fluid-work IRC Logs-2008-04-23

[06:39:31 EDT(-0400)] * phiggins (n=dante@c-68-34-199-67.hsd1.tn.comcast.net) has joined #fluid-work
[09:27:53 EDT(-0400)] * anastasiac (n=team@142.150.154.105) has joined #fluid-work
[09:47:59 EDT(-0400)] * theclown (n=theclown@142.150.154.101) has joined #fluid-work
[10:02:54 EDT(-0400)] * EricDalquist (n=dalquist@bohemia.doit.wisc.edu) has joined #fluid-work
[10:52:34 EDT(-0400)] <Bosmon> More shock JS performance news!!!
[10:53:13 EDT(-0400)] <Bosmon> Access to a variable not in the same scope costs 300ns (smile)
[10:53:31 EDT(-0400)] <Bosmon> But... the good news is, primitives on "stack frames" do not count as Objects
[10:53:46 EDT(-0400)] <Bosmon> At least for memory costing purposes
[10:53:48 EDT(-0400)] * EricDalquist thinks we should all just give up on JS and write applets (wink) (ok not really but it seems like that's what JS is going to become eventually, except with less standardization)
[10:53:52 EDT(-0400)] <Bosmon> Not that that is very much good news
[10:54:05 EDT(-0400)] <Bosmon> I would love to go back to writing applets (smile)
[10:54:15 EDT(-0400)] <Bosmon> If someone can write a JVM that can start up in ms rather than s (tongue)
[10:54:53 EDT(-0400)] <EricDalquist> it seems like if browsers ran a 'native?' JVM all the time that would be solved
[10:55:04 EDT(-0400)] <Bosmon> Yeah, but it would just crash all the time
[10:55:12 EDT(-0400)] <Bosmon> Noone ever really solved the JVM partitioning issue
[10:55:17 EDT(-0400)] <EricDalquist> true
[10:55:24 EDT(-0400)] <EricDalquist> which is too bad
[10:55:26 EDT(-0400)] <Bosmon> I mean, they did, just Sun never really cared about it
[10:55:36 EDT(-0400)] <Bosmon> Really the failure of Java is just an implementation quality issue (tongue)
[10:55:42 EDT(-0400)] <EricDalquist> I'd love to have A JVM running on my machine that I can load apps in and out of
[10:55:46 EDT(-0400)] <Bosmon> Whereas the failure of Javascript is *INTRINSIC*!!
[10:55:52 EDT(-0400)] <EricDalquist> lol
[10:56:06 EDT(-0400)] <Bosmon> Well, there was once this thing called "Hydra"
[10:56:22 EDT(-0400)] <Bosmon> But the requirements on Java code to run on it are just too "unexpected" for the average developer
[10:56:39 EDT(-0400)] <Bosmon> Although they were just "good practice" to anyone else (tongue)
[10:56:45 EDT(-0400)] <Bosmon> No modifiable statics etc..
[10:57:29 EDT(-0400)] <EricDalquist> ah yeah ... statics cause lots of problems ... we have some devs here that have big memory leaks when re-deploying WARS ... due to lots of statics
[10:58:27 EDT(-0400)] <Bosmon> Since noone has any requirements on shared visibility of artefacts in a JS "VM", it is perfectly acceptable to firewall their "ClassLoaders" (that is, frames) 100% and simply ditch them into the sea when they die
[10:58:42 EDT(-0400)] <Bosmon> Unfortunately that cuts too deeply into the Java programming idiom to be made widely acceptable
[10:58:49 EDT(-0400)] <Bosmon> That is, widely acceptable to anyone without backing from Sun (tongue)
[10:59:09 EDT(-0400)] <EricDalquist> (smile)
[10:59:22 EDT(-0400)] <EricDalquist> that would be a nice feature in Java though
[10:59:26 EDT(-0400)] <Bosmon> Yes
[10:59:31 EDT(-0400)] <Bosmon> It would make it actually acceptable (tongue)
[10:59:41 EDT(-0400)] <EricDalquist> then I could just have my dream of a single JVM just always running
[11:00:00 EDT(-0400)] <EricDalquist> instead of like 8 when I'm using all my different java based tools
[11:00:11 EDT(-0400)] <Bosmon> Well
[11:00:19 EDT(-0400)] <Bosmon> It would still be too devastating when they segfaulted
[11:00:45 EDT(-0400)] <Bosmon> Anyway, this stuff was considered "cutting edge" back in 1998, and here we are 10 years later and it still has not happened
[11:00:51 EDT(-0400)] <EricDalquist> (smile)
[11:00:58 EDT(-0400)] <EricDalquist> glad I could distract you from your actual problem (wink)
[11:01:45 EDT(-0400)] <Bosmon> erm yes
[11:01:47 EDT(-0400)] <Bosmon> Likewise (tongue)
[11:13:42 EDT(-0400)] * colinclark (n=colin@142.150.154.101) has joined #fluid-work
[11:14:12 EDT(-0400)] * jhung (n=Administ@H128.C195.cci.switchworks.net) has joined #fluid-work
[11:15:05 EDT(-0400)] <jhung> anastasiac: you there?
