**** BEGIN LOGGING AT Sat Jan 11 02:59:58 2014 Jan 11 03:00:44 but what does that mean if I'm building with SDK 17+, but with a minimum requirement of SDK lower than 17? Jan 11 03:12:51 it means you can use new APIs where appropriate but have to be mindful of supporting older versions Jan 11 03:24:42 Anyone have a suggestion for getting around InetAddress's caching behaviors? Jan 11 03:24:55 do to what, exactly? Jan 11 03:25:42 JakeWharton: I can't have it caching previous lookups. And it does. I realize in >4.0 it respects TTL, but there's still a lot of oru client base who is <4.0. Jan 11 03:41:16 hello Jan 11 03:41:34 JakeWharton retrofit is better with gson or jackson for performance? Jan 11 03:43:51 probably the same anywhere where you would compare gson to jackson Jan 11 03:45:06 Baluse: Retrofit streams the response to the converter so it doesn't really matter Jan 11 03:45:28 the delay of the bytes transferring over the network will be 10x slower than Gson or Jackson can turn them into JSON and write them to the object Jan 11 03:47:19 to anyone who has implemented a custom account. What is the point of having preferences directly in the accounts settings page if they are written to /data/data/com.android.settings/ so no application can even access them? Jan 11 03:47:43 not sure how deep of information you want, but Jackson is usually faster on raw performance but at the expense of hammering the string pool and less aware of security whereas Gson is 10-20% slower (remember how I said this didn't really matter?) but using normal string parsing and has defenses against certain attacks against Java/JSON Jan 11 03:51:56 awesome work on retrofit JakeWharton. Im writing basically an open source version of Parse & retrofit is an integral part of it. Jan 11 03:54:26 awesome Jan 11 03:54:38 Retrofit is one of my favs Jan 11 03:55:58 trying to install android sdk in linux. after being prompted for my default workspace it crashes. i'm getting this in the terminal http://dpaste.com/1546085/ Jan 11 03:57:25 u looked at the log? Jan 11 03:57:37 probably a jvm issue if i had to guess Jan 11 03:57:59 tho it says in native...so who knows Jan 11 03:58:29 no i havn't. is openjdk compatible with android sdk? Jan 11 03:59:17 http://askubuntu.com/questions/180555/eclipse-does-not-want-to-use-openjdk-7 Jan 11 03:59:22 seems to be related Jan 11 04:00:46 jug6ernaut: thanks. i'll check it out Jan 11 04:01:06 i don't need to have eclipse already installed right? Jan 11 04:01:26 4 what Jan 11 04:01:41 to launch eclipse from the android sdk bundle Jan 11 04:01:41 the sdk is just a zip Jan 11 04:01:50 no its included Jan 11 04:01:56 right right. Jan 11 04:02:06 tho, i believe its old. Jan 11 04:02:16 late response. Jan 11 04:02:29 probably better off downloading the latest eclipse and using that Jan 11 04:02:34 It seems json to object mapping is a little bottleneck Jan 11 04:02:48 to me but only at my old phone LG P500 Jan 11 04:03:05 Baluse how much of a bottleneck? Jan 11 04:03:10 cause at galaxy tab is like 1sec whereas lg is 4-5 secs Jan 11 04:03:14 how big the objects Jan 11 04:03:21 :| Jan 11 04:03:32 7 lists of objects 1000~ objects Jan 11 04:04:18 ah, might want to throw the 7 lists into different requests & thread it Jan 11 04:04:31 that helped galaxy tab Jan 11 04:04:44 a lot becuase server has 200ms latency Jan 11 04:04:54 but at LG P500 made almost no difference Jan 11 04:05:27 how old of a device is that? Jan 11 04:05:36 gingerbread Jan 11 04:05:38 i mean the gtab is pretty old itself Jan 11 04:06:10 2010 lg p500 Jan 11 04:06:11 ic Jan 11 04:06:19 galaxy tab isnt very old Jan 11 04:06:26 its pretty old :P Jan 11 04:06:39 how often are u gonna be pulling that much info? Jan 11 04:06:47 i'll try installing using the standard eclipse package. that's what the post's suggesting. Jan 11 04:06:48 april 2013 is old ? Jan 11 04:07:12 not as LG p500 Jan 11 04:07:20 the Galaxy Tab 1 is older then 2013 Jan 11 04:07:26 its tab 3 Jan 11 04:07:33 ah Jan 11 04:07:39 and its fast i dont have a real performance issue Jan 11 04:07:44 this is going to happen once as sync Jan 11 04:07:55 but for slower devices Jan 11 04:08:03 meh Jan 11 04:08:09 they should upgrade :P Jan 11 04:08:09 i am competing IOS developer here heh Jan 11 04:08:23 who is synchronising faster Jan 11 04:09:17 competing? Jan 11 04:09:22 networking seems bigger bottleneck than writing these data to internal storage Jan 11 04:09:53 u would expect it to be lol Jan 11 04:10:37 i have the following layout xml defined, but i can't seem to get it to fill the