off the shelves, when it goes volume. Feb 02 15:47:41 this looks like the same old free wifi drivers vs fcc regs discussion Feb 02 15:47:43 SpeedEvil: Yes you are. You are saying that we cannot freely program those things FIC presents us in the hardware, and you use some FUD about possible ITAR problems to stop us. It's very bad. Feb 02 15:48:52 SpeedEvil: yeah, but there are many "closed+binary system" devices which have later had linux installed on them Feb 02 15:50:26 I want neo to succeed - I want one - well - maybe of v2, I may not have cash for v1. I'm waiting on responses from some people who have worked in this area. Hopefully they may have more insight. Feb 02 15:50:34 so when someone figures out how to install linux on some random GPS (like the cheap ARM based one I got that runs WinCE, i bet would not be too difficult) that has been on the market a few years, does that make the manuf suddenly liable? Feb 02 15:51:03 I don't know - it depends on what you can get a court to decide. Feb 02 15:51:33 I would say almost certainly not, as it's closed, and the person doing it is not doing it with the makers approval. Feb 02 15:51:41 SpeedEvil: Good, that sounds progressive. FUD about possible ITAR problems in the future is in contrast not progressive, it's as negative as all FUD. Feb 02 15:52:29 that sounds about as much as fud as the playstation could be used in missiles stupidity Feb 02 15:52:49 people who build missiles can get proper gps for it, rest assured Feb 02 15:52:56 Go and look at the ITAR regs - they do in fact apply. Feb 02 15:53:00 Even if the GPS is civilian. Feb 02 15:53:13 If it can produce positions outside the limits. Feb 02 15:53:35 Anyway - I'll stop talking of this aspect till I get more info. Feb 02 15:53:46 then put that limits in an open source gpsd as well and it's legal Feb 02 15:54:24 IMO, that'd be a matter for the courts, as it's essentially a documented 'delimit' switch. Feb 02 15:55:05 Anyway - on another topic - has it been cleared up if the 'hole' in some of the pictures is a lanyard attach, or a button? Feb 02 15:55:15 its a hole Feb 02 15:55:35 :) Feb 02 15:55:45 SpeedEvil: I fail to see difference between bundled binary A and binary B both obeying limits Feb 02 15:56:02 Bundled binary B comes with a documented off-switch. Feb 02 15:56:07 (the source) Feb 02 15:56:27 I wish LCDs were not so damn fragile. Feb 02 15:56:52 Worn round the neck, with no cover, I'll break it. :( Feb 02 16:00:43 it's rather unlikely that this kind of logic will hold in court, most of us here can probably reverse engineer the communication protocol in the matter of days without any source whatsoever so the "off switch" in the binary blob case is only marginally harder to find Feb 02 16:03:11 SpeedEvil: *both* come with off switch: download from net binary C without limits Feb 02 16:06:44 courts and logic arent exactly close friends though Feb 02 16:11:09 I hope I don't get chewed out for this question... I'm not as tech savy as many of you are, but I was wondering I develop in flash and would there be a way to get flashlite on openmoko? Feb 02 16:12:30 I'm not worried about courts here, but about the project being FUD'd out of sight. FIC is a large company, they have lots of lawyers. Let them deal with possible ITAR issues please. We're given an open piece of hardware at the interfaces, and our job here is to improve it with open source work in all areas. Feb 02 16:14:57 ans: anything that you can get to work with OpenEmbedded should work on the Neo. Feb 02 16:32:15 and why worry about flash when you can use Moble SVG Feb 02 16:33:09 Mobile SVG Feb 02 16:33:28 SVG is a open standard Feb 02 16:35:09 http://mobilemonday-ny.com/ Feb 02 16:38:06 so is swf Feb 02 16:38:35 but so far the open source implementations are lagging... Feb 02 16:46:28 the SVG guys are moving