**** BEGIN LOGGING AT Fri Jul 22 02:59:57 2011
Jul 22 07:26:06 <ka6sox> jacques, linuxjacques you still here?
Jul 22 15:52:24 <Puffy_tp> dtzWill: ping
Jul 22 20:16:22 <dtzWill> Puffy_tp: pong
Jul 22 20:16:28 <dtzWill> also, looking through the SDL code
Jul 22 20:16:36 <dtzWill> i, uh, see why blitting via gl has always been faser
Jul 22 20:16:46 <dtzWill> they do it, badly, themselves anyway.
Jul 22 20:16:48 <dtzWill> if i understand it
Jul 22 20:16:54 <dtzWill> well "Badly" with respect to my use case anyway
Jul 22 20:17:05 <dtzWill> sooooo much overhead, phew.
Jul 22 20:17:23 <dtzWill> always bugged me that i haven't tried that again lately on newer builds (using just plain SDL for blitting a 2d surface....)
Jul 22 21:25:43 <elpollodiablo1> is there a reason why Xserver and Xterm are displayed on the app launcher even though they are unable to launch from there?
Jul 22 21:29:29 <dtzWill> elpollodiablo1: principle, we try to not install software that doesn't have an icon
Jul 22 21:29:35 <dtzWill> even kernels have launcher icons
Jul 22 21:29:46 <dtzWill> not sure i'm enitrely sure of the full reason, but i think that's the general idea of it :)
Jul 22 21:30:52 <ka6sox> dtzWill, okay thats good to know...
Jul 22 21:31:18 <elpollodiablo1> oh ok that makes sence, even though it might be a little confusing if the person doesnt know about xecutah
Jul 22 21:33:28 <elpollodiablo1> dtzWill: you ever get a chance to check out the key mappings? I want you to do a quick diff to see what I changed so you know specifically the changes and if I mixed anything up.
Jul 22 22:03:10 <ka6sox> dtzWill, anything common that would reduce the footprint of the xserver?
Jul 22 22:03:24 <ka6sox> since I want to run multiples (like 2-3 on a TP)
Jul 22 22:04:33 <dtzWill> ka6sox: on disk or in memory?
Jul 22 22:05:07 <ka6sox> memory
Jul 22 22:05:22 <ka6sox> since what I am doing is only changing which display its attached to.
Jul 22 22:06:12 <ka6sox> the veer seems to be able to do 2 of them...
Jul 22 22:06:17 <dtzWill> it's possible we cna trim down the raw xserver itself
Jul 22 22:06:18 <ka6sox> haven't tried the Pre2 yest
Jul 22 22:06:26 <ka6sox> yet
Jul 22 22:06:33 <dtzWill> and possible that somehow the libraries are being duplicated in memory, but i'd expect them to all be shared
Jul 22 22:07:06 <ka6sox> ya,
Jul 22 22:07:20 <ka6sox> the veer I'm doing 2 xserver and 2 apps
Jul 22 22:07:29 <ka6sox> haven't tried the TP yet
Jul 22 22:10:34 <dtzWill> btw my comment from before about updaterects in sdl being terrible, i'm less sure now.
Jul 22 22:10:53 <dtzWill> I thought it was going through SDL_GL_UpdateRectsLock
Jul 22 22:10:58 <dtzWill> which if it is, is terrible.
Jul 22 22:11:19 <dtzWill> but it's possible it's doing the 'right' thing and blitting straight to /dev/fb1 (when not offscreen) in which case all is good
Jul 22 22:11:53 <dtzWill> nope i'm worthless that's just for openglblit
Jul 22 22:11:58 <dtzWill> and is why openglblit is a disgrace
Jul 22 22:11:59 <dtzWill> great
Jul 22 22:12:01 <dtzWill> glad that's settled
Jul 22 22:17:05 <ka6sox> dtzWill, kewl.
Jul 22 22:20:28 <dtzWill> maybe :).
Jul 22 22:20:38 <dtzWill> trying to figure out how to make blitting 2d frames to screen as fast as i can
Jul 22 22:20:45 <dtzWill> b/c current method of uploading to texture sucks
Jul 22 22:21:48 <PuffTheMagic> dtzWill: i was just gonna say that i unbroke mendeley and 0.0.4 should  work
Jul 22 22:21:55 <dtzWill> oh, hooray! :)
Jul 22 22:21:56 <dtzWill> ty
Jul 22 22:22:12 <PuffTheMagic> i tried merging in your dl_notify changes but i need to refactor that some more
Jul 22 22:22:19 <PuffTheMagic> i started on it
Jul 22 22:22:22 <PuffTheMagic> but it was getting messy
Jul 22 22:22:27 <PuffTheMagic> so i need to think about a better way
Jul 22 22:22:31 <PuffTheMagic> for storing the data
Jul 22 22:22:42 <PuffTheMagic> really thinki need to get a db going before going further
Jul 22 22:25:32 <PuffTheMagic> did anyone email palm about the wrong TP kernel?
