**** BEGIN LOGGING AT Sat Nov 22 02:59:56 2008 Nov 22 03:05:47 I'm looking for 1) Something that does a simple POST request to dump data to a server 2) something that does a simple multipart HTTP upload request to dump something to a server Nov 22 03:05:56 opensource code snippet or something Nov 22 03:06:51 obviously 1) and 2) do not need to be the same code snippet or even the same program; i just need *some example* of the above Nov 22 03:09:21 check out anddev.org Nov 22 03:09:40 multipart upload tutorial was added today Nov 22 03:09:45 POST has been up there for a while Nov 22 03:10:10 umdk1d3: you here? Nov 22 03:19:43 ismarc: are you sure that multipart tutorial was added today? Nov 22 03:19:48 it looks like it was posted on oct 04 Nov 22 03:22:54 hello. Nov 22 03:23:25 hi Nov 22 03:23:39 how's it goin? Nov 22 03:24:11 ouch... froze Nov 22 03:24:51 theCarpenter: maybe it was just updated today, I just saw it on "view posts since last visit" Nov 22 03:27:38 ismarc: gotcha Nov 22 04:15:14 hi! Nov 22 04:15:21 hello Nov 22 04:15:24 I'm booting android for the first time Nov 22 04:15:43 but I'm stuck Nov 22 04:15:46 http://pastebin.com/m43527a63 Nov 22 04:15:47 any idea Nov 22 04:15:49 ? Nov 22 04:16:33 youre booting it on what? Nov 22 04:16:48 it looks to me like /system and /data are not at all right. or /, even. Nov 22 04:17:06 i'm booting on a omap2 board Nov 22 04:17:23 well, it definitely looks like you've not set up /data or /system as it expects. Nov 22 04:18:19 I just copy /data and /system to the initramdisk Nov 22 04:18:35 the initramdisk just have /init Nov 22 04:18:50 and empty /dev plus other things Nov 22 04:18:57 everything was in /system Nov 22 04:21:49 here is what I got in those dirs: http://pastebin.com/m6b254b3 Nov 22 04:22:21 \/data is empty Nov 22 04:22:34 data shouldnt be empty... Nov 22 04:22:47 (i dont think *grin*) Nov 22 04:23:27 hmm, I just compiled the android src code, was busy all this week with the kernel Nov 22 04:23:37 and now that boots something is missing :) Nov 22 04:23:48 mornin fellas Nov 22 04:23:59 mornin Nov 22 04:24:03 jasta: what's up? Nov 22 04:24:06 solca: hi Nov 22 04:24:56 i'm thinking about a twitter like app for g1 Nov 22 04:25:11 muthu_: or say just an interface to twitter? Nov 22 04:25:25 theCarpenter: no not an interface to twitter Nov 22 04:25:25 id be surprised if it hasn't been done already Nov 22 04:25:43 take the twitter idea and apply to something useful ;) Nov 22 04:25:49 http://stuffthathappens.com/blog/2008/11/01/browsing-android-source-in-eclipse/ -- hmm, this is interesting Nov 22 04:25:53 though apparently doesn't work. hmm. Nov 22 04:26:02 Hrm... if only twitter had a way to send arbitrary text messages to phones. Now that would be a great idea! Wait... Nov 22 04:26:22 hello Nov 22 04:26:24 twitter can be channelised imo Nov 22 04:26:43 twitter is a glorified message passing system Nov 22 04:26:57 right, can we make use of it for business? Nov 22 04:26:57 i'm not quite sure what you want here Nov 22 04:27:09 muthu_: can you make use of >message passing> for business... Nov 22 04:27:23 here's what i'm thinking: Nov 22 04:27:25 muthukrishnan Nov 22 04:27:32 theCarpenter: heh Nov 22 04:27:38 muthuxxxxx Nov 22 04:27:57 i know ten academics with that last name =] Nov 22 04:28:03 lol Nov 22 04:28:32 and another ten corporate programmers and sysadmins with the same Nov 22 04:28:32 so back to twitter.. Nov 22 04:28:35 anyway continue :) Nov 22 04:28:47 muthu is the most common name in tamilnadu Nov 22 04:29:11 >back to twitter Nov 22 04:29:30 why can't we use twitter for jobs? Nov 22 04:29:48 you know connect the employer <==> employee.. Nov 22 04:29:52 that's my line of thinking Nov 22 04:29:55 because you dont want confidential information sitting in a cloud service Nov 22 04:29:59 the idea itself is fine Nov 22 04:30:06 but you want some internal