big time into mobile Feb 02 16:47:33 and they are working on Joost ....and if you've seen Joost you know it wants to be a Mobile app Feb 02 16:53:07 hmm since when is swf open ? last time I checked it was something like "you can get the specs if you agree not to use them to write flash player." Feb 02 17:04:06 AFAIK that is not the case. I doubt there would be a GNU flash project if that were the case. Feb 02 17:04:38 the gnu project is based on reverse engineering, imho Feb 02 17:04:40 not on open specs Feb 02 17:05:10 Can I use the File Format Specification to create a SWF interpreter or player? Feb 02 17:05:10 No, the File Format Specification is provided for the specific purpose of enabling software applications to export to the Macromedia Flash File Format (SWF). Feb 02 17:10:16 Nearly as closed as sdal. ;o) Feb 02 17:30:55 Adobe recently announced that they were going to turn over the reigns of PDF to open standards committee, although people say that's because of pressure from MS's document formats. Well Adobe now own Macromedia, so perhaps flash will one day be properly open. But yeah, currently it's not. Feb 02 17:33:24 12/window new Feb 02 17:33:26 sigh Feb 02 17:34:29 * cjb recommends a better IRC client. Feb 02 17:36:14 Typos still appear. Feb 02 17:36:20 Unless you go to point+click Feb 02 17:36:21 Talking of documents ... I wonder what the best ebook/document reader will be for us, given the size of the screen. Feb 02 17:36:42 I've used xpdf - it's probably not best though. Feb 02 17:37:15 plus enscript Feb 02 17:38:07 Do we have an accurate number for the screen dimensions? Feb 02 17:38:09 cu Feb 02 17:38:57 I wonder if that info is on the xoo template Feb 02 17:39:04 I suppose it's reasonable to assume that the 2.8" is 4:3 Feb 02 17:39:11 It looks about tht. Feb 02 17:42:07 * SpeedEvil goes away to read a book on a 2.8" window. Feb 02 17:53:55 counter Feb 02 17:53:55 a week, 2 days 11:34:16 (9.482 days) for source for *all* developers and devices for selected developers (2007-02-11); a month, a week, 2 days (37.482 days) for anybody (targeting developers) (preorders: not yet) (2007-03-11); 7 months, a week, 2 days (221.482 days) for mass market (2007-09-11): see topic for more info Feb 02 17:58:26 good morning all Feb 02 18:02:28 right, finally sorted out the hw ecc Feb 02 18:02:36 http://www.fluff.org/ben/linux-26/2620/2620-rc6-s3c24xx-nand-hwecc.patch Feb 02 19:00:31 Cripes, Dell and Intel are in one hell of a heap of costly trouble. (Slashdot story from Reuters) Feb 02 19:06:29 Where do we sign up to get a quarter of a billion to put only intel CPUs into the neo? :) Feb 02 19:07:20 SpeedEvil: i think that kind of incentive went away when intel sold its arm dev to marvell Feb 02 20:08:06 http://lists.cs.uiuc.edu/pipermail/llvmdev/2007-January/007813.html Feb 02 20:08:28 * koen calls for volunteers to do some benchmarks Feb 02 20:30:59 koen: J2ME job posting :-) http://tinyurl.com/ypuwwo Feb 02 20:47:00 koen: ah ha! saw my article didja? Feb 02 20:52:52 cying: Did a quick search to see if I could find the article you mentioned, and this is what I got: http://en.wikipedia.org/wiki/Charles_Ying Feb 02 20:53:16 CM: hehe, that's not me :) Feb 02 20:53:34 CM: http://www.satine.org/archives/2007/02/01/connect-the-dots-iphone-graphics-os-x-llvm-arm-and-ruby/ Feb 02 20:53:34 cying: I kind of guessed that, but it was still funny.. :P Feb 02 20:53:49 Ah, thanks Feb 02 20:53:51 CM: yea... you would not believe how much crap i get for that Feb 02 20:54:19 CM: "OMG charles i heard you were dating Janice, is that true?" Feb 02 20:54:27 Hehe Feb 02 20:57:48 * cying misses his top google ranking for his name. Feb 02 21:06:11 * SpeedEvil is playing second fiddle to a polar bear researcher. Feb 02 21:06:31 I once got invited on an expedition (paid) to the pole, to consult on