**** BEGIN LOGGING AT Sat Oct 04 02:59:57 2008 Oct 04 03:16:34 erre Oct 04 04:45:23 oy Oct 04 05:17:06 *coding* :) Oct 04 05:19:29 jksdfds Oct 04 05:25:19 01:23:29 up 141 days, 58 min, 1 user, load average: 0.30, 0.27, 0.26 Oct 04 05:25:21 woot. Oct 04 05:27:23 22:25:55 up 111 days, 19:09, 4 users, load average: 0.07, 0.07, 0.02 Oct 04 05:29:17 22:27:46 up 298 days, 6:43, 1 user, load average: 0.06, 0.03, 0.00 Oct 04 05:29:18 base Oct 04 05:29:25 hmm. Oct 04 05:29:28 http://i33.tinypic.com/2hgqxkg.png Oct 04 05:29:30 oh and Oct 04 05:29:33 22:27:40 up 137 days, 11:37, 5 users, load average: 0.06, 0.06, 0.02 Oct 04 05:29:37 you got nothin' ;P Oct 04 05:29:48 these are all production servers btw. Oct 04 05:29:53 yeah, so is mine Oct 04 05:30:10 pft 137 days Oct 04 05:30:13 my load average is much higher than yours Oct 04 05:30:13 except, the 137 is actually my personal server at home. terabyte raid5 :) Oct 04 05:30:18 better resource usage. Oct 04 05:30:21 18:28:55 up 559 days, 5:49, 1 user, load average: 0.41, 0.33, 0.28 Oct 04 05:30:23 than any of your boxes Oct 04 05:30:27 wow im kind of proud of myself Oct 04 05:30:28 22:02:22 up 209 days, 1:05, 2 users, load average: 0.00, 0.04, 0.00 Oct 04 05:30:32 now, see, duey's I can respect Oct 04 05:30:33 never would have thought Oct 04 05:30:35 duey ftw Oct 04 05:30:36 his machine has been doing stuff. Oct 04 05:30:46 I see it as duey > kraln > * Oct 04 05:30:51 =p Oct 04 05:31:04 duey: what machine is that, and what os? Oct 04 05:31:04 it seriously needs a reboot Oct 04 05:31:10 but the uptime is to awesome Oct 04 05:31:13 its debian Oct 04 05:31:15 etch Oct 04 05:31:19 in a data center Oct 04 05:31:24 * yakischloba wonders what uptimes he could find on his network Oct 04 05:31:24 what does it do? Oct 04 05:31:30 its my personal box Oct 04 05:31:35 does email/web/stuff Oct 04 05:31:38 granted, but what does it do Oct 04 05:31:43 to give it that load Oct 04 05:31:45 or is it underpowered Oct 04 05:31:47 http://in.youtube.com/watch?v=HuwhTmM_QGQ&NR=1 Oct 04 05:31:53 Kraln: maybe my systems are just more efficient and need not waste so much time because of poor administration? :) Oct 04 05:31:58 Kraln, hosts a few websites Oct 04 05:32:03 friends etc Oct 04 05:32:09 I think some of them are quite popular Oct 04 05:32:55 jasta: uh, no? if your system load is consistently low, and you have 2+ machines, you're making poor use of your resources Oct 04 05:33:12 Kraln: it's called mission critical redundancy ;) Oct 04 05:33:22 yakischloba: no, not really Oct 04 05:33:23 does system load take into account I/o wait? Oct 04 05:33:28 duey: yes Oct 04 05:33:41 heh then Oct 04 05:33:45 it's a rough measure of how long things wait to get cpu time Oct 04 05:33:55 18:31:35 up 143 days, 3:21, 2 users, load average: 0.00, 0.00, 0.02 Oct 04 05:33:55 one main production machines @ work Oct 04 05:34:09 lol Oct 04 05:34:16 We have two commodore amiga's at our datacenter. If there was any correlation between age and uptime, I'd probably have a winner lol Oct 04 05:34:21 see, good production machines should not be over-utilized. Oct 04 05:34:22 18:32:50 up 3 days, 4:18, 1 user, load average: 0.40, 0.69, 0.66 Oct 04 05:34:22 db server Oct 04 05:34:33 (it just had software updates) Oct 04 05:34:35 there, that is a server doing its part. Oct 04 05:34:48 I see machines with load averages of like Oct 04 05:34:49 20 Oct 04 05:34:52 :| Oct 04 05:34:56 that's bad Oct 04 05:35:00 granted you can barely use a shell on them ;) Oct 04 05:35:06 I had a machine with 4 cores that'd get up to 20 Oct 04 05:35:07 yeah. thats our customers. Oct 04 05:35:09 if they got slammed Oct 04 