**** BEGIN LOGGING AT Tue Aug 26 02:59:57 2008 Aug 26 03:14:07 that is a good post. props to morrildl Aug 26 03:14:36 but lets get that bluetooth API happening pls Aug 26 03:15:53 xmpp would be great Aug 26 03:28:36 it's just 1.0 Aug 26 03:28:41 these APIs will be added in a future release Aug 26 03:29:48 cant wait for October 13 Aug 26 03:38:52 romainguy__: you work in python? Aug 26 03:39:42 I used to Aug 26 03:39:42 is it a good idea for an android developer to learn python? Aug 26 03:39:54 it's a good idea to learn Python period Aug 26 03:39:58 it has nothing to do with Android Aug 26 03:40:03 but it's a good and useful language Aug 26 03:40:09 gr8 Aug 26 03:40:32 any python resources you can't live without? Aug 26 03:40:43 the official docs :) Aug 26 03:40:48 that's all I ever used Aug 26 03:40:49 haha Aug 26 03:40:54 :) Aug 26 03:40:57 :) Aug 26 03:41:06 that's all we need.. Aug 26 03:41:48 why goog chose python? Aug 26 03:41:55 instead of perl or php or ruby.. Aug 26 03:42:35 does it slow you down too much to add the le in there..... goog-le? Aug 26 03:42:40 Ruby wasn't popular/very documented Aug 26 03:42:52 and PHP is rarely used outside of web servers Aug 26 03:43:05 michaelnovakjr_: yeah it terribly slooooooows me down :) Aug 26 03:43:21 and Python is a nice language :)) Aug 26 03:44:52 python is better than ruby? Aug 26 03:45:01 to me they're the same Aug 26 03:45:06 I prefer Python's syntax though Aug 26 03:45:44 yeah, looks cool Aug 26 03:47:06 GAE is influencing me a lot Aug 26 03:47:36 Android + GAE = rocks! Aug 26 04:21:37 You all see the new post about GTalk Services and Bluetooth 2.0 API being left out? Aug 26 04:22:33 yeah, nice post Aug 26 04:25:54 trigatch4: yeah, i kind of figured as much Aug 26 04:26:23 it's funny to me that they needed their security team to realize that there were problems with allowing remote peers to arbitrarily invoke code on your device :) Aug 26 04:26:32 the network operators are scared of the internets Aug 26 04:26:58 haha yeah Aug 26 04:27:06 jasta: thats not actually what its doing Aug 26 04:27:13 but they mentioned that had they included the Bluetooth 2.0 API as is.. .they would be locked into that for years to come Aug 26 04:27:18 SanMehat: that's part of what your official explanation said. Aug 26 04:27:28 and.. so... regardless if its a black eye on the process or not, i'm glad they didn't include it Aug 26 04:27:32 SanMehat: however i was just being quippy. it's funny that you needed your security team to realize any part of why that wouldn't work :) Aug 26 04:27:57 thats why they're there.. to catch stuff that us super-humans might miss when we're coding 36 hours a day Aug 26 04:28:15 san: 36 hour days are not approved for external disclosure at this time. Aug 26 04:28:23 swetland: oops sorry Aug 26 04:28:37 nothing to see here folks, move along. Aug 26 04:29:00 not approved to disclose how you bend the laws of spacetime? :) Aug 26 04:29:15 or about how google treats you like slaves? :) Aug 26 04:29:19 swetlands my boss Aug 26 04:29:25 halp =D Aug 26 04:29:32 it just means the security model was scheduled for 'later'...quite understandable and sensible. Aug 26 04:30:04 gambler: that is neither. security must always be considered throughout the design, never as an afterthought. Aug 26 04:30:12 hahaha Aug 26 04:30:13 the latter is Microsoft circa 1995. Aug 26 04:30:44 security is subtle and having a specific team doing extra review is very worthwhile. I'm happy we've got 'em. Aug 26 04:31:14 yeah, I for one am really glad we've got them to quadrouple-check the code Aug 26 04:31:15 swetland: certainly, but the flaws in the gtalkservice as discussed in that post seem painfully obvious, not