polar bear mating habits. Feb 02 21:06:40 0.o Feb 02 21:07:27 I even look sorta-similar. Feb 02 21:07:42 I suspected they'd have worked out it wasn't me though. Feb 02 21:08:04 hehe Feb 02 21:14:52 Better get all the research done before they go extinct :-( Feb 02 21:17:00 There is no-one else on the web with the same name as me Feb 02 21:17:06 I'm every single web hit baby Feb 02 21:17:38 * CM joins the ego-surfing Feb 02 21:18:05 cat is sitting on my shoulder like a pirate parrot Feb 02 21:18:13 yarrrr Feb 02 21:18:18 Yaaaarr. Feb 02 21:18:22 parley! Feb 02 21:18:24 parley! Feb 02 21:18:46 You need to train it to wave a little sabre. Feb 02 21:19:17 Hehe, nice image Feb 02 21:20:49 her body is curled around the back of my neck Feb 02 21:20:50 too Feb 02 21:21:02 so, neck warmer Feb 02 21:21:08 Pr0n on OpenMoko! Feb 02 21:21:33 you can be sure that's comming Feb 02 21:21:37 nothing like tamale's made with cat Feb 02 21:21:39 Hehe Feb 02 21:21:43 I do after all live in the worlds fastest growing porn city Feb 02 21:21:59 hmm, two mokos + bluetooth app that syncs them = stereo pron ;) Feb 02 21:22:00 what city is that? Feb 02 21:22:04 noidd: I seem to be the only one with my name on google too :) Feb 02 21:22:05 Las Vegas? Feb 02 21:22:07 rochester? Feb 02 21:22:07 Charlotte, NC Feb 02 21:22:15 oh right Feb 02 21:22:15 the buckle of the bible belt Feb 02 21:22:29 I already suggested a porn image (perhaps background), changing according to your location Feb 02 21:22:58 did they give you a free neo to implement that? Feb 02 21:23:05 no, those bastards Feb 02 21:23:06 * cying didn't get a free neo. Feb 02 21:23:08 * SpeedEvil ponders the possibilities of 'interest finder' + bluetooth + gps. Feb 02 21:23:14 on the neo. Feb 02 21:23:19 * cying *may* develop his graphics framework on a greenphone instead. Feb 02 21:23:21 SpeedEvil, indeed. Feb 02 21:23:28 ping if someone matching your stated interests is in range. Feb 02 21:23:48 did you get a free greenphone? Feb 02 21:24:09 not yet Feb 02 21:24:12 first app, i'll write is a rss feed grabber to read the daily news in the subway Feb 02 21:24:14 i've been promised one Feb 02 21:24:28 :) Feb 02 21:24:30 what framework is this? Feb 02 21:24:39 * SpeedEvil ponder when the first neo will hit ebay. Feb 02 21:24:47 me? Feb 02 21:24:49 a day after the p0 launch Feb 02 21:24:50 yah Feb 02 21:24:55 someone is going to sell it, Feb 02 21:25:09 50 people, there has to be at least one person who fancies their chances Feb 02 21:25:15 Stephmw: a non sucky next gen UI framework for mobile apps Feb 02 21:25:15 i've just searched for cheap grps rates in germany. there is none :/ Feb 02 21:25:23 cying: unnamed as yet? Feb 02 21:25:26 Stephmw: unnamed Feb 02 21:25:30 all around 0.2 eur/mb Feb 02 21:25:38 err, 0.1 Feb 02 21:25:39 Stephmw: prototype first, then name :) Feb 02 21:25:58 though if the openmoko folks come up with a VMware simulator image, that might be just as good Feb 02 21:26:07 with all the tools set up and everything Feb 02 21:26:07 * mjr wouldn't buy a p0 second-hand phone... Those people should be hacking them, not selling them. Feb 02 21:26:21 * koen still wonders what people want simulated Feb 02 21:26:43 koen: did you check out that link? Feb 02 21:26:46 Greenphone may end up being just as open, depends. If it turns out that documented access to the GPS serial port is denied us, then I guess Neo is no more open than the Greenphone. We'll just have to see. Feb 02 21:26:47 toolchain setups, linkers, projects Feb 02 21:26:56 so i can just boot, load and start hacking Feb 02 21:27:12 koen: the setup is the big bugger... it's all about the usability to get the developers in Feb 02 21:27:13 Morgret, "no more"? I beg to differ, even in that case. Feb 02 21:27:28 (as for access to the serial port, it would really boggle me if there wasn't) Feb 