05:35:18 btw, load averages are based on 1, 5, and 15 minute periods Oct 04 05:35:31 so, being 10:30pm on a Friday, it is not hard to see why my work servers haven't seen any action Oct 04 05:35:37 jasta: protip: load averages are an average cpu load Oct 04 05:35:38 yeah same here Oct 04 05:35:47 they are all fairly low, weekend Oct 04 05:36:31 Kraln: it's more complex than that, but its based on averages over relatively short periods of time Oct 04 05:36:46 any reasonable average for a production machine would be based on a monthly cycle, not the last 15 minutes Oct 04 05:36:46 it's not really even averages Oct 04 05:37:11 my average load average is about 1 Oct 04 05:37:15 http://i33.tinypic.com/2hgqxkg.png (see top graph) Oct 04 05:37:45 * jasta goes back to hacking Oct 04 05:39:50 hmm Oct 04 05:40:19 * umdk1d3 looks for small spinning wheel animation Oct 04 05:45:37 umdk1d3: progress bar? Oct 04 05:46:01 is there any generalization to do a wait for a monitor lock with a timeout not bothered by an interrupt? Oct 04 05:46:45 something to generalize the timeout -= (now - then) if (timeout <= 0) { timed out! } foo.wait(timeout);? Oct 04 05:47:04 look in java.util.concurrent Oct 04 05:47:08 there's everything you need Oct 04 05:47:50 hmm, i've looked before, but didn't find it? Oct 04 05:49:38 what i'm wanting to do is Object#wait up to some timeout, but such that the wait is resumable if interrupted before the timeout has been reached. Oct 04 05:50:03 its a simple like 5 or 6 line loop to do, but feels clumsy not to have been generalized somehow... Oct 04 05:51:33 if a wait is interrupted, there's probably a good reason Oct 04 05:51:53 yeah, probably, but if there wasn't, i still want my code to work as it was designed. Oct 04 05:52:04 that feels really weird to me to do this Oct 04 05:52:19 but it reminds me of the pattern for wait/join Oct 04 05:52:25 why? Oct 04 05:52:36 where you do use a loop with a condition to protect the wait from spurious wakeups Oct 04 05:52:53 right, which is what you should do Oct 04 05:53:14 well reading your code above it seems like you want to ensure you'll wait for a given period of time right? Oct 04 05:53:45 the basic logic is: Oct 04 05:54:11 while (somethingHappened == false) { somethingHappenedLock.wait(timeout); } Oct 04 05:54:29 actually, that doesn't express it well without all the other junk Oct 04 05:54:46 ok, so why the need to check for the timeout if wait was interrupted? Oct 04 05:54:52 ah ok I see Oct 04 05:54:55 what i'm doing is waiting up to 1s for an event to occur Oct 04 05:55:04 muthu_: perfect :) progressbar works i thought there was an animation or something Oct 04 05:55:06 if the wait is interrupted before 1s, and the event has not occurred, i want to *keep waiting8 Oct 04 05:55:23 which is what you do with the while (!somethinghappened) already Oct 04 05:55:28 in particular, if woken up 500ms in, i want to wait with a timeout of another 500ms. Oct 04 05:55:33 but you might wait for more than 1s Oct 04 05:55:46 romainguy_: except that if the event DOESNT occur after 1s, i want to abort and give up. Oct 04 05:56:34 yeah I got that Oct 04 05:56:56 the logic is really simple. you just take the current time, measure it after the wait and if the condition is false, and wait the remaining time Oct 04 05:57:10 er, yes, that's what you wrote above Oct 04 05:57:14 but it also makes a very simple expression of a timed wait into something much longer. Oct 04 05:57:26 and not very general Oct 04 05:57:53 not really sure how i'd generalize this... Oct 04 06:23:42 hmm i think i ran into this before, but is there a way of forcing an activity into