requiring some specialist to discover. Aug 26 04:31:24 * swetland shrugs Aug 26 04:31:38 jasta: you on a witch hunt or something? Aug 26 04:31:59 no, it just tickled me is all. do you honestly not agree that those were obvious points? Aug 26 04:32:14 honestly i've been so busy i hadn't thought of it Aug 26 04:32:18 san: we are not confirming or denying the existence of witches on the team or support or non-support for witches in the APIs Aug 26 04:32:26 because i've got a zillion other more important things to worry about. Aug 26 04:32:37 SanMehat: well, i didn't attack you so don't take it so personally. Aug 26 04:32:46 i'm not. Aug 26 04:32:57 i'm cool like school Aug 26 04:33:01 somebody somewhere wasn't thinking, and that made me chuckle. Aug 26 04:33:06 it happens Aug 26 04:33:15 cool like cuil Aug 26 04:33:25 groan Aug 26 04:33:35 haha Aug 26 04:33:39 gambler: whats your game of choice? Aug 26 04:33:57 jasta_: google treating them as slaves, come on! Aug 26 04:34:07 SanMehat: on a side note, i do think it is kind of interesting how a lot of the android engineers don't get the opportunity to see the platform like we do. Aug 26 04:35:04 SanMehat, ill bet on which raindrop falls down a window faster, but I spent a few years losing money playing internet poker and options trading Aug 26 04:35:25 gambler: options trading is too complicated for my simple mind Aug 26 04:35:36 so is poker, but i play it anyways Aug 26 04:35:50 SanMehat, stakes? Aug 26 04:35:59 i like 5c slot machines.. hehe Aug 26 04:37:18 they were showing lasvegas highrollers the other day, amazing how much these folks spend gambling Aug 26 04:41:27 SanMehat: *poke*, privmsg Aug 26 04:51:24 lame, my new aidl is 400 lines long. now i have to implement this beast ;) Aug 26 04:51:27 shit, tuesdays are so boring Aug 26 04:51:32 there are 4 types of listeners too, *yargh* :) Aug 26 04:52:15 400 lines.. how many methods? Aug 26 04:53:55 i don't know, a lot :) Aug 26 04:54:13 lots of comments though, too :P Aug 26 05:04:02 gambler: $1/2 for fun... $5/10 in vegas Aug 26 05:10:13 tuesdays are the best days Aug 26 05:10:31 for what? Aug 26 05:11:08 its not the beginning, not the middle and not the end.. hehe Aug 26 05:19:11 the best for business Aug 26 05:19:30 mondays people are getting re-acquainted from the weekend Aug 26 05:19:44 fridays are the best for business Aug 26 05:19:48 tuesday they are concentrated and refreshed and online Aug 26 05:19:56 depends what business Aug 26 05:20:09 not for online business Aug 26 05:20:15 right Aug 26 05:28:30 yakischloba: how was your first day? :) Aug 26 05:29:46 jasta: the guy that was supposed to train me went home sick before I got there. I spent about 3 hours being taught some basic stuff and the next 5 figuring some stuff out on my own, wandering around the data center looking at stuff, trying not to fall asleep Aug 26 05:30:33 seem like it's gonna be a good gig though? Aug 26 05:30:44 yeah it'll be exciting once I get rolling Aug 26 05:30:49 coo Aug 26 05:31:20 I can see by browsing through the ticket system that the level I was hired for will be a breeze Aug 26 05:31:22 gah i really wish i could figure out what was wrong with ServiceTestCase :( Aug 26 05:31:32 hopefully move up pretty fast Aug 26 05:31:50 yakischloba: excellent, just gotta show the right attitude and it should be easy Aug 26 05:32:01 mmhmm Aug 26 05:33:29 thrilling to have a zillion dollars of fiber networking stuff blowing hot air in my face Aug 26 05:33:38 lol Aug 26 06:09:59 jasta: i'm pretty sure you can find your answer to ServiceTestCase by asking on AndroidForums.com Aug 26 06:10:07 ;) i kid, i kid... Aug 26 06:31:20 . Aug 26 06:31:48 .. Aug 26 06:31:56 yo Aug 26 