message passing system with a good web interface Nov 22 04:30:11 ok, make it secure/private.. yada yada Nov 22 04:30:17 indeed! Nov 22 04:30:24 how does im work in g1? i dont own one yet Nov 22 04:30:31 that's what i was thinking.. secure private twitter Nov 22 04:30:36 to connect people Nov 22 04:30:36 itshould support different models - publish-subscribe, push-to-box, etc Nov 22 04:30:50 muthu_: It almost seems like you want a Simple Message Transfer Protocol Nov 22 04:30:57 lol Nov 22 04:31:01 :P Nov 22 04:31:04 does it keep a xmpp connection open at all times? Nov 22 04:31:11 meh.. smtp is so old school.. Nov 22 04:31:32 people like twitter Nov 22 04:31:45 twitter is just a new interface on an old idea Nov 22 04:31:54 twitter for jobs, movies, dining.. Nov 22 04:31:56 whatever Nov 22 04:31:56 just use SMTP and tack on a social interface and you're set Nov 22 04:32:07 you're discussing interfaces and use patterns Nov 22 04:32:17 that's an entirely different world than the underyling protocols/tech Nov 22 04:32:19 what do you think? Nov 22 04:32:26 * jasta checks out mobeegal *hehe* Nov 22 04:32:28 your idea is a good one, just be aware of exactly what you're trying to do Nov 22 04:32:28 yeah, forget about protocol/tech Nov 22 04:32:47 jasta: the new idea will replace mobeegal Nov 22 04:32:53 people try and reinvent wheels because they're convinced their idea is so awesome that noone could possibly have thought about it before Nov 22 04:32:54 i'm just poking fun Nov 22 04:33:02 haha.. i know Nov 22 04:33:10 anyway im off gentleman Nov 22 04:33:53 jasta: what you think of this twitter idea? Nov 22 04:35:31 i haven't been listening, but i'm sure in your usual fashion you are just firing at random hoping to get traction rather than thoughtfully working through real problems. Nov 22 04:35:50 very thoughtful Nov 22 04:36:08 there exists a real problem Nov 22 04:36:13 spare me Nov 22 04:36:18 heh Nov 22 04:36:28 its weekend jasta! Nov 22 04:37:36 jasta: what's a good whiskey to try? Nov 22 04:40:27 wow, repo sync went from 1.4 to 2.6.27 Nov 22 04:42:37 actually, 1.4 is the repo version, 2.6.27 is the linux version. Nov 22 04:42:44 oh ok Nov 22 04:42:51 lot of activity today Nov 22 04:43:26 someone went on a tag spree Nov 22 04:44:07 jbq: you have a lot of experience with HttpClient, right? I'm having a very strange issue that I can't figure out. Nov 22 04:44:54 jasta: I have a little bit of experience, yeah. Nov 22 04:45:52 it would seem that my aborted requests are not getting reabsorbed into the connection pool, causing the third and all subsequent requests to block in HttpClient#execute Nov 22 04:46:32 specifically they block in ConnPoolByRoute#getConnection, waiting for a connection to become available Nov 22 04:47:45 in my abort code, i merely call HttpMethod#abort(), then in the download thread i call the equivalent to HttpEntity#getEntity()#close Nov 22 04:47:59 err HttpEntity#close :) Nov 22 04:48:12 or something. no, it's getContent(). anyway... Nov 22 04:48:36 according to the HttpClient docs, that is all you should need but something clearly isn't working right Nov 22 04:48:55 i was hoping to debug it more aggressively but i can't figure out how to attach source to the Android eclipse plugin. Damn :) Nov 22 04:48:55 You might need to close() them as well, though I'm not sure. I don't think that my code tries to execute more than one request per client, so I don't run into those isses. Nov 22 04:49:00 s/isses/issues/ Nov 22 04:49:01 jbq meant: You might need to close() them as well, though I'm not sure. I don't think that my code tries to execute more than one request per client, so I don't run into those issues. Nov 22 04:49:45 jbq: and i wasn't running into it when i did the same. but according to the HttpClient devs, reusing the HttpClient instance is important for performance, and in my case