screen height. can anyone help please? Jan 11 04:10:46 heh Jan 11 04:11:06 hope you wont paste the whole xml here :p Jan 11 04:11:12 :| Jan 11 04:11:18 jug6ernaut, god day Jan 11 04:11:29 it's here http://pastebin.com/2YGKLnhV Jan 11 04:11:30 good even Jan 11 04:11:31 yo Jan 11 04:11:44 http://i.imgur.com/sQYBGVb.jpg Jan 11 04:12:20 arcode fill_parent is deprecated, and is this being applied as the root element? Jan 11 04:13:06 arcode why on earth do u have linear_layouts wrapping single views? Jan 11 04:13:29 because i want those things on separate rows Jan 11 04:13:33 jug6ernaut, i hope to god you never look at any of my UIs Jan 11 04:13:42 infact, let me go delete some stuff from github Jan 11 04:13:56 lol Jan 11 04:14:39 it's the only layout being applied in the acitvity so it should be the root element Jan 11 04:15:56 when you say its not full screen, what is it doing Jan 11 04:16:06 & what are you wanting Jan 11 04:17:37 i want each row to be equally spaced vertically so it fills the height of the screen. i'll make a screen of what it looks like now Jan 11 04:20:50 do u want the views to grow in size? Jan 11 04:20:56 or just be equally spaced? Jan 11 04:21:14 equally spaced Jan 11 04:21:45 add android:layout_weight="1" to each of your "rows" Jan 11 04:22:21 also remove the linear_layouts wrapping single views, they arnt needed. Jan 11 04:23:15 ok i will try it thanks Jan 11 04:23:23 with this simple of a layout it doesn't matter, but with more complicated layouts nested layouts will degrade performance. So you want to remove them when able, especially if they arnt necessary :P. Jan 11 04:25:20 may actually need the nested linear_layouts... Jan 11 04:25:37 since layout_weight will force the height to not be wrap_content Jan 11 04:26:45 ok i will put them back in :) Jan 11 04:27:10 lol Jan 11 04:30:10 i changed the xml to this http://pastebin.com/NMvuwiad, and it still looks like this http://imgur.com/ZrX8sUt Jan 11 04:31:26 so adding android:layout_weight="1" didn't change anything Jan 11 04:32:53 arcode u got other issues, the layout is fine Jan 11 04:32:58 i just tried it and it looks fine Jan 11 04:33:29 i am loading the layout in a baseadapter and then in my activity i use setListAdapter(), could that be the problem? Jan 11 04:33:47 say what Jan 11 04:34:00 yes Jan 11 04:34:19 if its being loaded in a list view row Jan 11 04:34:46 why are u doing that? Jan 11 04:34:51 ah, so how should i load it in my main activity? Jan 11 04:35:06 i thought it was the correct way to load a linearlayout Jan 11 04:35:27 just setContentView(R.layout.layoutName); Jan 11 04:37:03 JakeWharton: Thanks for the response. I know that I have to use new APIs and be mindful of old devices, but the wording confuses me. If I build the app using the @JavascriptInterface annotation (from 17+), will older devices still be vulnerable? Jan 11 04:37:21 i dont' know Jan 11 04:37:38 ok, fair enough. Shouldn't be too hard to build a test app to find out Jan 11 04:37:49 arcode https://www.dropbox.com/s/39biws4h46me2nt/Screen%20Shot%202014-01-10%20at%2010.36.46%20PM.png is what its suppose to look like Jan 11 04:38:07 many thanks jug6ernaut it's working Jan 11 04:38:15 vn Jan 11 04:39:18 My gut tells me that only devices with 17+ available will be OK, but then again, it's something that happens with the development sdk, so maybe it happens at build time, and builds the class in a safe way... Jan 11 04:39:27 anyway, i'll try it out Jan 11 04:50:14 ah, on another link: "Note: If the linked SDK has been built for an API lower than 17, the vulnerability exists – even if the application using the SDK has been built for API 17 or above. T" Jan 11 04:50:19 well thats too bad Jan 11 04:50:50 what vulnerability Jan 11 04:51:19 running arbitrary code from javascript if you use addJavascriptInterface on a WebView Jan 11 04:51:29 ah webview Jan 11 04:51:34 yeah :/ Jan 11 04:51:58 just another reason making apps from webviews is a terrible idea Jan 11 04:52:13 quite a long list :P Jan 11 04:52:51 tell me about it...but unfortunately my company insists of making like 80% of the apps webview based Jan 11 04:52:58 because price? Jan 11 04:52:58 otiose, that vulnerability depends on a few things Jan 11 04:53:05 oh? Jan 11 04:53:25 easist way to fix it is build with a newer api Jan 11 04:53:42 and using ssl Jan 11 04:53:46 but doesnt the quote i just pasted say that that doesnt work? Jan 11 04:53:51 avoid http, and validate your certificates Jan 11 04:54:24 otiose, i believe if you target api 17 and up Jan 11 04:54:24 making apps out of