06:32:11 how goes it folks Aug 26 06:33:04 pretty well Aug 26 06:33:43 good good Aug 26 06:34:06 romainguy: do you think you could do me a favor? *grin* Aug 26 06:34:15 depends Aug 26 06:34:26 "READ THE ISSUE TRACKER!" :) Aug 26 06:34:50 i ran into a small problem using ServiceTestCase where it seems callbacks are not supported. I doubt this is really true, i'm just failing to set up some part of my test environment correctly Aug 26 06:34:56 f00f-: haven't you noticed the recent activity? :) Aug 26 06:35:10 :P Aug 26 06:35:12 Do you know of any tests in Android's framework that use this (callbacks)? Could you peek into one and figure out if there's some trick? Aug 26 06:35:31 no I don't know of any test that does that Aug 26 06:35:45 and I have absolutely no will to look into this right now Aug 26 06:36:54 i think monkey is a lot useful than instrumentation Aug 26 06:37:10 muthu: there are both useful in different ways Aug 26 06:37:27 yeah, but you get monkey for free ;) Aug 26 06:37:30 romainguy: ok, fair enough. Aug 26 06:37:47 but I don't see why callbacks would not be supported Aug 26 06:37:54 an instrumented activity is just an activity Aug 26 06:38:23 romainguy: me neither, but i can't see that i am doing anything "weird". except, i am blocking the test thread. but i don't think that should have anything to do with it. Aug 26 06:44:41 romainguy: i dunno, i posted to the groups so hopefully someone responds Aug 26 06:44:52 that's what they're for Aug 26 06:45:27 i half expect i'm going to get a "don't use undocumented systems" response, though Aug 26 06:45:50 its 2 different process Aug 26 06:45:56 so callbacks might not work Aug 26 06:46:18 muthu: what do you mean? Aug 26 06:46:28 what does it being 2 processes have to do with anything? Aug 26 06:46:59 may b threading issues? Aug 26 06:47:12 i doubt very much that it is simply broken Aug 26 06:48:00 if the test environment doesn't match the real environment sufficient enough to make recursive binder calls work, then it's not worth very much IMO. Aug 26 06:50:37 i dropped instrumentation coz at this stage its a waste of time Aug 26 06:51:20 i don't have a clue what you're talking about Aug 26 06:53:02 haha Aug 26 06:53:07 that doesn't surprise me ;) Aug 26 06:54:34 me either. i don't know why you would think instrumentation is a waste of time just before handset launch. Aug 26 06:54:43 QA is the most important thing you could be doing right now Aug 26 06:55:12 without any documentation? Aug 26 06:55:36 there's really not much to instrumentation Aug 26 06:56:11 romainguy: right, but it really doesn't work in my case. there must be some special trick that you have to do to get recursive binder transactions working Aug 26 06:56:31 i imagine most of Android is tested with this stuff, right? tons of code must be using it internally Aug 26 06:56:50 probably Aug 26 06:56:54 don't think so Aug 26 06:56:58 we definitely use it for the UI framework Aug 26 07:01:12 gm guys Aug 26 07:01:16 morning Aug 26 07:01:36 romainguy: i get the sense that i am doing a lot of esoteric things with Android in my designs for some reason :) Aug 26 07:01:38 So I'm leaving for breakfast now. :) Aug 26 07:16:35 anno^da: what's for breakfast Aug 26 07:16:39 ? Aug 26 07:26:09 in chennai, idly sambar is the standard breakfast Aug 26 07:35:50 it would sure be nice if you could get aidl to generate you a RemoteCallbackList stub for callback interfaces ;0 Aug 26 07:35:56 writing these sucks so much :) Aug 26 07:37:34 more pictures of htc dream leaked Aug 26 07:38:13 http://androidguys.com/wp-content/uploads/2008/08/g1_official_image.png Aug 26 07:38:17 for anyone interested Aug 26 07:39:52 so that little tilt at the bottom is