02 21:27:29 Stephmw: heh, another nokia job offer Feb 02 21:27:42 cying: ehm, a toolchain is not a simulator Feb 02 21:28:10 * SpeedEvil would buy a p0 phone - if it was cheap, and do some dev work on it. Feb 02 21:28:11 koen: well when i run the built executable, i'd want it to come up on screen Feb 02 21:28:17 cying: it's saddening people keep giving the same answers as you did when I ask what they want to have simulated Feb 02 21:28:40 koen: oh, well then if you want simulator features: Feb 02 21:28:43 mjr: it boggles me too, since the whole point of Neo is openness. But we'll just have to wait a few days and see, unless Sean speaks up before then. Feb 02 21:29:18 koen: hey, it's the glamorous world of API development :) Feb 02 21:29:21 koen: emulated screen, touch input, and either build cross binaries or emulate the processor speed Feb 02 21:29:44 koen: so i can do "hello world" w/o much setup Feb 02 21:29:51 cying: the only actually useful thing for a simulator would be a gsm module Feb 02 21:30:24 telling yourself you need a simulator for 'hello world' is just sad Feb 02 21:30:45 koen: why? is screen access standardized across all linux distributions? Feb 02 21:30:52 koen: what do i do to blit to the screen? Feb 02 21:30:57 koen: and what do i do to get mouse input? Feb 02 21:31:05 cying, Neo will run X Feb 02 21:31:33 and looking at the currently available emulators for arm makes wanting one foolish as well Feb 02 21:31:37 mjr: do i learn X to get at the screen buffer? Feb 02 21:31:41 and X is pretty much standardized across most linux distributions; some other embedded ones don't use it, but anyway Feb 02 21:32:06 cying, not really, you should learn SDL instead for things like that Feb 02 21:32:18 SDL is a layer over X. Feb 02 21:32:19 mjr: and does SDL work with openmoko hardware? Feb 02 21:32:24 ahhh Feb 02 21:32:26 that is a bit simpler to use. Feb 02 21:32:38 sdl is a layer over pretty much anything Feb 02 21:32:38 so if i built a SDL codebase, would that just port over to openmoko? Feb 02 21:32:43 It brings up a SDL window, and you write to that. Feb 02 21:32:46 The only thing that bothers me is that qemu arm doesn't work well with gcc-4.x.x Feb 02 21:32:48 well, a layer over X and a whole lot of other stuff; yeah, what koen said Feb 02 21:33:01 qemu in general that is.. Feb 02 21:33:31 You probably can kill Neo's X and just use sdl on the framebuffer. Not that I'm suggesting it. Feb 02 21:33:35 mjr: so really, this would be the recommended way to write a openmoko app? Feb 02 21:33:40 cying: what I'm trying to say is this: "In contrary to other shitty environments, most openmoko stuff doesn't need to be emulated for you to test" Feb 02 21:33:54 cying, it would be if the app in question wants to access a framebufferish thingy Feb 02 21:34:10 if you want widgets and stuff, you use the openmoko libs, based on gtk Feb 02 21:34:49 koen: what i'm trying to say is that "packaging matters". from my POV, having extensive background in RTOS phone development, but 5 year old UNIX programming background, the less i need to learn to start writing openmoko targetted apps the better Feb 02 21:35:09 cying: I agree on that Feb 02 21:35:09 mjr: hmmm Feb 02 21:36:02 koen: from my POV, a VMware image that has linux, a simulator, and the openmoko package all set up in directories and such, with one script you run to get something built, is very very nice Feb 02 21:36:52 mjr: that's good info to know... maybe all i need to get started is just an SDL library Feb 02 21:37:44 cying: do you *need* a simulator, or do you _think_ you need a simulator? Feb 02 21:38:50 since openmoko is a huge break for conventional platforms, conventional paradigms will most likely be invalid Feb 02 21:39:12 koen: i need something that does this: when i launch ./neosim, i get a window that shows me the openmoko