touch mode instead of kbdmode? Oct 04 06:23:50 last time i think i didnt find a solution :/ Oct 04 06:24:13 umdk1d3: easy, don't Oct 04 06:24:49 well its going into the activity in keyboard mode, which is selecting my footerview by default Oct 04 06:25:04 when i refresh the adapter with the "page" of results, it keeps the original footer view selected Oct 04 06:25:10 which continues triggering ALL the next-pages :/ Oct 04 06:25:17 but in touch mode, it stays at the top of the list Oct 04 06:25:19 then set the selection to the position 0 Oct 04 06:25:25 but do NOT try to force touch mode Oct 04 06:25:28 just don't Oct 04 06:25:31 kk Oct 04 06:25:39 or I'll personally come kick your butt :) Oct 04 06:25:51 :P Oct 04 06:26:34 i should have this finished this weekend at some point and then ill gpl it and throw an apk out there Oct 04 06:26:42 your remote? Oct 04 06:26:45 or something different? Oct 04 06:26:45 yep Oct 04 06:26:49 have you talked to xav? Oct 04 06:26:54 he showed me an app he's working on Oct 04 06:26:59 using zeroconf in pure Java Oct 04 06:27:04 ooh nice :) Oct 04 06:27:05 thought you might be interested Oct 04 06:27:10 does it work over Wi-Fi? Oct 04 06:27:12 yea im borrowing the jmdns pure-java mdns package Oct 04 06:27:17 jasta: yea it kinda /has/ to use wifi Oct 04 06:27:19 that's what he's using Oct 04 06:27:49 umdk1d3: well, it doesn't have to. it could work over EDGE/3G. Oct 04 06:27:49 ive got it pairing with itunes over the air fine now :) there was somethign wierd in the hotel with my vmware bridge keeping it from working Oct 04 06:28:01 jasta: lol home stereo control over edge Oct 04 06:28:14 then youd have to tell them all about port fwding, etc Oct 04 06:28:19 well actually, yes. it wouldn't be that bad, actually Oct 04 06:28:24 hmm Oct 04 06:28:40 well it certainly wouldn't be a bad idea to support it at some point. remember, Wi-Fi is no guarantee on an Android phone. Oct 04 06:29:03 hm true Oct 04 06:29:07 not every application has to be easy for dolts to use either. something to keep in mind :) Oct 04 06:29:51 and an easy way to filter out those features is a simple "Oh by the way, check the README for other features." :) Oct 04 06:30:00 filter out those people, i mean. Oct 04 06:31:29 umdk1d3: in case you're interested, I added a new feature to Photostream Oct 04 06:31:39 it shows how an app can "enhance" web sites by responding to their URL Oct 04 06:32:04 hm so responding to their authority? Oct 04 06:32:07 yes Oct 04 06:32:10 i could overload google.com! Oct 04 06:32:10 =D Oct 04 06:32:11 muihaha Oct 04 06:32:16 in this case, you click a flickr.com/photos/ link and you get to choose between seeing it in the browser or in photostream Oct 04 06:32:25 sure, but you will get a dialog first :)) Oct 04 06:32:28 very nice, i was wondering if that was possible for end-users Oct 04 06:32:36 i know the maps.google.com thing is intercepted Oct 04 06:32:46 all apps are equal :) Oct 04 06:32:51 ^.^ Oct 04 06:38:13 hm Oct 04 06:38:29 so i have my footerview and im View.GONE'ing it, but it still takes up the space in the list Oct 04 06:38:58 might just do removeFooterView Oct 04 06:39:00 * umdk1d3 tries Oct 04 06:39:25 just set the selection to 0 when you get at least one item in the adapter? Oct 04 06:39:43 ah sorry Oct 04 06:39:45 misread Oct 04 06:39:50 yeah ListView doesn't care about GONE Oct 04 06:40:06 yea ive got that part working now, i just selection=0 on the first page of results thats returned Oct 04 06:41:32 yay removeFooter worked :) Oct 04 06:42:45 hmkay something ive been meaning to write for along time is something that loads http images from a