it matters since i have the opportunity to reuse the connections Nov 22 04:50:09 close() what btw? I am definitely closing the InputStream returned by HttpEntity#getContent Nov 22 04:50:32 I don't remember, it's been many months since I did this. Nov 22 04:50:35 :( Nov 22 04:50:47 well, i think i could probably be more productive if i could interactively debug Nov 22 04:50:53 any idea how to attach the source to Eclipse? :) Nov 22 04:51:12 Nope, I've never used Eclipse. Nov 22 04:51:32 sigh, normally it's quite easy but the android plugin disables that functionality Nov 22 04:52:24 Jasta, I don't know how with Eclipse, however, I can get you pointed in the right direction Nov 22 04:52:32 ? Nov 22 04:52:36 jdb -sourcepath$WORKING/src -attach localhost: Nov 22 04:52:49 that'll attach the java debugger to the running package Nov 22 04:52:59 and with your sourcepath, voila Nov 22 04:53:17 hmm. Nov 22 04:53:33 i think it might be more productive to try to look at the android eclipse plugin, which apparently h as some support for this it's just totally undocumented Nov 22 04:54:59 hi guys, Im writing a basic app that reacts when the phone takes a picture Nov 22 04:55:23 b0nn: what it does? Nov 22 04:55:33 i want to know where i find info about when a phone takes a pic, ie. what signals are available Nov 22 04:56:14 muthu_: nothing fancy, it's a practice run to learn about android Nov 22 04:56:38 just for the moment I need to detect that a picture has indeed been taken Nov 22 04:57:44 b0nn, I'd start by looking at the Camera app source, but maybe I'm just a masochist :) Nov 22 04:58:42 are there no ways to interrogate the OS for events? Nov 22 04:58:49 or to be alerted of events? Nov 22 04:59:26 events are all done through broadcast listeners Nov 22 04:59:41 ok Nov 22 04:59:57 could you point me toward some quality documentation on those? Nov 22 05:00:14 the only documentation is android.com Nov 22 05:00:28 the api that comes with your sdk Nov 22 05:10:44 jasta: sigh, normally it's quite easy but the android plugin disables that functionality << I asked our SDK guys and apparently Eclipse doesn't make that easy and/or obvious to do Nov 22 05:10:53 it's not on purpose Nov 22 05:11:21 romainguy: well, there seems to be some feature here to make it work Nov 22 05:11:24 i just cant understand the magic Nov 22 05:11:33 IClasspathEntry classpathEntry = JavaCore.newLibraryEntry(android_lib, Nov 22 05:11:33 android_src, // source attachment path Nov 22 05:11:41 and i found this: Nov 22 05:11:43 http://stuffthathappens.com/blog/2008/11/01/browsing-android-source-in-eclipse/ Nov 22 05:11:47 but it doesn't work for me and i don't know why Nov 22 05:12:52 i think at that point it's just that eclipse expects me to do something i'm not doing. not sure how to work through that :) Nov 22 05:15:08 don't use Eclipse :) Nov 22 05:15:15 or don't use the plugin :p Nov 22 05:15:24 i'm gonna try to figure it out :) Nov 22 05:16:43 jasta: set a normal java project and debug Nov 22 05:23:33 If I want to have something constantly poll a resource (... tail a file, poll a webserver... etc) - and then pass a very simple update to the UI - I take it I should use a service, right? What's the lightest weight way to pass data from that service to the UI layer? a static variable? Nov 22 05:24:06 RyeBrye: you only need a service if you care that this operation lives longer than the activities would normally. Nov 22 05:24:19 jasta: something that may fix your HttpClient issue (I'm not sure, haven't used the connection pool stuff with httpclient at all yet) is to do an HttpClient.getConnectionManager().releaseConnection() for the one you want to abort Nov 22 05:24:43 ismarc: yeah im aware of that method, but i had thought (and httpclient's documentation *seems* to confirm) that it is unnecessary now. Nov 22 05:25:12 that just calling close() on the