UI and my mouse will do touchscreen-ish things on that UI, built from the same code base Feb 02 21:39:21 I guess the "OpenMoko Simulator Howto" page in the official wiki will be a favourite.. Feb 02 21:40:24 CM: indeed Feb 02 21:40:34 koen: but i am curious, why i don't need this Feb 02 21:40:51 koen: it's just that from my POV, this is what i would want to get started Feb 02 21:40:57 you can get the same using Stefan's Xoo skin Feb 02 21:41:07 no need for a simulator Feb 02 21:41:09 since Feb 02 21:41:22 a) nothing available can emulate the cpu properly Feb 02 21:41:31 I want to know at the moment - does it simply look like a 'normal' X window with normal mouse? Feb 02 21:41:34 b) you don't have a 300dpi screen Feb 02 21:41:41 c) a mouse isn't a touchscreen Feb 02 21:42:10 what does his xoo skin do? Feb 02 21:42:11 b and c are annoying. Feb 02 21:42:26 a) can sort of be worked round by limiting cpu timeslice. Feb 02 21:42:27 cying: It just makes it look a bit like the open moko Feb 02 21:42:34 You can at least get pessimistic. Feb 02 21:42:37 CM: "it" ? Feb 02 21:42:37 SpeedEvil: nope Feb 02 21:42:37 And sets the resolution for Xoo correctly Feb 02 21:42:45 It's a png and an xml Feb 02 21:42:49 oh XOO Feb 02 21:42:58 SpeedEvil: qemu and armulator don't emulate unaligned access faults for example Feb 02 21:42:58 * cying googles Xoo Feb 02 21:43:18 cying: It's on the o-hand page Feb 02 21:43:19 oh see? Feb 02 21:43:21 SpeedEvil: so a+b+c make 'simulation' nothing more but a toy Feb 02 21:43:21 xoo is basically a pretty Xnest Feb 02 21:43:21 I mean _roughly_ - as in 1/bazth of a Pfoo/barGhz Feb 02 21:43:22 this is what i want Feb 02 21:43:25 afaik Feb 02 21:43:42 Horribly inaccurate, should be within a factor of 5 though. Feb 02 21:43:46 koen: i think that can be set Feb 02 21:43:53 koen: on armulator at least Feb 02 21:44:01 For much stuff. Feb 02 21:44:11 SpeedEvil: so people that say they *need* a simulator are fooling themselves Feb 02 21:44:29 SpeedEvil: or are in for a big surprise when they finally break out the real hardware Feb 02 21:44:46 "hey, this screen is smaller than on my CRT" Feb 02 21:44:49 Hehe Feb 02 21:45:14 "performance profiles are completely different, why's that?" Feb 02 21:45:14 koen: if you're doing serious app development, having a desktop development environment helps a lot Feb 02 21:45:18 Appropriate prescription glasses, and sitting back can ssort of help. Feb 02 21:45:20 But it's true, I guess the scrollbars and such will look huge on a normal monitor Feb 02 21:45:21 koen: even if the profiles are different Feb 02 21:45:33 cying: I didn't say it didn't help Feb 02 21:46:38 depending on a simulator is like not taking driving lessons put playing GTA:SA instead Feb 02 21:46:43 koen: so create a VMware linux distrib bundle with Xoo and openmoko's base system packaged and ready to go Feb 02 21:46:57 cying: right, that's already available :) Feb 02 21:47:08 koen: well shit, point me at it! Feb 02 21:48:04 did tha person ever pay up the cash for it koen? Feb 02 21:48:05 cying: MACHINE=qemuarm in OE generates an environment like that Feb 02 21:48:23 noidd: I have no idea what [g2] and hrw agreed on Feb 02 21:48:36 koen: can never please everyone. Half will say this phone is smaller than the xoo image, and half will say this phone is too big. :-) Feb 02 21:49:00 <[g2]> hey koen Feb 02 21:49:19 good grief Feb 02 21:49:26 I couldn't care less what its size is. It's open. Feb 02 21:49:30 there's about a 20 step process on the openembedded site Feb 02 21:50:14 g2: did you cough up the cash to hrw in the end? :-) Feb 02 21:50:17 and where is openmoko's core GUI then? Feb 02 21:50:23 is that just OE? Feb 02 21:51:05 <[g2]> noidd I had talked about a bounty and then I was going to write something up Feb 02 21:51:05 oh Feb 02 21:51:09 there it is on the home page