separate thread and fills them in as they arrive (into a listview) Oct 04 06:43:09 currently ive written it as a pair (image->view) that gets thrown off to fetch the image Oct 04 06:43:33 then a handler takes an event back with the image->view pair and updates them Oct 04 06:43:42 umdk1d3: UserTask is your friend :) Oct 04 06:43:46 it's exactly what Photostream does Oct 04 06:44:07 but this causes an issue when we recycle views, as a recycled view will be filled with a stale image Oct 04 06:44:27 how so? Oct 04 06:44:46 * umdk1d3 peeks at usertask first Oct 04 06:44:59 so i pass off the imageurl->view to a separate fetching thread Oct 04 06:45:22 if the user scrolls past and that view happens to be recycled, the thread will shoot off the first image to the now-recycled view Oct 04 06:45:22 you should pass the position instead of the view Oct 04 06:45:40 and then do something list getChildAt(position - getFirstVisiblePosition()) Oct 04 06:45:43 to get the View Oct 04 06:46:07 oh and check to see if the incoming request is outside of view Oct 04 06:46:14 *outside of the visible scroll range Oct 04 06:46:17 exactly Oct 04 06:46:36 okay cool that should work Oct 04 06:47:49 Good morning guys. :-) Oct 04 06:53:31 oh usertask is AWESOME Oct 04 06:56:04 umdk1d3: told you :p Oct 04 06:57:07 why hasnt it been rolled into the framework yet?! :P Oct 04 06:57:19 I wrote it too late for 1.0 Oct 04 06:57:24 but it's on my list for 1.1 Oct 04 06:57:39 I just need to rename it, clean up the doc a bit and it should be good to go Oct 04 06:58:32 hmm Oct 04 06:59:03 ohbtw i couldnt get any of that caching stuff to work :/ Oct 04 06:59:08 webkit appears to be jealous Oct 04 06:59:17 *android.webkit.CacheManager that is Oct 04 06:59:36 just make your own cache :p Oct 04 06:59:51 well yea, i have that written already, but i dont think theres anyway it could use the /cache partition Oct 04 07:00:04 no you probably can't Oct 04 07:00:09 use /sdcard if available Oct 04 07:00:23 or use Context.getCacheDir() or something like that Oct 04 07:00:34 ooh! maybe we can write to cache then Oct 04 07:01:02 getCacheDir() returns a path to /data if I remember correctly Oct 04 07:01:04 but give it a try Oct 04 07:01:06 maybe I'm wrong Oct 04 07:01:42 oh ouch :/ Oct 04 07:02:26 at least the cache dir is managed by the system Oct 04 07:03:31 yea getCacheDir() is in /data Oct 04 07:03:42 unless its some sort of transparent symlink over to /cache Oct 04 07:03:48 I doubt it Oct 04 07:04:24 that doesnt make sense tho--we should avoid caching on /data, and yet getCache returns there :P Oct 04 07:05:08 another day.. another ubuntu problem Oct 04 07:05:17 scim refuses to work.. any ideas Oct 04 07:05:24 :( Oct 04 07:05:37 man, my hours are spend looking at linux Oct 04 07:05:53 install / update / remove / reinstall Oct 04 07:06:12 fedora / ubuntu / back to fedora / again ubuntu Oct 04 07:06:22 muthu_: thats kinda why i swtiched to gentoo Oct 04 07:06:35 i was always having apt repos get into broken states Oct 04 07:06:47 woah gentoo Oct 04 07:06:48 was a huge headache, esp because i liked to stay on cutting edge of stuff Oct 04 07:06:59 gentoo is debian based Oct 04 07:07:03 ? Oct 04 07:07:09 umm lol gentoo is gentoo based :P Oct 04 07:07:15 hehe Oct 04 07:07:23 umdk1d3: you can cache on /data with getCacheDir because the cache dir content is cleaned up by the system when storage runs low Oct 04 07:07:37 but probably not recommended? Oct 04 07:07:53 hmm Oct 04 07:07:58 taht I don't know Oct 04 07:07:59 ask Dianne Oct 04 07:08:03 im thinking ill just use getCacheDir() Oct 04 07:08:10 if it changes to use /cache in the future, perfect ;) Oct 04 07:08:19 I use the sdcard in my books app Oct 04 07:08:23 works fine :) Oct 04 07:09:33 for the image cache i wrote, it had a db and code to handle deleting old stuff, but i could just rely on getCacheDir() to handle all that Oct 04 07:10:17 db + sdcard is what the pictures app does Oct 04 07:15:06 http://clanbase.ggl.com/news.php?nid=299290 Oct 04 07:15:06 cool Oct 04 07:15:09 that was a nice game Oct 04 07:22:46 gah /me wishes java had tuples Oct 04 07:22:56 well Oct 04 07:22:59 it's not as good Oct 04 07:23:05 but you have varargs :) Oct 04 07:23:12 doesn't work for return values though Oct 04 07:23:26 exactly, im trying to pass the position+image as a Result ;) Oct 04 07:23:36 use an array Oct 04 07:23:41 or a custom class with 2 public fields Oct 04 07:23:42 Object[] true Oct 04 07:24:00 i was writing it as a new class with 2 fields, but object[] might be faster Oct 04 07:24:12 not necessarily Oct 04 07:24:18 it's less readable :( Oct 04 07:24:26 I don't think it's worth the speed different Oct 04 07:24:33 but in the end it's really up to you Oct 04 07:30:28 romainguy_: How much does using GPS drain the battery in the G1. Or differently how much differs the battery life of GPS off vs. GPS on in the G1. In the end is it possible to use apps that need GPS all the time. Oct 04 07:30:55 swetland could give you precise numbers Oct 04 07:31:06 but having the GPS on all the time drains the battery pretty quickly Oct 04 07:31:14 anno^da: i doubt the G1 is terribly different than other devices with GPS. GPS drains a lot. Oct 04 07:31:27 and I mean *quickly* :) Oct 04 07:31:30 Well yeah I thought so. :-) Oct 04 07:31:31 the 3G radio is also a massive drain on battery life Oct 04 07:31:44 3G is also really not that bad Oct 04 07:31:50 at least the G1 does very well Oct 04 07:32:27 i'm not sure i buy that...i've had quite a few devices with 3G radios and it's significant for sure. Oct 04 07:32:34 I'm just thinking about some apps for example Locale. They are partly based on GPS so how do they save battery life? Setting up profiles based on your location. Oct 04 07:33:05 They need your current location all the time. Oct 04 07:33:14 they could poll from time to time Oct 04 07:33:19 anno^da: well, Locale probably uses triangulation and periodic polling. Oct 04 07:33:24 Well ok. Oct 04 07:33:27 location based on the cell ID is probably enough Oct 04 07:33:27 google's "my location" is pretty cheap, actually. Oct 04 07:33:41 it's just triangulation with data the cellular radio has anyway Oct 04 07:34:04 the only piece that the radio can't tell you is the location of the towers, but since you aren't often moving between towers, it won't need to update that often either. Oct 04 07:34:18 Yeah thats true. Oct 04 07:34:44 google's "my location" almost certainly works by measuring the delay in communication between the cell towers that the phone is communicating with, then triangulating. Oct 04 07:34:50 okay Oct 04 07:34:52 How long does the G1 take to get a position? (cold starting GPS) Oct 04 07:34:52 so simple math on top of numbers that are always available to the radio Oct 04 07:34:53 so i had a genius moment Oct 04 07:35:05 with mickrobk the other day Oct 04 07:35:16 we freaking realized how google is doing some of their triangulation stuff Oct 04 07:35:25 anno^da: it varies Oct 04 07:35:37 all of their streetview vehicles were wardriving! :evilgrin: Oct 04 07:35:45 it TOTALLY make sense Oct 04 07:36:08 they would have an internal database of all the wifi networks to get better detail in certain areas =D Oct 04 07:36:14 umdk1d3: actually, they did it with devices that had both GPS and