InputStream rturned from HttpEntity#getContent was equivalent now Nov 22 05:27:07 jasta - so if I don't need a service (since I don't care if it runs in the background longer than the activity) - should I just use a timertask or something to do the polling? Nov 22 05:27:32 i'd just use a worker thread that is doing a sleep. Nov 22 05:27:43 Ok, sounds simple enough. Nov 22 05:27:55 jasta: or use an alarm Nov 22 05:27:58 but do realize that if the CPU falls asleep, your sleep isnt going to return until it wakes back up again. Nov 22 05:28:09 yeah, this wont be too crucial Nov 22 05:28:14 romainguy: well, they're not interchangable. you use one for one sort of task and the other for another. Nov 22 05:28:17 not for this basic app I'm toying with at least Nov 22 05:28:29 jasta: if you need to do something periodically an alarm is as good as it gets Nov 22 05:29:04 romainguy: as i said, they solve different problems. alarms wake the cpu, sleeps and waits do not. it isn't a good idea to just use alarms without thinking about what you want. Nov 22 05:29:30 in his case, if the cpu sleeps, he probably doesnt want to wake it to poll a file to update a screen the user isn't looking at Nov 22 05:30:07 so if that's true, a non-realtime timer (as with Thread#sleep) would be much better. Nov 22 05:30:29 I'm realizing now how woefully un-savvy I am with threads :) Nov 22 05:31:25 does an handler wake up the cpu? Nov 22 05:31:30 jasta: I agree, I just didn't look at what was needed :) Nov 22 05:32:08 * RyeBrye wonders if some apps on his phone are using alarms improperly... because his cpu seems to never sleep Nov 22 05:32:24 RyeBrye: you can check the cpu sleep time in the phone info screen Nov 22 05:32:52 i thought the apm sucked on the g1? Nov 22 05:33:08 romainguy: Right now my asleep time is 340917:10:45.... I'm not quite sure that is correct Nov 22 05:33:10 muthu_: no, handlers use non-realtime timers (wait, in particular) Nov 22 05:33:20 ok, thx Nov 22 05:33:33 My uptime is 340925:33:20 :) Nov 22 05:33:35 RyeBrye: what is the awake time percentage? Nov 22 05:33:55 muthu_: to my knowledge, the only realtime timer in android is with the AlarmManager service. so, everything else you should assume does not wake the cpu. Nov 22 05:34:01 romainguy - well... since the awake time seems to be the only one being listed properly it is 0% since it's got 8:23:02 awake time Nov 22 05:34:20 I'm talking about what's reported in Settings > About Phone > Status Nov 22 05:34:24 jasta: ok Nov 22 05:34:28 Oh, I'm in device info Nov 22 05:35:13 peoples, anyone care to give some tips of changing G1 startup logo (not android logo)??? Nov 22 05:35:34 muthu_: generally, that's a good thing. and something you ought not concern yourself with. Nov 22 05:36:40 only in special circimustances would you actually want to use an AlarmManager instead of another simpler timing mechanism. for instance, if you were implementing a connection keep-alive, you would need an AlarmManager because you really mean to run your code in x minutes, not whenever the phone feels like it. Nov 22 05:37:22 similarly if you wanted to retry failed downloads in the background. you would need an alarmmanager Nov 22 05:37:34 romainguy - this is the screenshot of my device info: http://www.ryebrye.com/images/strange_uptime.png but don't think too much of whatever is causing that strange error... It's probably something unique to my phone Nov 22 05:37:40 speaking of which, we're thinking of changing AlarmManager to gather several alarms together Nov 22 05:37:59 It's unlikely that I have had my phone on for 38 years Nov 22 05:38:06 (as my uptime seems to think) Nov 22 05:38:11 for instance you would just say that you need an alarm every day, or every hour and the manager would try to wake all apps at the same time Nov 22 05:38:37 romainguy: while you're at it, make a keep-alive