real? i always wondered why that was there Aug 26 07:40:02 it's going to make it fit in people's pockets funny Aug 26 07:58:17 muthu: now I'm back Aug 26 07:58:56 I had bread with cheese and marmelade and some ham. Aug 26 07:59:07 and fresh apples from the garden Aug 26 07:59:26 and a big glass of apple juice Aug 26 08:15:32 ugly device Aug 26 08:15:43 i wont buy the first one then Aug 26 08:16:15 Zohan, :) Aug 26 08:19:14 i'm not too sure what i feel about this, actually Aug 26 08:19:34 this doesn't seem like it's gonna win any hardware design awards :( Aug 26 08:19:49 jasta, you talk about the dream? Aug 26 08:19:55 are there new pics out ? Aug 26 08:19:59 i guess i'll have to wait and see how it feels in the hand and all that, but it looks kind of like recycled HTC crap Aug 26 08:20:08 or are you just imagine how it will look like Aug 26 08:20:12 <_avatar> cutmasta: < trigatch4> http://androidguys.com/wp-content/uploads/2008/08/g1_official_image.png Aug 26 08:20:19 HTC makes some good phones, but they've also got plenty of really generic ones. this sorta seems like one of those more than their really fancy ones Aug 26 08:20:51 jasonchen, the diamond pro is nice deisgn Aug 26 08:20:54 jasta Aug 26 08:20:56 :) Aug 26 08:20:59 right, exactly Aug 26 08:21:03 fuck autocomplete Aug 26 08:21:05 sorry Aug 26 08:21:06 :) Aug 26 08:21:14 <_avatar> i really don't have a problem with the way it looks. that being said, it doesn't appear to be the most attractive device out there Aug 26 08:21:45 _avatar: http://www.htc.com/www/product.aspx?id=49518 -- the same company made this. Aug 26 08:22:06 <_avatar> personally, because I'm a cheapskate, I'd rather buy a functional, slightly ugly phone, than something twice as expensive that does the same thing but looks better. Aug 26 08:22:12 and that device seems much more attractive. perhaps the face plate will get some love, though Aug 26 08:22:18 <_avatar> so as long as the handset is decently priced i'm sold, i think Aug 26 08:22:18 _avatar, you are right Aug 26 08:22:31 but in my opinion they should sell a competitor to iphone Aug 26 08:22:38 regarding functionality AND design Aug 26 08:22:48 _avatar: yeah but a phone with wi-fi, gps, huge touchscreen, slide-out keyboard, etc is _NOT_ going to be cheap. Aug 26 08:23:03 gps not certain Aug 26 08:23:08 oh and throw in a 3D accelerator chip in there. Aug 26 08:23:10 <_avatar> jasta: true, but a sexier phone will be more expensive Aug 26 08:23:23 i'll go for a reasonably priced phone Aug 26 08:23:35 muthu: for real? are we also speculating as to whether it will have a cellular radio? Aug 26 08:24:16 why on earth would there be no GPS? last i checked google's "my location" was still terribly inaccurate. Aug 26 08:24:42 donno, might be some device constraints Aug 26 08:24:59 <_avatar> seems like it'd be silly not to have a GPS given the tight maps integration Aug 26 08:25:12 yeah, what a clusterfuck that would be to not include GPS Aug 26 08:25:14 stop speculating guys, just wait for the devices and you'll see what's inside Aug 26 08:25:22 <_avatar> :) Aug 26 08:25:36 (on the other hand it's fun watching you speculate :) Aug 26 08:25:40 lol Aug 26 08:25:43 <_avatar> i'm sure it is Aug 26 08:25:43 haha Aug 26 08:25:56 this channel would die, if we don't speculate ;) Aug 26 08:26:06 jasta would have nothing to do.. hehe Aug 26 08:26:20 muthu: oh he would Aug 26 08:26:28 I already know whats in the devices Aug 26 08:26:30 i could get back to coding, of course. Aug 26 08:26:31 the diamond pro is too thick imo Aug 26 08:26:36 the diamond touch is nice Aug 26 08:26:45 its an iphone running android Aug 26 08:26:46 obviously Aug 26 08:26:54 hehe. Aug 26 08:26:59 this would