webviews ftl Jan 11 04:54:33 pfn++ Jan 11 04:54:41 I installed Bower using NPM on Windows. How does Windows know how to handle the bower command? It isn't in the Path system variable. Jan 11 04:54:44 yeah i really dont think we'll be moving to ssl soon; our server architecture is so screwed up to begin with, it would take a long time to merge our normal+secure servers Jan 11 04:54:47 changing target shouldn't make a difference Jan 11 04:55:00 otiose, then they shouldnt be using webviews at all imo Jan 11 04:55:00 jcase even if you build against newer apis, on older devices the old version of the web view would still be used wouldnt it? Jan 11 04:55:55 jug6ernaut, i blieve they are mitigating it with the newer support library Jan 11 04:55:56 jcase: yeah, i agree with you. i fought the idea of going to webview, but thats all in the past now Jan 11 04:55:58 but i could be wrong Jan 11 04:56:11 ohh that is possible Jan 11 04:56:21 otiose, demonstrate exploiting it to your supervisor Jan 11 04:56:23 and document it Jan 11 04:56:34 else it may be your ass that gets blamed at some point Jan 11 04:56:40 yeah, we're talking about it on tuesday Jan 11 04:57:00 given teh age of webkit on most devices Jan 11 04:57:03 its not the only worry Jan 11 04:57:11 im the iOS guy... i showed him that I can just display the ouput of 'ls /sdcard' onto a browser Jan 11 04:57:24 and he's just like 'meh, i dont think its a problem' Jan 11 04:57:31 xD Jan 11 04:57:40 and im like wtf dude, this guy in the video _literally_ got root access to the phone with this Jan 11 04:58:47 true, if we only display our own pages, with no links to anywhere else, and we use ssl (which we don't) then we might be sort of OK Jan 11 04:59:06 otiose, why does your webview app need permission to external st orage? Jan 11 04:59:20 but still, man... knowing about something like this and purposely not do anything is dumb Jan 11 04:59:39 jcase: iirc, your app doesnt need those permissions for the exploit to get to it Jan 11 04:59:52 using addJavascriptInterface? yes it does Jan 11 05:00:04 but our app has a bunch of native features, one being taking/saving pictures Jan 11 05:00:06 unless it is using another vulnerabilityu after the fact Jan 11 05:00:26 yeah even if its an exploit of the view its still sandboxed. Unless it does actually get root. Which would be crazy... Jan 11 05:00:28 are your webviews using javascript? Jan 11 05:00:38 yep Jan 11 05:00:52 give me a few Jan 11 05:00:57 lets see if i can fix Jan 11 05:01:00 jug6ernaut: let me see if i can find the video where the guy got root; im not sure if i can find it again Jan 11 05:02:24 jug6ernaut, he used a secondary exploit Jan 11 05:02:45 :| Jan 11 05:02:57 * jug6ernaut avoids WebViews Jan 11 05:03:13 * jug6ernaut things the only valid use for a WebView is authentication Jan 11 05:04:09 ah ok, using just this exploit, they only have the sandboxed info Jan 11 05:04:10 otiose, i saw a workaround for at at somepoint Jan 11 05:04:36 but apparently you can drop in some other root exploits Jan 11 05:04:44 correct, they are using reflection to download additional files Jan 11 05:04:50 then execution the downloaded payload Jan 11 05:05:06 http://50.56.33.56/blog/?p=314 states gingerbreak, rageagainstthecage, zergRush, psneuter, etc Jan 11 05:05:24 :\ Jan 11 05:05:26 zergRush, ratc andpsneuter wouldnt work Jan 11 05:05:39 not by default Jan 11 05:05:45 zergrush might work with modification Jan 11 05:05:53 but i believe the other two are property space exploits Jan 11 05:06:04 and are unsetting ro.secure Jan 11 05:06:05 well, i dont know anythign about them... just what the article says... im an iOS guy who just does android on the side :D Jan 11 05:06:10 only granting root over the adbd Jan 11 05:06:12 I was just thinking about using UserDefinedFileAttributeView to identify changes in a file tree. Does anybody know if that works with android? Jan 11 05:06:38 otiose, the author apparently didnt either Jan 11 05:06:41 gingerbreak would work for sure Jan 11 05:06:45 we could change the we we're using it to use shouldLoadUrl or whatever its called Jan 11 05:06:50 on pre 2.3.4 Jan 11 05:07:08 dhulke look up FileObserver Jan 11 05:07:19 but with our current codebase, it might take a while to implement... but not impossible Jan 11 05:07:27 if i have time this weekend i might mess around with it Jan 11 05:07:39 * jug6ernaut wishes he worked on android apps at work Jan 11 05:09:10 ahow the hell do i set the alignment of a section WITHOUT fucking with the ndk-build linker scripts Jan 11 05:09:27 this is