alarm :) Nov 22 05:38:38 RyeBrye: what does the other screen say? the Status screen? Nov 22 05:38:46 so i can wake my app up at the same time GTalk does ;) Nov 22 05:38:52 lazy :p Nov 22 05:39:12 romainguy - same thing Nov 22 05:39:41 I'm thinking somehow my phone thinks it was booted at 0, because it's been 38 years since 1970, right? Nov 22 05:39:45 RyeBrye: since you rooted and played with the firmware I blame you :)) Nov 22 05:39:57 Yeah, It's almost definitely my fault :) Nov 22 05:40:14 I'll go back to stock stuff later Nov 22 05:41:27 jasta - worker threads look really easy, thanks for the direction Nov 22 05:41:49 i didn't actually imply any particular class by hte way. Nov 22 05:41:53 i just meant a thread which does work :) Nov 22 05:42:19 a thread that possibly all your activities can share in some way. Nov 22 05:42:21 Yeah, I know :) but I just realized I can do it all with an inner class very easily Nov 22 05:42:24 or maybe not, depends on your design i suppose Nov 22 05:42:35 RyeBrye: btw, i have implemented tail :) Nov 22 05:42:47 RyeBrye: btw, if you need a worker thread that talks to the UI, I suggest you to look at the UserTask in Photostream's source code Nov 22 05:42:48 Jasta - where is tail implemented? Nov 22 05:42:59 romainguy - sounds good, I'll go check that out Nov 22 05:43:03 (part of a future version of Android, under the name AsyncTask) Nov 22 05:43:04 http://code.google.com/p/five/source/browse/trunk/five-music/src/org/devtcg/five/music/util/streaming/TailStream.java Nov 22 05:44:06 this class is particular to Five, of course, but i'm sure you could learn something from it Nov 22 05:44:19 Yes, that is helpful Nov 22 05:44:22 thanks Nov 22 05:44:52 i encapsulate it as an InputStream so that i can just do blocking reads from it elsewhere. i expect your usage will want to be a bit different Nov 22 05:45:05 in particular, you'd probably want a handler to communicate with the activity that some new data is available Nov 22 05:45:10 or something. it's up to you :) Nov 22 05:48:06 romainguy: lol, i feel so stupid. my issue attaching the source was that i just didn't hit refresh on the library after i followed that guys howto... Nov 22 05:48:12 lol Nov 22 05:48:22 i restarted eclipse, which is what i assumed to be sufficient Nov 22 05:48:53 * RyeBrye has been bit by the eclipse refresh thing more than once Nov 22 05:49:08 well this is pretty convenient now. i get android source code in eclipse :) Nov 22 05:54:34 jasta: what's the status on your email changes btw? Nov 22 05:54:37 romainguy - so if I'm doing something that will just post strings out of a file using userTask - but will do it indefinitely, it'd just be setting up a Usertask right? Nov 22 05:55:10 yes Nov 22 05:55:14 you use the progress values Nov 22 05:55:19 except the types are Void not void Nov 22 05:55:35 ok Nov 22 05:55:48 yeah, that UserTask looks pretty slick Nov 22 05:55:51 you should get a raise Nov 22 05:55:52 ;) Nov 22 05:56:47 romainguy: currently working, but not yet finished. i need to fix 2 more known bugs and finish merging everything into a separate process Nov 22 05:57:41 i have been talking with someone who has more experience with IMAP than I and has given me a few useful suggestions to handling the connection failure case though. i think i might integrate his suggestions this weekend to fix 1 of those known bugs. Nov 22 05:58:15 splitting the process will be a big pain though, and i'm not sure it's wise to do yet. stadlero (sp?) has cautioned me about pretty big upcoming changes being merged in Nov 22 05:58:28 sp? Nov 22 05:58:32 spelling Nov 22 05:58:39 ah yes Nov 22 05:58:41 Andy Stadler Nov 22 05:58:54 we are working hard to get our mainline onto git Nov 22 05:59:02 we finally have a perforce branch with the git layout Nov 22 05:59:21 it's pretty damn annoying to juggle between the old p4 layout, the