be really cool Aug 26 08:27:07 publish an installer for iphone Aug 26 08:27:19 that would kick ass ;) Aug 26 08:27:23 nice device Aug 26 08:27:24 that would really kick SJ ass :) Aug 26 08:27:30 call it "the iphone youve been waiting for" :) Aug 26 08:27:31 with extremly nice OS Aug 26 08:27:41 dont think I would want to run android on an iphone.. Aug 26 08:28:07 hm, in my opinion the iphone weighs to much Aug 26 08:28:28 design is nice Aug 26 08:28:33 big filespace Aug 26 08:28:40 but Aug 26 08:28:40 android on iphone would be a nice Aug 26 08:28:48 the battery runs out fast Aug 26 08:28:54 if you really USE it Aug 26 08:28:57 has anyone hacked linux onto an iphone currently? Aug 26 08:29:36 if they have, and the driver status is good, then it would probably be trivial to load Android on there. Aug 26 08:30:18 but i doubt very much that anyone has done that. Aug 26 08:30:22 and i bet the work involved would be staggering Aug 26 08:30:32 yeah, not even rumors around Aug 26 08:30:36 theyre running some version of bsd Aug 26 08:30:56 muthu: "rumors"? do you find a lot of open source hackers that don't publish their findings? Aug 26 08:31:27 Zohan: Loading a form of BSD onto the iPhone is a much less impressive feat, since the iPhone alraedy runs a form of BSD. Aug 26 08:31:55 thats what i meant, the osx base is some bsd version Aug 26 08:32:17 that doesn't translate at all to loading Linux onto the device with good driver support. Aug 26 08:32:22 btw, i dont publish my findings Aug 26 08:32:25 all the iphone drivers are assuredly closed source. Aug 26 08:33:21 the iphone is probably nightmarishly properietary too. probably no other phones with even similar hardware except maybe the baseboard/cpu. Aug 26 08:33:48 and with it also being the only hardware to run OS X, there are probably no tools to explore the hardware exisitng Aug 26 08:34:09 the task would be incredibly huge for anyone to take on. it's just not happening... Aug 26 08:34:18 and noone should support apple by buying an iphone and putting a good OS on it! Aug 26 08:35:24 especially given the outrageous iphone pricing Aug 26 08:35:45 the iPhone is just simply not an attractive target for tinkering Aug 26 08:36:02 the chiphone is Aug 26 08:36:05 you guys seem to think this sort of stuff is magic just because you're used to someone doing all the real work for you. Aug 26 08:36:24 hmm, any idea how to resolve the path/filename of a raw resource? Aug 26 08:38:59 embedded systems are embedded systems Aug 26 08:39:05 yo swetland Aug 26 08:39:09 they're not particularly magical Aug 26 08:39:29 though sometimes it's a pita to get actual databooks from the silicon vendors Aug 26 08:39:36 'evening Aug 26 08:39:54 swetland: you're good at yelling at them ;-) Aug 26 08:40:09 yelling at who? us? Aug 26 08:40:13 romainguy: I mean for joe random third party hacker Aug 26 08:40:20 jasta: no, sillicon vendors Aug 26 08:40:26 swetland: ah yes :) Aug 26 08:40:27 oh, of course. Aug 26 08:41:15 the hard part is all in the last 5% though. saving 0.1mA here and there, finding the flaky stability issues in the hardware or the modem or whatever Aug 26 08:41:38 swetland: and then see these @!