Feb 02 21:51:10 okay Feb 02 21:51:20 well a VMware image would help, but i suppose this will do for now Feb 02 21:51:36 <[g2]> noidd I never wrote anything up, but hrw went ahead and developed something Feb 02 21:52:04 <[g2]> noidd I haven't tested what hrw created Feb 02 21:52:59 roger that Feb 02 21:53:05 get testing :-) Feb 02 21:53:08 <[g2]> noidd have you tried it ? Feb 02 21:53:53 No, didn't know it was there until a few lines ago Feb 02 21:54:16 <[g2]> noidd I've been distracted by two issues: 1) ixp4xx stuff, 2) moving my base build environ from edgy to xfce Feb 02 21:55:01 <[g2]> and 3) a bunch of video/media streaming things Feb 02 21:55:40 * [g2] has been wondering about the 2/11 OE/Openmoko release Feb 02 21:57:55 I'm guessing we'll get another mtn feed Feb 02 22:00:19 <[g2]> noidd dunno, but it was a 4hour+ monotone conversion for that last one I pulled Feb 02 22:11:53 wow Feb 02 22:16:20 hmm, what's this terminal laying here in the corner for? someone must have lost it. Feb 02 22:56:37 Nice joke, actually: http://degredo.net/ Feb 02 23:13:17 Greetings. Feb 02 23:14:36 howdy Feb 02 23:14:50 is there anyway to play with some emulator of the openmoko Feb 02 23:15:05 to give the feel for the kind of technology it is Feb 02 23:15:20 * pr3d4t0r came to learn about the tech from the /topic. Feb 02 23:15:54 counter Feb 02 23:16:16 hmh, the bot is down Feb 02 23:16:27 anyway, cprf, in about a week the platform source will be released Feb 02 23:16:46 that'd be all the "emulator" you'd likely need, except for perhaps some details Feb 02 23:16:47 * newmedian still votes for preorders for 3-11 Feb 02 23:17:42 sweet Feb 02 23:21:51 cprf: I read the wiki regarding hardware, etc. Is there a formal process to get a hold of the hardware? Feb 02 23:22:21 cprf: I've led development for commercial mobile applications. I'm curious about the actual hardware, even if it is in prototype form. Feb 02 23:22:40 s/cprf/mjr/ Feb 02 23:22:44 mjr: ^^^ Feb 02 23:24:01 mjr: The wiki only says "select developers" - I'm working with a group of people who have interested in the OpenMoko platform and we'd like to evaluate it not only the software/OS but also the hardware. We're based in San Francisco. Feb 02 23:24:11 I'm not with FIC; anyway, phase 0 where free phones are distributed is decided, phase 1 (wolrdwide sales targeting developers) starts 11th March Feb 02 23:24:28 (Phones given to select developers on Feb 11, available for sale to developers on Mar 11, and general public in Sept, from my memory of the counter) Feb 02 23:24:33 anyone can buy one for $350 Feb 02 23:24:34 mjr: Yes. Feb 02 23:24:52 The price 6 months on drom then has not been disclosed AIUI, and will presumably depend on the market then. Feb 02 23:24:54 mjr: It's not clear if there is a qualification process for 11.March. Feb 02 23:25:05 and when I say decided I mean they've decided already who gets the phase 0 phones Feb 02 23:25:07 probably a valid credit card. ;) Feb 02 23:25:29 pr3d4t0r, going by what they've said on the mailing list, no Feb 02 23:25:29 mjr: Phase 1 would be fine with us if we can put acquiring a couple of devices in our dev plan. Feb 02 23:26:08 anyone should be able to buy on March, just that they're expected to know they're getting a development device Feb 02 23:26:28 mjr: Cool, thanks. Feb 02 23:26:40 pr3d4t0r! Feb 02 23:26:44 slava! Feb 02 23:26:56 slava: What are you doing here? Feb 02 23:27:04 pr3d4t0r: i'm interested in the device Feb 02 23:27:07 slava: This is the last channel I'd expect to find you at. Feb 02 23:27:12 slava: So is my posse. Feb 02 23:28:06 pr3d4t0r: i'm porting my stuff to ARM/Linux Feb 02 23:28:14 or rather I already have Feb 02 23:28:16 slava: Factor? Feb 02 23:28:20 yes Feb 02 23:28:38 slava: Sweet. Feb 02 23:28:42 pr3d4t0r: i have one of these: http://gumstix.com/store/catalog/product_info.php?products_id=167 Feb 02 23:29:18 slava: Cool project btw, I'm impressed :) Feb 02 23:29:24 thanks Feb 02 23:29:48 slava: Nice. I met one of the ARM designers on an airplane a couple of months ago and stayed in touch with him. Feb 02 23:30:06 slava: In case we need one of those in-a-hurry responses not in the manuals yet :) Feb 02 23:30:22 pr3d4t0r: the manuals are pretty good, as far as CPU instruction set references go. Feb 02 23:30:34 slava: Yeah. ARM is a pretty nice architecture. Feb 02 23:30:40 pr3d4t0r: i'm looking for a usb<->serial adapter for my mac; why suggestions for a reputable online store that sells this type of thing? Feb 02 23:31:01 seems hard to come by. nobody cares about rs232 anymore Feb 02 23:31:13 slava: Check if Central Computer has them. If they do check the price. If you don't like the price shoot me an email and I'll get the stuff for you with my discount and ship it to you. Feb 02 23:31:29 slava: http://www.centralcomputer.com - try these guys. Feb 02 23:31:34 google found 'em :) Feb 02 23:31:37 slava: ;) Feb 02 23:31:55 slava: Their official list on-line and what I get may differ. But they'll ship anywhere in the world and have superb customer service. Feb 02 23:32:28 slava: Don't order from the web site; I'll give you the phone number for my sales lady here in SF and call her directly. Feb 02 23:33:07 pr3d4t0r: its only 20 bucks Feb 02 23:33:12 slava: Ah. Feb 02 23:33:26 won't break the bank if there's no discount :) Feb 02 23:33:35 ebay has them for around half that. Feb 02 23:33:36 slava: My discount kicks in at $100 anyway :) Feb 02 23:33:48 (incl delivery) Feb 02 23:37:31 slava: For other gear these guys rock. If you need anything more expensive let me know. Jonathan and I (and almost every geek I know) buy our kit here. Feb 02 23:37:43 sure Feb 02 23:37:46 slava: Feb 02 23:37:52 pr3d4t0r: Feb 02 23:37:53 slava: They are very Linux/*BSD friendly. Feb 02 23:38:16 slava: And if you need Apple-compatible kit they'll get it for you as long as you know what it is you want. Feb 02 23:38:42 pr3d4t0r: i'm going to upgrade the RAM in my G5 at some point in the future so i'll take you up on your offer then. Feb 02 23:41:13 slava: OKi. Feb 02 23:41:25 slava: They got me a good deal on Kensington memory for my MacBook :) Feb 02 23:41:56 slava: And if that future is more or less around early March, I may be in Canada and ship the memory to you from there -- no duties. Feb 02 23:42:06 slava: That's when I go skiing to Whistler for a week. Feb 02 23:42:07 whereabouts in canada? Feb 02 23:42:09 ah Feb 02 23:42:13 slava: Vancouver. Feb 02 23:42:22 fun Feb 02 23:42:26 slava: Yeah :) Feb 02 23:42:31 slava: I only ski once/year. Feb 02 23:42:36 slava: But when I do, I go all out :) Feb 02 23:43:03 pr3d4t0r: i gotta go out to dinner. talk to you later Feb 02 23:43:09 say hi to your g1rlz for me. Feb 02 23:43:31 slava: Heh. Bon appetit. Feb 02 23:54:14 Thanks everyone - have a nice wknd. Feb 03 02:40:23 anyone know if the cell tower ID will be accessible in the phone? Feb 03 02:41:20 Probably not. Feb 03 02:41:38 We only get AT commands to the phones modem. Feb 03 02:42:10 I've not seen a GSM modem that supports that. Feb 03 02:42:16 I may of course be completely wrong. Feb 03 02:43:34 The phone part is a closed-source TI chip. Feb 03 02:50:48 :=( Feb 03 02:54:51 Well - a truly open phone with the GSM bit open too would actually be arrestable for being in possession of in the UK. Feb 03 02:55:04 And selling them has severe penalties. Feb 03 02:55:42 Admittedly, it's not happened before, but people have been imprisoned for supplying phone hacking tools. Feb 03 02:56:39 you'd think the misuse would be the part that's illegal Feb 03 02:59:36 In the UK, it's an arrestable offense to refuse to be searched. **** ENDING LOGGING AT Sat Feb 03 02:59:59 2007