new p4 layout and git :) Nov 22 05:59:35 yeah, so i'm thinking i might put all the functional changes i need together and wait for that big dump to come. then work on splitting the processes. Nov 22 06:00:14 the latter issue isn't really hard to do it's just that the diff is going to be nightmarish. just copying code around mostly. Nov 22 06:00:29 yeah tell me about it Nov 22 06:00:53 I have two branches with big changes in Home that make each merge a real pain in the ass Nov 22 06:01:11 my gf is going to be gone for 3 weeks in December for her work, so i bet all of these things will get done then :) Nov 22 06:01:37 jasta: we need to send your gf on vacation, to make your productive :)) Nov 22 06:02:40 romainguy: the design of IMAP is somewhat unfortunate when combined with android's wifi behaviour Nov 22 06:02:52 we should fix the IMAP RFC then ^^ Nov 22 06:02:56 in particular, each time the idling connection is broken and restored, i need to perform part of a full sync. Nov 22 06:03:17 with wifi on, this means each time the phone sleeps and wakes back up this will happen Nov 22 06:03:17 I know there are changes that are going on with WiFi, I don't know if that's gonna help you though Nov 22 06:03:59 i think what i will do is implement some sort of throttling feature. so that it will only do the partial resync so many times per hour Nov 22 06:04:27 if you can make a push that doesn't suck the battery out I'm sold Nov 22 06:04:34 or maybe ill just say "if you use wifi, your battery life will be sucky." :) Nov 22 06:04:35 I had to disable push email on my iPhone Nov 22 06:04:46 it was just awful Nov 22 06:04:54 it would drain the battery on EDGE in half a day Nov 22 06:05:10 romainguy: it's a really hard problem to solve, but it can be done. it works best, of course, if you dont get emails that frequently Nov 22 06:05:19 true Nov 22 06:05:33 but I'm talking about an account that didn't get that many emails per day Nov 22 06:05:37 20 to 50 Nov 22 06:05:43 hence the suck Nov 22 06:05:47 the best thing you can do is just optimize the hell out of the case when idle wakes you up Nov 22 06:05:57 yep Nov 22 06:06:22 well i get about 20 to 50 a day and i can say my code at least beats your experience with the iphone :) Nov 22 06:06:33 I certainly hope so Nov 22 06:06:37 the iphone probably lazily is performing a partial resync each time its woken up Nov 22 06:06:42 because I get in the hundreds a day on my gmail account Nov 22 06:07:34 one thing that can be done is to avoid waking yourself up until some short period after you are tickled Nov 22 06:07:55 that is, when you are alerted to a new message, wait a minute, then react. this helps for the case when you get flooded with lots of messages all at once. Nov 22 06:08:19 i can tell the gmail app on android does this. either that or its really slow ;) Nov 22 06:08:27 you can see it get tickled and then like a minute later say something about it Nov 22 06:09:05 waiting for a minute (or more) seems like a sensible approach Nov 22 06:09:11 it's not like you need this to be true realtime Nov 22 06:09:26 anyway, decent battery life with push e-mail is a possibility. i will do my best ;) Nov 22 06:09:29 and with high frequency mail account like the ones we have at google, that must help a lot Nov 22 06:09:49 however, the _best_ possible battery utilization would happen only if GTalk and my/other apps could share the CPU :) Nov 22 06:10:07 *hint hint* Nov 22 06:10:09 if its's high frequency, wouldn't pull email on a regular schedule work ok? Nov 22 06:10:53 RyeBrye: that's up to the user to decide. personally i would find that on a high volume inbox i'd rather not have my phone use push. i would find it pretty annoying :) Nov 22 06:13:03 it is annoying :) Nov 22 06:13:21 actually I have to admit I don't find push email that useful Nov 22 06:13:32 a 30 minutes/1 hour pull