[11:15:50 EDT(-0400)] <anastasiac> yes, hi jhung
[11:16:12 EDT(-0400)] <jhung> Question about Blockers
[11:16:19 EDT(-0400)] <anastasiac> go ahead
[11:16:48 EDT(-0400)] * colinclark thinks the only Applet worth running in a browser is Corel Office for Java.
[11:17:00 EDT(-0400)] <colinclark> The sweetest crashes I've ever seen while trying to get some work done. (tongue)
[11:17:22 EDT(-0400)] <jhung> Michelled emailed me asking about the blocker FLUID-411. The task isn't marked as a blocker in Jira, so where is it documented as such?
[11:18:23 EDT(-0400)] <anastasiac> looks like 411 is a duplicate of 381, which is a blocker
[11:19:24 EDT(-0400)] <jhung> 411 is different from 381.
[11:19:24 EDT(-0400)] <anastasiac> hm... 411 is a subtask of a ucamp planning task...
[11:19:35 EDT(-0400)] * michelled (n=team@142.150.154.197) has joined #fluid-work
[11:20:09 EDT(-0400)] <anastasiac> ah, ok, I'm looking at 411 more closely now... michelled is not sure why we considered it a blocker
[11:20:16 EDT(-0400)] <anastasiac> it might have been downgraded
[11:20:18 EDT(-0400)] <jhung> ah.
[11:20:22 EDT(-0400)] * anastasiac checks
[11:20:45 EDT(-0400)] * davidb (n=davidb@142.150.154.101) has joined #fluid-work
[11:20:54 EDT(-0400)] <davidb> where is the lightbox demo (URL) again?
[11:20:57 EDT(-0400)] <davidb> i can't find it
[11:21:01 EDT(-0400)] <davidb> anastasiac: ^
[11:21:11 EDT(-0400)] <anastasiac> davidb: build.fluidproject.org
[11:21:17 EDT(-0400)] <davidb> ta
[11:21:18 EDT(-0400)] <anastasiac> Standalone Lightbox
[11:21:58 EDT(-0400)] <anastasiac> jhung: it seems michelled was just confused (smile)
[11:22:15 EDT(-0400)] <anastasiac> so no, this is not a blocker for this release
[11:22:20 EDT(-0400)] <anastasiac> it's not even part of this release
[11:22:29 EDT(-0400)] <jhung> ah
[11:22:30 EDT(-0400)] <jhung> ok
[11:22:40 EDT(-0400)] <jhung> So 381 is a blocker that should be done.
[11:23:10 EDT(-0400)] <anastasiac> yes, that's the one I pinged you about yesterday, that you should talk to Daphne about
[11:23:10 EDT(-0400)] <jhung> Is that what michelled was thinking of?
[11:23:15 EDT(-0400)] <jhung> yes
[11:23:37 EDT(-0400)] <jhung> In progress. Not sure how long it will take. Will be talking to Daphane about it today.
[11:23:47 EDT(-0400)] <jhung> Maybe another day or 1.5 days?
[11:24:09 EDT(-0400)] <anastasiac> no problem. we're cutting the beta this week, but it's up to Daphne as to the importance of getting the pattern into the beta
[11:25:05 EDT(-0400)] <jhung> ok. I'll ask her about that.
[11:25:25 EDT(-0400)] <colinclark> jhung: To clarify, Daphne is is working on this task but could use a hand with the accessibility section for the pattern.
[11:27:39 EDT(-0400)] <jhung> yep
[11:27:53 EDT(-0400)] <jhung> Been working on that.