a FUCKING nightmare Jan 11 05:09:39 jug6ernaut: oh ok. I'm trying to create a folder sync app to sync with a computer. I tried to use WatchService but just figured out that i cant rename a folder if it has a subfolder being watched... Jan 11 05:10:29 makes sense Jan 11 05:10:37 either way, i have to do some more cleaning -- gotta leave to pick someone up from the airport in about 1.5 hours, and its still a bit of a mess ;D Jan 11 05:10:56 thanks for you knowledge, jcase, jug6ernaut Jan 11 05:11:00 np Jan 11 05:11:30 jug6ernaut, is probably a better person to listen to for the programming advice, mine is fairly limited in scope (you just happened to touch my area) Jan 11 05:11:43 lol Jan 11 05:12:10 jcase: you deal with webviews, security, or exploits? Jan 11 05:12:25 lol not webviews Jan 11 05:12:28 never used one before Jan 11 05:12:46 :') you are a fortunate individual, keep it that way Jan 11 05:13:24 yeah, well im pushing my luck, ive gotta go get an ipad and iphone for work project Jan 11 05:13:44 objective...c.... Jan 11 05:14:04 jug6ernaut, lucky me no writing for hte device Jan 11 05:14:10 just dealing with it over usb Jan 11 05:14:16 agh nice Jan 11 05:14:18 ahh* Jan 11 05:14:23 jug6ernaut: is an awesome language that is a pleasure to work with Jan 11 05:14:29 is what you mean to say, right? ;) Jan 11 05:14:36 xD Jan 11 05:14:57 idky they felt the need to fix c's syntax Jan 11 05:15:07 & in such a backwards way Jan 11 05:15:34 c has no objects, and they liked smalltalk, i guess Jan 11 05:15:56 if they didn't fix c's syntax, you'd be writing objc_sendMsg(obj, message, params ... ) every time Jan 11 05:16:29 & I've only done a very little iOS dev, but that shit is :S. At least with what i did to much of the code was generated, and i had no idea how views were being connected to there logic. Which i couldn't stand. Jan 11 05:16:48 otiose opposed to what in obj-c? Jan 11 05:16:55 (comparable operation) Jan 11 05:17:57 well instead of obj_sendMsg(array, setObject:atIndex:, object, index); (not sure about how parameters are send in atm) Jan 11 05:18:06 you would just do [array setObject:object atIndex:index]; Jan 11 05:18:15 super readable Jan 11 05:18:32 java might be something like array.setObject(object, index) or something, idk Jan 11 05:19:20 i just hated it when got nested function calls Jan 11 05:19:49 well, for a lot of things they are starting to embrace dot syntax as well Jan 11 05:20:24 anyways i out Jan 11 05:20:26 peace Jan 11 05:20:49 so you used to have to do do [array setObject:[self myObject]] but now you can do [array setObject:self.myObject] Jan 11 05:20:51 ok see ya Jan 11 05:21:25 * jug6ernaut much prefers the . syntax Jan 11 05:22:35 fucking hate the way every fucking person in the world developing "Development tools" fucks over people who try to develop in ASSEMBLER Jan 11 05:24:34 trying to call a DO NOTHING assembler function that I have wrapped up with a C wrapper. do a jni call on the c wrapper with the asm call commented out... works fine Jan 11 05:24:43 include the call to the DO NOTHING assembler funciton..... Jan 11 05:24:46 fucking segfault Jan 11 05:25:02 in fucking libc.o Jan 11 05:25:11 .so i mean Jan 11 05:26:25 how do i specify a custom linker script for a NDK library Jan 11 05:51:00 <_genuser_> guys, my user can click on listitems and it will load a new fragment. The fragment will launch asynctask to fetch json data. User can keep click on listitems. Jan 11 05:51:12 <_genuser_> this causes a huge delay and lot of asynctasks floating about. Jan 11 05:51:37 <_genuser_> eventually teh fragments take forever to actually respond. I think android prevents the new ones from running some of the old ones are done. Jan 11 05:51:48 <_genuser_> any ideas on how to tackle this situation best? Jan 11 05:51:59 don't run multiple ... Jan 11 05:52:08 <_genuser_> you see that too, eh? Jan 11 05:52:26 <_genuser_> I have moved it to MainActivity as static for now. but I wasn't sure if that's a good thing. Jan 11 05:53:21 <_genuser_> the asynctask also has a refernce to the fragment to notify it of completion. once the fragment is gone, it can't be notified and new ones can't start. Jan 11 05:54:02 <_genuser_> so I'm thinking of notifying the mainactivity of completion and let it store a reference to the latest fragment so it can notify it that it can go ahead with the request. Jan 11 06:02:34 if I were to get a contractless android phone for development purposes, today, which one should I get? Jan 11 06:02:55 anything from google Jan 11 06:03:02 nexus 5 then huh? Jan 