# framework devs ruin all of your efforts with fancy features :)) Aug 26 08:41:49 but watch those guys who are getting android (and other mobile linux stuff) fired up on a bunch of existing handsets with no docs Aug 26 08:42:05 swetland: /me was contributing to that effort on the Vogue... Aug 26 08:42:08 determination and an insane amount of free time goes a long way ^^ Aug 26 08:42:18 and actually, you're missing something about those folks... Aug 26 08:42:28 they were all using HaRET, an existing tool which basically is doing all the work for them Aug 26 08:42:40 it's not quite all that simple Aug 26 08:42:55 then just trying to guess at the differences between the code *you*'ve been writing and what is appropriate for the other hardware :) Aug 26 08:42:56 though having existing tools to help reverse engineering is a plus Aug 26 08:43:14 where did you change voltages? Aug 26 08:43:20 jasta: and disassembling the wince stuff and picking it apart, etc, etc Aug 26 08:43:57 swetland: HaRET can actually do most of the work on certain devices. Aug 26 08:44:45 not sure what you mean by "the work". I am reminded of people requesting a device driver wizard for beos where you "just punch in the irqs and addresses and it writes the driver for you" ^^ Aug 26 08:45:15 it looks like haret is damn handy for sorting out gpio assignments and so on, which is certainly nice Aug 26 08:45:30 swetland: it was basically that easy. i know you have no free time to look, but look at the diff for the vogue against your tree ;) Aug 26 08:46:05 yes, 90% of the diffs are a bitmap for the onscreen keyboard Aug 26 08:46:19 well that's not what i meant :) Aug 26 08:46:29 i just meant that much of it is a copy/paste job on your code, modifying the small details that are necessary. Aug 26 08:46:32 but the work picking apart the quite different shared memory interface and getting the at/ppp channel up was unlikely to be magically done by haret Aug 26 08:46:35 most of those details were discovered from HaRET Aug 26 08:46:38 i talked to dzo quite a bit about this ;) Aug 26 08:47:02 oh yes, that piece was damn impressive. i have no clue how he made that work. Aug 26 08:47:11 I do. Aug 26 08:47:16 it remains impressive though ^^ Aug 26 08:47:23 im on jasta's side with this, without knowing both codes Aug 26 08:47:34 my "side"? Aug 26 08:48:07 the copy n paste wasnt much work side Aug 26 08:49:41 driver work is all about the details. and sorting out those details in absence of actual docs or reference code can be quite time consuming, even if you do have good tools to help. don't make the mistake of thinking just because there's not a lot of code, it's trivial Aug 26 08:50:03 swetland: i didn't make that mistake. i am telling you, i discussed this with him :) Aug 26 08:50:35 but regardless, it was still really incredible and that is with good tools. porting something like the iPhone is dizzying. you're starting totally from scratch there. Aug 26 08:50:47 by the time you succeeded to get a stable, working stack, the iphone would be old news :) Aug 26 08:50:56 yet people do stuff like that Aug 26 08:51:24 sure, but generally the more interesting hobbyist projects come from more attractive targets than the iphone Aug 26 08:51:44 or maybe not interesting, but usable :) Aug 26 08:52:06 it's an arm core with flash, ram, standard embedded peripherals, a radio module Aug 26 08:52:20 hi there Aug 26 08:52:29 not much different than many other devices people pick apart ^^ Aug 26 08:52:56 at least the iphone has no SD card slot, like the kaiser, where the hackers are struggling with driver issues Aug 26 08:53:15 I always was amazed at the desire to use SD Aug 26 08:53:23 the nand driver was one of the first things we published Aug 26 08:53:30 that sdcc is just plain evil Aug 26 08:54:03 jeez, i really have to go to bed Aug 26 08:54:09 this is ridiculous being up at 2am Aug 26 08:55:13 come on over here, 11 am here :) Aug 26 08:55:35 i am finally to a place where i have to implement that stupid local HTTP server work-around and i'm just dreading it Aug 26 08:55:38 so i'm not even coding anymore Aug 26 08:57:03 anyway, night folks Aug 26 08:57:48 gn8 Aug 26 09:00:57 swetland: so you're saying the nand driver