GSM radios and mapped out the cell tower locations. Oct 04 07:36:15 no need to lease it from someone else Oct 04 07:36:39 jasta: well gsm cellid has been around for awhile, and i think you might just license that from the cell providers Oct 04 07:36:46 yahoo has had an api for years iirc Oct 04 07:36:49 this is not new technology by any stretch. my friend worked with this at Intel research *years* ago. Oct 04 07:37:07 in fact, he had a functioning version of it on early Windows Mobile phones in the Seattle metro area. we used it all the time. Oct 04 07:37:34 umdk1d3: yeah, but you don't *have* to license it from them. Oct 04 07:37:55 besides, the licensing from the carriers is absurd. or, it used to be. you'd have to pay a fee *per* lookup. Oct 04 07:38:25 this is what inspired Intel research to start figuring out a way to subvert them. I mean really, this technology was functional in 2003. Oct 04 07:38:39 romainguy_: Hmm whats the average. What would you say. Oct 04 07:39:05 there were two big problems though: the cell radios were black boxed so you couldn't get this data on most phones, and the cell tower database was huge and hard to map. Oct 04 07:39:21 anno^da: I honestly don't know, I've seen fixes obtained with a delay of a few seconds up to a few minutes Oct 04 07:39:27 depends on the place :) Oct 04 07:39:30 the former i think is still true, but the latter is a problem google can solve with its army of interns Oct 04 07:39:40 lol army of interns Oct 04 07:39:44 ok thanks. :-) Oct 04 07:41:49 * jasta wonders if he was actually supposed to talk about the fact that Intel research was doing this Oct 04 07:42:03 i guess i never thought to ask if that information i got was through a breach of NDA :) Oct 04 08:19:08 O.O Oct 04 08:19:08 10-04 02:16:26.016: ERROR/AndroidRuntime(8570): java.util.concurrent.RejectedExecutionException Oct 04 08:21:05 hmm Oct 04 08:28:17 http://gizmodo.com/5058030/nokia-5800-xpressmusic-hands+on-with-nokias-first-s60-touch-phone <- has also some nice hardware specs. Oct 04 08:29:14 hmm Oct 04 08:29:39 i wonder if there is a good way to figure out if a getView() from an adapter is really "important" Oct 04 08:29:50 like if its going to be in the final landing spot of a fling Oct 04 08:30:11 I know it is Symbian but 3,5mm jack, 8GB card included etc. sounds pretty nice. Oct 04 08:30:11 so that i coudl skip loading images for "in-between" views Oct 04 08:32:11 umdk1d3: use the OnScrollListener for this Oct 04 08:32:19 aha Oct 04 08:32:20 I use this for this exact reason in my books app Oct 04 08:32:26 there's an API demo of it too Oct 04 08:32:31 that shows exactly this use case Oct 04 08:32:48 and should i be using WeakReferences to try creating an inmemory cache instead of reading from disk each time? Oct 04 08:33:05 trying to figure out Weak vs Soft Oct 04 08:33:22 no, soft references for a cache Oct 04 08:39:48 umdk1d3: sounds like youre working on something interesting over there :) Oct 04 08:50:02 holy CRAP!@!!! this thing is **FAST** Oct 04 08:50:06 thanks romainguy_ =D Oct 04 08:50:15 what is fast? Oct 04 08:50:27 im pulling album covers as part of the itunes search thingy Oct 04 08:50:34 ah :)) Oct 04 08:50:35 and with both disk+memory caching it sings ^.