11 06:05:41 <_genuser_> AsyncTaskLoader vs. AsyncTask, is either better to use? Jan 11 06:11:55 the task loader is around to generally fix issues of asynctasks. mainly involved in configuration changes Jan 11 06:13:08 there are a few limitations of asynctaskloaders, but i can't find a reference that lists them Jan 11 06:13:52 <_genuser_> Yeah from googling, I see both can fetch data in the background and update UI. Jan 11 06:14:16 <_genuser_> I think I'm doing to put a notify list in my AsyncTask. So it can notify all interested parties that it's done. The ones that are null have been unloaded. Jan 11 06:14:28 <_genuser_> the one that is not null is the one currently up and waiting for it to complete so it can request it's own data. Jan 11 06:14:32 <_genuser_> *its Jan 11 06:33:20 <_genuser_> if a fragment passes a refrences to itself to an AsyncTask, after a screen rotate would the AsyncTask have a valid reference? Or is the activity completely new? Jan 11 06:36:30 ok so i'm building cyanogenmod recovery(made a device tree for my galaxy trend) Jan 11 06:36:41 ionutica-99: wrong channel Jan 11 06:36:47 and all I get is a white screen flashing with random pixels on it Jan 11 06:37:16 bankai_: what channel should it be? Jan 11 06:37:27 #android-root or cm channel Jan 11 06:37:47 _genuser_: you're going to have a bad time on config change Jan 11 06:38:11 ok thanks Jan 11 06:38:13 <_genuser_> bankai_: I don't understand. Jan 11 06:38:34 <_genuser_> problem on config change due to which part of my design. Jan 11 06:40:48 your use of Android Jan 11 06:41:18 you want a loader because it will not retrigger the execution after config change Jan 11 06:41:31 <_genuser_> AsyncTaskLoader? Jan 11 06:41:47 <_genuser_> I'd be glad to change my code to it, if that's a benefit it offers. Jan 11 06:42:11 right. so on rotation (or any other config change) you will request the same loader be fired Jan 11 06:42:31 the loader manager will not re-execute the task because you haven't forced loading and because you already initialized it before Jan 11 06:42:46 <_genuser_> I see. that's perfect! Jan 11 06:42:48 it will either wait until the existing task is finished or it will redeliver the same result Jan 11 06:43:03 <_genuser_> provided, of course, that I used the same id, correct? Jan 11 06:43:22 <_genuser_> for init Jan 11 06:45:30 yep Jan 11 06:46:29 <_genuser_> alright, I have some major re-arranging to do, then. :) Jan 11 06:47:02 <_genuser_> I think this wouldn't have been problematic if my app wasn't launching a new fragment for each list item selection to request new data. Jan 11 06:47:35 <_genuser_> because the old fragment is destroyed, the new one would have to sit in a while loop waiting for a homegrown .isRunning() to finish. Jan 11 06:47:52 <_genuser_> and STILL have problems with rotate. Jan 11 06:47:53 <_genuser_> lol. Jan 11 06:48:18 <_genuser_> thanks again for your help. You have really helped me learn a lot of android stuff with just the right pointers. Jan 11 06:59:30 <_genuser_> lol, no need for isRunning(), and all that. Jan 11 07:00:59 is there any reason why a native library cannot call the mmap system call? Jan 11 07:01:08 can native libraries NOT make system calls? Jan 11 07:02:03 this fucking shit is fucking me over Jan 11 07:02:43 can anyone tell me if a native library can make system calls? Jan 11 07:09:17 You mean like system("fooCmd"); from c++? Jan 11 07:09:42 no Jan 11 07:09:45 i mean swi 0 Jan 11 07:09:52 a direct linux system call Jan 11 07:10:08 yes, of course they can Jan 11 07:10:15 maybe you are calling mmap with some silly arguments? Jan 11 07:10:53 no because the code works perfectly on my beagle board xm. and worked perfectly when i ran it as a native process under adb shell Jan 11 07:11:06 if i comment out the swi 0 the code does not segfault Jan 11 07:11:15 if i leave in the swi 0 it segfaults Jan 11 07:11:51 this was previously a stand alone native arm linux executable. i converted it to an android dynamic library Jan 11 07:12:25 in my JNI_OnLOad i allocate a 4k buffer - the code that does this here also does it in the native executable version. Jan 11 07:12:35 the code is not wrong. there is something else going on here Jan 11 07:13:09 why can i not make system calls Jan 11 07:14:23 mov r0, #0 ldr r1, #size mov r2, #3 mov r3, #0x22 mov r4, #-1 mov r5, #0 mov r7, #0xc0 swi 0 Jan 11 07:14:45 the #size here is a constant Jan 11 07:15:47 also. why is this freeking shit not letting me write into a section i declared with a "w" flag Jan 11 07:18:31 what is libdvm.so ? Jan 11 