would be usable for the kaiser as well, even right now? well, and if it is, actually making use of it wouldn't that imply flashing some experimental linux (w/wo android on top) to the rom? but what if it goes wrong? what a bout the bootloader? is there a way back? whereas "playing" with haret booting from SD seems reversible... Aug 26 09:02:35 not writing to the onboard flash is certainly lower risk Aug 26 09:02:54 but I have to say if we had to get sd working first it would have delayed us impressively ^^ Aug 26 09:03:38 swetland: I should use the same excuse for ListView bugs :p Aug 26 09:03:48 you blame San? Aug 26 09:03:59 no Aug 26 09:04:00 does he know? Aug 26 09:04:04 I blame myself Aug 26 09:04:17 (we nailed another one today) Aug 26 09:04:27 "I don't fear diseases, ListView will kill me first." -- Romain Aug 26 09:04:30 lol Aug 26 09:04:51 I feel the same way about the AMSS shared memory interface Aug 26 09:04:59 just the name makes me crap my pants Aug 26 09:05:06 at least we don't have a floppy controller Aug 26 09:05:09 thank the gods Aug 26 09:05:55 though the usb client controller was pretty hateful in its own special ways Aug 26 09:06:06 well, you used the word USB Aug 26 09:06:13 so I was kinda expecting hatred to be involved Aug 26 09:07:03 lol, :) just because a floppy drive would buckle the pocket more than the phone itself Aug 26 09:08:06 z: at Be, ownership of the floppy driver (which was incredibly hateful) was considered a dire punishment Aug 26 09:11:02 why on earth did i install an irc client on my phone Aug 26 09:11:08 :) Aug 26 09:14:13 because its great? Aug 26 09:14:14 :) Aug 26 09:16:53 no, not true :) Aug 26 09:20:33 :-) Aug 26 09:20:48 you have to go to sleep :) Aug 26 09:20:54 jasta Aug 26 09:42:09 romainguy: you there Aug 26 09:42:53 I know that you like your mac, have you tested that: http://www.foozoodesign.com/lockdown.html . Pretty funny (ok it is somehow senseless but just a fun app) Aug 26 11:57:27 nice, finally found a way to "deploy" my nativ lib with the .apk, and prolly gonna be killed for that by any android dev ;) Aug 26 12:03:58 tric: who? you can install native libs with .apk? Aug 26 12:07:02 well. "install", i put them in as raw resource, and on start before loading i check for filename, if its not found i use the resource inputstream and write them to the fs. but of course not in /system/lib, cause ro mount and root permissions, but /data/anr is world writable and System.load(fullpath) can load this lib Aug 26 12:07:44 works only with a single jni lib, so you have to link all your jni libs into one static. else the linker wont find the dependencies Aug 26 12:08:24 buster: but thats of course a hack. prolly wont work with next sdk, and for sure not with a real device (which isnt hacked) Aug 26 12:08:57 but for testing this is nice, finally no need to push before running the service after emu start. Aug 26 12:15:55 .o(i guess i should have told this, i bet next sdk release System.load is removed) Aug 26 12:18:13 in the next it is open ;) Aug 26 12:19:26 but not in 1.0 or on real devices. Aug 26 12:21:20 thats true :) Aug 26 12:22:03 fun fact: the android security model does not depend on the vm for security Aug 26 12:22:28 some parts do Aug 26 12:22:33 true, it depends alot on the kernel. Aug 26 12:22:47 zohan: which parts do? Aug 26 12:23:08 Oh after looking at the pictures of the G1 I hate the logo placement. I'd like to have a device without the logos. Even without the ugly T-Mobile one. Aug 26 12:23:10 no, the security design of the system is around one-uid-per-app and kernel enforced process boundaries Aug 26 12:23:22 you can of course crash your app very easily via jni, but only your app gets