frequency works great for my use Nov 22 06:13:43 guys, in android mtd2 image there is a file logo.rle, anyone know what that file is??? is it the first boot G1 logo???? Nov 22 06:13:56 romainguy: really, you guys should consider offering a way for apps to hook into a reasonable alarm to avoid connection resets. Nov 22 06:14:05 file a feature request Nov 22 06:14:14 I do pixels and widgets :) Nov 22 06:14:23 this way the platform could abstract network-specific details which developers are unlikely to ever acquire Nov 22 06:14:27 :) Nov 22 06:14:27 hehe Nov 22 06:17:55 bbl, back to fixing my busted HttpClient usage :) Nov 22 06:22:45 ohhhh, interesting :) Nov 22 06:23:27 source code sure did help a lot on that one. stumbled upon a comment while stepping that tells me exactly why my usage is wrong Nov 22 06:24:01 if you call HttpMethod#abort from a separate thread in order to unblock the I/O, you still must also call the same method on the same object in the downloading thread Nov 22 06:24:27 you mean you need to call abort() from both threads? Nov 22 06:24:29 the method behaves differently depending on which thread its called from. only in the connection thread will it release the connection. Nov 22 06:24:56 romainguy: yes, abort() from a separate thread will merely unjam the blocking thread. called from the jammed thread, it will release the connection. Nov 22 06:25:22 it will throw an IOException, which you are required to respond to with method.abort() Nov 22 06:25:40 I have to admit HttpClient is powerful Nov 22 06:25:44 but I don't like the APIs Nov 22 06:25:48 i thought i was being clever by saying if isAborted == false, abort Nov 22 06:28:17 hah, market comments got disabled? Nov 22 06:30:36 * jasta tests... Nov 22 07:04:10 romainguy: as it turns out, i think you are are actually not supposed to call HttpMethod#abort again, but rather in the case that you shutdown from a separate thread you must work out your own way to call releaseConnection Nov 22 07:04:38 which is pretty bizarre to me, but whatever Nov 22 07:05:11 normally they provide a handler ofr IOException, calling checkAbort which will abort and release your connection Nov 22 07:05:40 it checks if the connection has already been aborted and does nothing if it has. your separate thread sets this flag, so the only thread that could handle releasing the connection won't do it. Nov 22 07:05:59 very odd Nov 22 07:06:13 i realize now that their example is not like mine, because their example is single threaded Nov 22 07:32:27 romainguy: i give up, im calling this a bug in HttpClient :\ Nov 22 07:32:45 im gonna just detect this case specially (abort) and tear down the whole HttpClient instance. Nov 22 07:53:27 hi there. i'm not sure if there'll be some guide on hacking android's source Nov 22 07:53:29 ?? Nov 22 08:12:57 romainguy - that UserTask is really sweet Nov 22 08:13:13 romainguy - thanks :) I think it does exactly what I was looking for :) Nov 22 08:13:46 glad you like it Nov 22 08:15:02 It's perfect for those of us who are thread-retards :) Nov 22 08:15:26 to me it was useful mostly to avoid having Handlers all over the place Nov 22 08:16:20 Yeah, that's nice about it too Nov 22 08:16:48 I'm just doing a basic throw-away app now to get my feet wet, but I'll definitely use that more when I start doing real apps Nov 22 08:17:57 I need to finish my app @!