07:25:08 dalvik vm Jan 11 07:25:24 thats where the segfault is happening. Jan 11 07:29:19 this shit is now segfaulting inside my .so at a place that should not be executing Jan 11 07:29:41 my JNI_OnLOad tries to write into a WRITEABLE data section and i segfault Jan 11 07:29:55 but the reported address is utter bullshit Jan 11 07:31:15 maybe you should show us the code Jan 11 07:31:30 "Swi 0" Jan 11 07:31:32 thats the code Jan 11 07:31:45 *shrug* Jan 11 07:32:09 anything else other than that is not important, i KNOW it works. it works just fine on my hardware if i run the original executable in adb shell Jan 11 07:32:11 NO PROBLEM Jan 11 07:32:17 so i know the code is just fine Jan 11 07:32:23 evidently it's not Jan 11 07:32:29 whats NOT fine is for some reason my library is not allowed to swi 0 Jan 11 07:32:48 come on, you've got these things called logic and reason, stop raging and think Jan 11 07:32:49 maybe i need to list some permissions im not aware of Jan 11 07:33:20 also. i have a .section user_data, "wa" Jan 11 07:33:30 i try str r0, [r1] where r1 points into that section and Jan 11 07:33:32 segfault Jan 11 07:33:39 you said yourself that the address is, quoting, "utter bullshit" Jan 11 07:33:54 you also should know that assembly without context is useless Jan 11 07:34:05 so please show us the actual code or stop venting Jan 11 07:34:16 .section userData, "aw" Jan 11 07:34:17 user_data: Jan 11 07:34:17 .space 0x00100000 Jan 11 07:34:22 thats the section Jan 11 07:34:28 ldr r0, =user_data Jan 11 07:34:36 str r0, [r0] Jan 11 07:34:39 segfault Jan 11 07:35:03 yes i know allocating that empty space in the library is silly. but it proved the flags are working Jan 11 07:35:23 so the address space should be +w Jan 11 07:38:49 so first your problem was with calling mmap, then it was executing some code that "should not be executing", and now this Jan 11 07:38:58 what -is- it really? Jan 11 07:40:06 BOTH i tried to temporarilly get around the mmap issue by not allocating the buffer at run time but allocating it at assemble time Jan 11 07:40:19 one way gives me one problem, the other way gives me a different problem Jan 11 07:40:22 BOTH need to be fixed Jan 11 07:40:37 is this hand-written assembly or is there some source code it's compiled from? Jan 11 07:40:44 hand written assembly Jan 11 07:40:52 oh, then you probably have many problems Jan 11 07:41:10 you can take a look at the native linux executable at www.isforth.com/a4.tar.bz2 that contains the soruces Jan 11 07:41:14 WORKING linux sources Jan 11 07:41:23 but its not EXACTLY what im using here Jan 11 07:41:28 but its the basis for it. Jan 11 07:42:00 that a4.tar.bz2 runs on my beaglbe board xm running gentoo. it runs on my phone in adb shell Jan 11 07:42:11 well, you're trying to turn it into a shared library Jan 11 07:42:12 my phone is rooted, not sure if that is a requirement for it to work there Jan 11 07:42:16 yes Jan 11 07:42:20 that's a whole other bucket of headache Jan 11 07:42:27 yup Jan 11 07:42:48 and you're probably screwing up the ABI somewhere Jan 11 07:43:18 no i dont think thats the problem Jan 11 07:43:19 again, assembly without context is useless. Just because some fragments look fine doesn't mean -anything-. Jan 11 07:43:51 have you ever done any androd development in asm? and i dont mean "inline asm" Jan 11 07:44:00 i mean in asm. pure asm. no c. Jan 11 07:44:02 luckily, no Jan 11 07:57:12 unrecognized .section attribute: want a,e,w,x,M,S,G,T Jan 11 07:57:39 which is fucking annoying because im looking at the GNU docs for as and half of those fucking flags arent even listed Jan 11 07:58:28 maybe you need to use 'info' :| gnu man pages are often elided Jan 11 07:58:39 im looking at the gnu WEB page Jan 11 07:58:59 time to look at the source ;) as isn't too bad Jan 11 08:00:10 no way in HELL im looking at the source code to ANYTHING produced by the GNU foundation, they would just make me thoroughly CRINGE. Jan 11 08:00:20 cluster fuck source format Jan 11 08:00:22 no comments Jan 11 08:00:39 heavilly obfuscated macros using undocumented assembler features (im sure) Jan 11 08:00:48 the sources wouldnt help me Jan 11 08:00:52 lol Jan 11 08:02:58 <_genuser_> JakeWharton: wow, that was pretty cool. not unnecessary refresh on screen rotation. !!! Jan 11 08:03:12 yep. that's the idea Jan 11 08:04:26 :( Jan 11 08:04:28 <_genuser_> I have another strange behavior that I just can't describe. Jan 11 08:04:48 can someone PLEASE tell me why the FUCK i cannot write into a data section i have declared as +w in my sources? Jan 11 