[11:57:23 EDT(-0400)] * aaronlev (n=chatzill@pD9E4E1D2.dip.t-dialin.net) has joined #fluid-work
[12:02:19 EDT(-0400)] * davidb (n=davidb@user147-241.wireless.utoronto.ca) has joined #fluid-work
[12:10:02 EDT(-0400)] <davidb> hi aaronlev
[12:10:08 EDT(-0400)] <aaronlev> hi
[12:10:20 EDT(-0400)] <davidb> aaronlev: welcome to the fluid room
[12:10:41 EDT(-0400)] <davidb> colinclark: aaronlev might ask some questions... feel free to ignore him and work on your workshop (smile)
[12:11:12 EDT(-0400)] <colinclark> aaronlev: Welcome. Feel free to ask away.
[12:11:36 EDT(-0400)] <aaronlev> i'm on the phone with rich – he's giving me an overview
[12:12:03 EDT(-0400)] <colinclark> Great. Say hi to Rich for me. (smile)
[12:21:13 EDT(-0400)] * davidb wonders what rich is saying about fluid
[13:13:51 EDT(-0400)] <Bosmon> A couple more performance updates:
[13:14:01 EDT(-0400)] <Bosmon> i) indexOf() is really pretty damn performant (as you might expect)
[13:14:16 EDT(-0400)] <Bosmon> Can search a 50K string in 50us
[13:14:36 EDT(-0400)] <Bosmon> There appears to be a standard "native call overhead" of about 15us
[13:15:20 EDT(-0400)] <Bosmon> So you don't actually start to notice the size of the target String until you are searching about 5K or so
[13:15:32 EDT(-0400)] <Bosmon> These costs can therefore be considered pretty much "flat rate"....
[13:39:22 EDT(-0400)] * colinclark (n=colin@user145-105.wireless.utoronto.ca) has joined #fluid-work
[13:57:06 EDT(-0400)] <Bosmon> OK
[13:57:17 EDT(-0400)] <Bosmon> So no special bonuses for "inline string concatenation"
[13:57:26 EDT(-0400)] <Bosmon> Of the form "a + a + a + a + a" etc (tongue)
[14:02:15 EDT(-0400)] <colinclark> Bosmon, You mean, as opposed to concat()?
[14:12:00 EDT(-0400)] * davidb (n=davidb@142.150.154.101) has joined #fluid-work
[14:19:04 EDT(-0400)] <Bosmon> Well, as opposed to doing it one at a time
[14:19:32 EDT(-0400)] <Bosmon> So, interesting... my first implementation of a "heap concatenator" behaves a lot worse than serial concatenation
[14:19:58 EDT(-0400)] <Bosmon> Although it is slightly faster than join()
[14:20:00 EDT(-0400)] * colinclark (n=colin@142.150.154.101) has joined #fluid-work
[14:20:04 EDT(-0400)] <Bosmon> But probably only because it has a bug in it
[14:20:50 EDT(-0400)] <Bosmon> I guess since each "+" is obliged to allocate a new object, you are essentially doomed
[14:21:41 EDT(-0400)] <Bosmon> Stupid bloodly language
[14:21:47 EDT(-0400)] <Bosmon> It actually has no performant primitives whatever
[14:22:05 EDT(-0400)] <Bosmon> Except, that is, for putting individual primitives on the "stack"
[14:22:10 EDT(-0400)] <Bosmon> Which is little bloody use to anyone
[14:52:39 EDT(-0400)] * aaronlev (n=chatzill@pD9E4E1D2.dip.t-dialin.net) has joined #fluid-work
[16:00:55 EDT(-0400)] * aaronlev (n=chatzill@pD9E4E1D2.dip.t-dialin.net) has joined #fluid-work
[16:29:17 EDT(-0400)] * michelled (n=team@142.150.154.197) has left #fluid-work
[16:54:31 EDT(-0400)] * jhung (n=Administ@H128.C195.cci.switchworks.net) has left #fluid-work
[17:23:31 EDT(-0400)] * colinclark (n=colin@142.150.154.101) has joined #fluid-work
[17:29:14 EDT(-0400)] * theclown (n=theclown@142.150.154.101) has left #fluid-work
[20:01:23 EDT(-0400)] * colinclark (n=colin@bas1-toronto09-1279475430.dsl.bell.ca) has joined #fluid-work