killed, i tried that alot ;) Aug 26 12:24:10 whats the permission good for, then? Aug 26 12:24:53 the system servers can map the uids of calling apps to their package info and use that to enforce whatever Aug 26 12:25:27 direct hardware / filesystem access is generally granted by setting additional groups when the app process is launched (for permissions that map to direct driver access, etc) Aug 26 12:25:33 zohan: ok, true. those INTERNET permissions are good for nothing. but you could add a JNI permission ;) Aug 26 12:25:54 tric: actually those internet permissions are pretty important Aug 26 12:26:30 swetland: well, but you can (currently) use jni to access the net i guess. Aug 26 12:26:36 well, a programm could use a internet and jni permission then. and then read the position via native access on the gsm/gps modul Aug 26 12:27:02 there are around 2000 native methods. i guess you can do nearly everything native Aug 26 12:27:06 tric: I think the emulator kernel may not currently be build with CONFIG_PARANOID_NETWORK Aug 26 12:27:18 http://git.android.com/?p=kernel.git;a=blob;f=net/ipv4/af_inet.c;h=97e4f081abdd045ea6b65cb1a4040741c3f34ff2;hb=a9b254e2f99fa52d2f2841a27d1eddef12bdac2c Aug 26 12:28:11 ah, interesting, so those permissions are handed over to the kernel aswell. Aug 26 12:28:26 so enable jni support! ;) Aug 26 12:28:49 the process start from zygote arranges to put the app in the appropriate additional groups. the kernel merely enforces access with them Aug 26 12:29:01 drop C code, execute it, exploit some app to gain root privs and install rootkit :) Aug 26 12:29:24 there is a reason very few things run as root Aug 26 12:29:26 zohan: some app, thats the problem, there are only very few ;) Aug 26 12:30:39 you can do it if you really want Aug 26 12:30:44 but you must try, try and try Aug 26 12:31:16 you dont have to try and try, you have to read alot when the source code is open ;) Aug 26 12:32:06 thats what some people get paid for Aug 26 12:32:18 I'm sure even without that people would happily pick through the binaries to look for fun stuff Aug 26 12:32:20 reading source code, finding exploitable areas Aug 26 12:33:17 in any case, the point is system security is not designed around the VM being the last line of defense (or even that it be reliable at all, though the vm guys work hard to make it so) Aug 26 12:34:11 native code support is more an issue of ABI/API support -- which was a bit out of scope for 1.0, but obviously something we're thinking about Aug 26 12:34:36 native code support is a must Aug 26 12:34:46 or else someone will jailbrake the gphone :p Aug 26 12:34:47 well, for my use, only deployment needs to be solved. some local lib path for every .apk. Aug 26 12:35:02 Zohan: you dont need to jailbreak an open platform Aug 26 12:35:32 I'm sure if it's not supported people will devote considerable energy to working around that. Aug 26 12:35:34 buster: the platform is open. maybe the phones are not! Aug 26 12:35:34 if youre bound to a VM Aug 26 12:36:08 swetland: yeah, i found a way in 0.9_r1, only a writable directory is needed. Aug 26 12:37:08 the main thing to keep in mind is that it is not supported, and there are no guarantees that even if you do whatever and run native code that that native code will work the same on different android devices, future versions of the platform, etc Aug 26 12:37:15 tric the way you "found" is new File(); ? :D Aug 26 12:38:02 people running native code tend to know this swetland. they arent stupid Aug 26 12:38:30 I don't recall saying they were ^^ Aug 26 12:39:30 zohan: with input to fileoutputstream, yes. and i never treid System.load before, always used System.loadLibrary Aug 26 12:40:31