# Nov 22 08:18:09 Which app? Nov 22 08:19:18 http://progx.org/users/Gfx/shelves.mp4 Nov 22 08:19:53 (this version had drawing bugs) Nov 22 08:29:41 yes, you do Nov 22 08:29:54 :) Nov 22 09:09:28 <[ShadowGhost]> yoooo Nov 22 10:06:37 romainguy: you there? Nov 22 10:07:14 * RyeBrye finished making a drawing app that will track two fingers... (kind of) and is off to bed Nov 22 10:07:36 i hate multitouch actually :) Nov 22 10:07:50 I want it for zoom, and that's about it Nov 22 10:07:54 i played with it just yesterday on an iphone and i think it is awkward Nov 22 10:08:01 oh, and 2-player air hockey Nov 22 10:08:37 Yeah, it's not the pinnacle of human existence or anything, but it's nice to have in some situations Nov 22 10:09:12 I hate those damn zoom buttons on the browser... I frequently have to interact wth stuff that they cover up so I have to scoot the page over just to hit a submit button or something Nov 22 10:09:18 jasta: yup? Nov 22 10:09:28 I don't like multitouch either Nov 22 10:09:29 romainguy: i'm curious for your opinion on something Nov 22 10:09:34 it's nice when you try it Nov 22 10:09:39 then never use it again :) Nov 22 10:09:47 the only multitouch I like is two fingers to scroll with a trackpad Nov 22 10:09:52 jasta: sure, shoot Nov 22 10:10:11 i've been thinking about downloading thumbnails during ifve's initial sync (instead of the full sized versions), and downloading the full-sized versions async when it needs to display album art Nov 22 10:10:20 sounds good Nov 22 10:10:31 except the problem is that five of course might be also trying to stream the music you chose Nov 22 10:10:40 so you certainly wouldnt want to steal bandwidth away from that Nov 22 10:10:52 what size of thumbnail were you thinking about? Nov 22 10:10:59 1920 x 1080? Nov 22 10:11:01 ;) Nov 22 10:11:04 * RyeBrye runs off to bed Nov 22 10:11:13 romainguy: the thumbnails are 64x64, full-sized are 300x300 Nov 22 10:11:13 because a 64x64 jpg should use little bandwidth Nov 22 10:11:31 especially if you use high levels of compression Nov 22 10:11:32 but the full-sized ones would be the ones that need to be downloaded on demand, as with potentially the music itself. Nov 22 10:12:23 what if you use a low priority thread? Nov 22 10:12:34 it will take time to download the art, but that's not the most important thing Nov 22 10:12:35 would that mean anything? Nov 22 10:13:03 without knowing anything about your scheduler, i would be surprised to find that it could impact low speed download performance Nov 22 10:13:21 we do use priorities in mainly places Nov 22 10:13:22 the issue is forcing the radio to compete between downloading album artwork (not important) and streaming music (very important) Nov 22 10:13:33 I know a lot of work went into the scheduler to give us better control on the threads Nov 22 10:14:24 hmm. i would be surprised if that could do anything meaningful generically Nov 22 10:15:06 could you cook a quick test? Nov 22 10:15:18 possibly. i might have to do that. Nov 22 10:15:57 the initial sync still has to download something like 25MB worth of artwork for my collection (70GB of music). which is a lot :) Nov 22 10:16:12 on 3G it's not that much Nov 22 10:16:27 the total throughput ends up being about 32MB with database. Nov 22 10:16:49 but i suppose t-mobile's 3g is about 700-900kbps, so that would be... Nov 22 10:17:07 well I regularly download 50 MB podcasts in the shuttle in the morning Nov 22 10:17:13 and listen to them in the shuttle when the download is over Nov 22 10:17:25 and that's on a shuttle driving on the highway Nov 22 10:18:10 hmm. how long does that take usually? i just calculated my case to be about 6 minutes... Nov 22 10:18:29 hmm in my case, because it switches to edge from time to time, it takes about 20 minutes Nov 22 10:18:38 which I would find tolerable for a first time sync Nov 22 10:18:47 depends on what you want the user experience to be Nov 22 10:19:00 I personally wouldn't mind doing the first sync over WiFi for instance Nov 22 10:19:09 yeah i suppose Nov 22 10:19:11 anyway, time to go to bed Nov 22 10:19:14 good luck :) Nov 22 10:19:15 night Nov 22 12:05:10 HI Nov 22 12:36:36 hi Nov 22 13:07:04 chillen Nov 22 13:54:35 ? Nov 22 13:57:51 what Nov 22 13:58:44 in connectbot, how do i detach a screen? (ctrl-a d) Nov 22 13:58:52 are there any IRC clients for android? Nov 22 13:59:12