^ Oct 04 08:50:42 ^^ Oct 04 08:50:49 its unreal how fast its working Oct 04 08:55:31 LOL its faster than the ipod touch Oct 04 08:55:39 the touch looks choppy next to it :) Oct 04 08:57:13 umdk1d3: awesome :))) Oct 04 08:58:32 what are you comparing? Oct 04 08:58:33 lists? Oct 04 08:59:28 yea scrolling through lists that have album cover art in them Oct 04 08:59:33 nice Oct 04 08:59:34 both fetched from same itunes on the backend Oct 04 08:59:38 yay for ListView :p Oct 04 08:59:42 :) Oct 04 09:04:32 i'm really happy with what ListView lets us get away with Oct 04 09:04:41 like that SEARCH intent is so easy to provide knowing how efficient ListView is. Oct 04 09:05:00 just toss up a list with everything your app supports and use a filter query. Oct 04 09:05:27 very tidy for both user and dev Oct 04 09:07:05 at some point we even played with a ListView of WebViews for Gmail Oct 04 09:07:17 that was a bit too much though :) Oct 04 09:07:34 (the initial layout of a WebView is too expensive for nice scrolls in a ListView) Oct 04 09:07:47 of course itw asnt listviews fault :) Oct 04 09:08:00 not at that time Oct 04 09:08:07 you have no idea how much work went into ListView :p Oct 04 09:08:34 but when the initialization of a WebView takes several hundred milliseconds, it's hard to get a smooth scroll Oct 04 09:09:27 i can imagine, actually. i've attempted something like it a long time ago Oct 04 09:09:42 a fast, "windowed" list view...i failed, however :) Oct 04 09:09:44 list view is great Oct 04 09:09:56 muthu: still needs some work though Oct 04 09:10:12 the default scroll bar coulda been better Oct 04 09:10:23 well that has nothing to do with ListView :) Oct 04 09:10:26 i sort of like it being small and sexy, actually Oct 04 09:10:55 and i like the fade to show you that there are more items at the top/bottom Oct 04 09:11:05 a lot more built in layouts for listviews would be nice Oct 04 09:11:13 umdk1d3: do a screencast :-) Oct 04 09:11:15 the fade was fun to implement Oct 04 09:11:37 romainguy_: Are there any plans to allow adapters to invalidate at a more granular level? Oct 04 09:11:39 romainguy_: it's actually great because it's both visually rich and informative. Oct 04 09:11:40 and a nice headache to make it fast during a ListView fling Oct 04 09:11:56 jasta: my favorite feature about it is how it grows smoothly Oct 04 09:12:09 hmm? Oct 04 09:12:14 the invalidation is a bit buggy Oct 04 09:12:14 tomgibara: it's in a bug somewhere, but very low priority Oct 04 09:12:20 muthu: er, no Oct 04 09:12:27 hehe, romain will kill you Oct 04 09:12:31 haha Oct 04 09:12:31 jasta: go to the top of a list, then scroll slowly Oct 04 09:12:37 you will see the fade gradually grow Oct 04 09:12:42 it doesn't toggle Oct 04 09:13:11 romainguy_: understandable, it hasn't actually caused me any performance problems Oct 04 09:13:25 tomgibara: we'd rather make a full screen invalidation fast first Oct 04 09:13:39 and then add a fine grained invalidate :) Oct 04 09:13:57 the problem with the fine grained invalidate is when the adapter is backed by a database Oct 04 09:14:33 okay Oct 04 09:15:36 hey, noticed that the cursor adapter does not auto update the list view Oct 04 09:15:41 when the content changes Oct 04 09:15:51 do we have to request layout? Oct 04 09:16:08 you have to call notifyDatasetChanged Oct 04 09:16:16 why is it? Oct 04 09:16:19 and one of the constructors of CursorAdapter allows you to setup auto requeries Oct 04 09:16:28 hmm, ok Oct 04 09:16:29 muthu: to notify ListView that something has changed Oct 04 09:16:37 right, but thought that's the default Oct 04 09:17:04 this view/adapter/cursor thing is nice Oct 04 09:17:12 vac patter Oct 04 09:17:16 pattern Oct 04 09:18:08 avc rhymes more with mvc Oct 04 09:18:42 hmm with expandablelists, can we use getChildAt(int) where the int is a packed position? (packed positions are long though :/) Oct 04 09:19:06 yeah, the expandlist demo shows that Oct 04 09:19:12 there are methods to do the conversion between packed/unpacked positions Oct 04 09:20:35