08:05:01 .section userData, "w" @progbits Jan 11 08:05:27 <_genuser_> in method, populateInfo(...), I check for a value, and if it's 0, I set the background color to red'ish and foreground color to brighter red. Jan 11 08:06:01 <_genuser_> well, sometimes when the said value is not 0, the background is changed, however, the foreground is unchagned. It is the original intended color. Jan 11 08:06:31 <_genuser_> sometime, when I just inflate the view, one of the fields, that is hardcoded to have a greenish background also get the redish color. Jan 11 08:06:41 <_genuser_> all these colors are defined in @color/xxx Jan 11 08:14:05 i cannot seem to be able to declare a section as writeable Jan 11 08:14:10 fuck Jan 11 08:21:48 _genuser_: sounds like incremental build problems Jan 11 08:22:03 the unique IDs in the R file are being shifted so you're getting adjacent values Jan 11 08:22:37 30/? I/DEBUG? signal 11 (SIGSEGV), code 2 (SEGV_ACCERR), fault addr 80500000 <--- TOTALLY not the fucking address where the segv happens Jan 11 08:22:44 <_genuser_> JakeWharton: oooh, should I go over it and make sure, there's no collisions? Jan 11 08:22:50 <_genuser_> or is this an eclipse problem. Jan 11 08:22:55 Eclipse problem Jan 11 08:23:03 <_genuser_> or should I just do a clean? Jan 11 08:23:05 yep Jan 11 08:23:28 <_genuser_> what's happening, is that on the same run, 5 times, I'll click the nav drawer list and ti will load the fragment just right. on the 6th, it will load it incorrectly. Jan 11 08:23:35 <_genuser_> and then it will stay botched until app close. Jan 11 08:24:04 <_genuser_> the 5th,6th times being examples. it can go on for more times working as expected. Jan 11 08:25:02 ive noticed that eclipse screws up with builds if you dont do make clean at least every 3 or 4 builds Jan 11 08:25:10 but that was with the TI version of it Jan 11 08:25:59 but ive never been able to run any version of eclipse for longer than a couple of hours before totally uninstalling it and primising to never install it ever again Jan 11 08:26:23 <_genuser_> I'll say this of eclipse.... Jan 11 08:26:28 i think the android NDK ignores section atribute flags Jan 11 08:26:40 specify writeable... not writeable Jan 11 08:26:42 <_genuser_> it has a delight launch screen. I have never seen any other app's launch since netscape, but.... Jan 11 08:26:44 segrault on write Jan 11 08:27:44 and pointless filing any BUG report with google. probably requires a G+ accoun t anyway. seems like google needs more data to sell to the spammers Jan 11 08:28:37 <_genuser_> eclipse's ctrl+I to format is the worse. Jan 11 08:29:46 just rm -rf eclipse and install intellij idea community edition Jan 11 08:29:59 <_genuser_> is that an ibm product? Jan 11 08:30:14 no Jan 11 08:30:38 <_genuser_> 'coz I was about to say they've got some odd products. not very impressive. Jan 11 08:30:41 its the greatest java development environment ever devised by god Jan 11 08:30:50 <_genuser_> I'm thinking of moving to android studio. once this app is done. Jan 11 08:30:59 android studio IS intellij idea Jan 11 08:31:11 <_genuser_> well, there that's nice. Jan 11 08:31:45 Does android studio come with maven? Jan 11 08:38:31 so a simple 10 minute job has thus far taken 3 days Jan 11 08:38:48 wish the fuck the GNU development suite would just disappear Jan 11 08:38:50 permanantly Jan 11 08:40:17 <_genuser_> if I may venture a guess, I'd say you're just a tad irritated... Jan 11 08:41:07 cant write into a writeable section. section declared as nobits takes up 1 megabytes worth of file size Jan 11 08:41:12 yes im pissed off Jan 11 08:41:52 <_genuser_> yeah, I was just being overly formal like Arthur Spooner from King of Queens. Jan 11 08:42:38 dont watch it, no idea what it is Jan 11 08:42:42 <_genuser_> but I'm fairly ticked at my app, also. Jan 11 08:42:45 dont have a tv, have not had one in close to 20 years Jan 11 08:42:50 no intention of ever getting one Jan 11 08:43:38 <_genuser_> don't know but when I hardcode a value #01ff01 into a field, it's background is being changed in an if-statement that does background and foreground both. Jan 11 08:43:58 <_genuser_> but only the background is changed, and the data clearly indicates that it should not be even in the if statement. Jan 11 08:44:17 <_genuser_> think I really need to read this function carefully. Jan 11 08:49:01 http://www.foxnews.com/tech/2014/01/10/google-integrates-google-with-gmail/?intcmp=features Jan 11 08:49:10 ya google can be trusted... honest! Jan 11 08:50:22 hello everybody and happy new year. I got an OEM tablet, a very cheap one, and yesterday the manufacturer gave me a rom upgrade in .img format. They told me i should enter update mode but no luck there. I managed to install cwm though to have a recovery environment. Can you tell me how to install the new rom? Thank you in advance Jan 11 08:51:13 if it was in zip format it would be easy, and i googled about converting the img into zip but it's a lot of trouble :( Jan 11 08:56:14 "Hello, I have a format suffix that represents a collection of files, and another format suffix that also represents a collection of files. How would I convert one to the other?" Jan 11 08:56:50 why on earth did this guy install CWM rather than learn to use the provided update facility correctly. Jan 11 09:02:46 <_genuser_> is it not standard jva behavior that if your method doesn't catch an exception it bubbles up to the caller? Jan 11 09:03:03 <_genuser_> and if the caller has a try catch block around the method and one of the catches is (Exception e), it should be caught? Jan 11 09:04:24 _genuser_: it bubbles up until someone catches it, or terminates the program Jan 11 09:05:11 <_genuser_> alankila: yep, the caller has a try{ func_call(); } catch (Exception e) { ; }. So it should be caught and not crash the android app. Jan 11 09:05:14 and formally, Exception is just one of the subclasses of Throwable. All your normal exceptions inherit from it. Jan 11 09:05:28 <_genuser_> exactly. So it's a catch -all. Jan 11 09:05:33 Errors are another type of throwable that should not be caught because they aren't generally recoverable Jan 11 09:05:57 <_genuser_> java.lang.IllegalStateException that's def an exception tho. Jan 11 09:06:19 in principle you can do just what you did there, the try-catch block that catches Exception is a common sight after all Jan 11 09:06:28 it isn't the best possible coding practice though Jan 11 09:06:47 <_genuser_> you mean react to each situation individually, etc? Jan 11 09:06:51 and you should not swallow exceptions silently. At the very leat, Log.error(TAG, "Something bad happened while doing x", e); Jan 11 09:06:54 or was it Log.e Jan 11 09:07:15 <_genuser_> right, I'm doing Log.i there. that was just to spare the channel long code lines. Jan 11 09:07:48 but in general yeah I do a lot of catch-Exception stuff, though I most often throw it right again but this time as RuntimeException Jan 11 09:07:49 <_genuser_> I was actually letting them bubble up to the call, so the caller can actually display it on the activity. Jan 11 09:08:16 this exception class does not need to be declared and handled by caller, so it streamlines programs and usually without any loss Jan 11 09:08:46 my own exceptions also often inherit from RuntimeException Jan 11 09:08:55 <_genuser_> I have come to loathe exceptions. Jan 11 09:09:31 <_genuser_> mainly because of devs who don't catch exceptions. especially if you throw them for their benefit. Jan 11 09:09:44 <_genuser_> and it crashes the whole thing and the boss is looking you because it crashed in your unit. Jan 11 09:11:04 well, that is a social problem. Shouldn't guide what the right design is. Jan 11 09:11:38 I prefer writing "bug-free" algorithms as much as possible, and one of their characteristics is that they can never throw and all input combinations are sensible and do something useful Jan 11 09:12:12 <_genuser_> alankila: yes and no. I mean if your code is breaking in production and you're the one staying late fixing it 'coz the caller won't catch it or has a clueless look then the boss only knows one thing Jan 11 09:12:15 still, i/o is one of those things that generally causes IOExceptions to be declared as thrown and they actually do get thrown. No way around that. Jan 11 09:12:16 <_genuser_> who knows how to fix it. Jan 11 09:12:22 I dislike it when APIs use unchecked exceptions Jan 11 09:12:40 i dislike it when development tools fuck me over Jan 11 09:12:42 <_genuser_> I have decided on a compromise. Jan 11 09:12:51 like not honoring WRITEABLE attributes on sections i create Jan 11 09:13:03 <_genuser_> I catch all exception by frameworks and by your code. and store them in a var for you to query. then I return true/false, or obj/null. Jan 11 09:13:21 ew Jan 11 09:13:29 <_genuser_> if you get a false or a null, you can care to check the last seen exception. if not, it will only crash if you use the null without any checking. Jan 11 09:13:38 O.o Jan 11 09:13:43 that is C-like error handling Jan 11 09:13:47 <_genuser_> what this causes is, application not reacting at all. Jan 11 09:13:49