Jul 22 22:39:50 <dtzWill> more ramblings
Jul 22 22:40:10 <dtzWill> /dev/fb1 is the framebuffer for currently active app
Jul 22 22:40:14 <ka6sox> PuffTheMagic, yep
Jul 22 22:40:16 <ka6sox> I already did
Jul 22 22:40:34 <dtzWill> scribbling directly on it seems rather faster than going through gl/gpu?
Jul 22 22:40:38 <dtzWill> not sure if that'll work for X-ness
Jul 22 22:41:00 <ka6sox> please send me the Doctor that works on the veer (since you provided the source) or at least send me the 2.6.35 source and patches...
Jul 22 22:41:00 <dtzWill> but if we're cute and catch sdl active/hidden (or w/e they're called) events to stop rendering to it when we're not active app
Jul 22 22:41:10 <dtzWill> we might be in good shape
Jul 22 22:41:25 <dtzWill> i think /dev/fb1 is layered then put to /dev/fb0
Jul 22 22:41:29 <rwhitby> elpollodiablo1: both XServer and XTerm tell the user to use Xecutah in the documentation that appears when they click on them.
Jul 22 22:41:30 <dtzWill> so we'll still see notifications, etc
Jul 22 22:42:10 <dtzWill> especially fun is can just mmap the framebuffer to the x rendering routines
Jul 22 22:42:15 <dtzWill> still no hardware accel
Jul 22 22:42:23 <dtzWill> but this means we don't have to do all the texture upload craziness
Jul 22 22:42:29 <dtzWill> rotation would be a challenge however
Jul 22 22:42:57 <rwhitby> ka6sox: veer?  they've already released the source for the veer.
Jul 22 22:43:21 <ka6sox> rwhitby, they released the Veer kernel source and said its in 3.0.0
Jul 22 22:43:41 <ka6sox> the linuxkernel is 2.6.29 for the "3.0.0" repo
Jul 22 22:44:09 <ka6sox> so either they got it wrong (most likely) or they have a version of 3.0.0 for the veer they aren't telling us about.
Jul 22 22:44:21 <rwhitby> ka6sox: sorry, I thought you were answering PuffTheMagic's question about whether anyone had asked for TouchPad kernel source yet.
Jul 22 22:45:16 <ka6sox> http://opensource.palm.com/3.0.0/index.html
Jul 22 22:45:18 <ka6sox> wrong kernel
Jul 22 22:45:40 <rwhitby> correct, they have not released touchpad source yet.
Jul 22 22:46:12 <rwhitby> ka6sox: what did you mean by "08:05 ka6sox: please send me the Doctor that works on the veer (since you provided the source) or at least send me the 2.6.35 source and patches..." ?
Jul 22 22:46:52 <ka6sox> the kernel source in the 3.0.0 repo that I just posted is for the Veer.
Jul 22 22:47:51 <rwhitby> oh, I see now.
Jul 22 22:55:48 <PuffTheMagic> ka6sox: i did too
Jul 22 22:56:08 <ka6sox> PuffTheMagic, the more the merrier....
Jul 22 22:56:55 <PuffTheMagic> rwhitby: did you decide if you are gonna be using Mendeley screenshots?
Jul 22 23:23:30 <dtzWill> okay
Jul 22 23:23:43 <dtzWill> linuxjacques_: yt? :)
Jul 22 23:27:09 <linuxjacques_> howdy dtzWill
Jul 22 23:27:19 <linuxjacques_> I lost my webchat sometime during the meeting
Jul 22 23:33:57 <rwhitby> PuffTheMagic: unlikely at this stage
Jul 22 23:38:16 <dtzWill> mplayer benchmark down from 10.5 seconds to ~3.5
Jul 22 23:38:17 <dtzWill> haha
Jul 22 23:38:18 <dtzWill> \o/
Jul 22 23:39:42 <linuxjacques_> dtzWill: what are you up to?  :-)
Jul 22 23:42:42 <linuxjacques_> dtzWill: stop having fun and tell me what you did.
Jul 22 23:44:05 <dtzWill> linuxjacques_: 1 sec O:)
Jul 22 23:44:13 <linuxjacques_> :-)
Jul 22 23:46:28 <dtzWill> lol friends and gf are supposed to come over in a bit
Jul 22 23:46:37 <scoutcamper|lan> dtzWill, did you make mplayer work on the TP?
Jul 22 23:46:37 <dtzWill> and i'm having too much fun with X and framebuffers
Jul 22 23:46:38 <dtzWill> lol
Jul 22 23:47:00 * dtzWill dodges: scoutcamper|lan it always worked, right?
Jul 22 23:47:12 <scoutcamper|lan> ummm, no?
Jul 22 23:47:23 <dtzWill> via sdl backend, or so? heck it even works with X via x11 xD
Jul 22 23:47:25 * scoutcamper|lan hasnt tried to much
Jul 22 23:47:34 * scoutcamper|lan is lazy
Jul 22 23:47:35 <dtzWill> it's terribly slow b/c decoding is slow without hardware support
Jul 22 23:47:54 <dtzWill> but i'm not fixing that, just using it as a "blit things fast" benchmark
Jul 22 23:48:09 <linuxjacques_> and that's a great improvement
Jul 22 23:48:15 <linuxjacques_> 3x
Jul 22 23:50:41 <ka6sox> linuxjacques, the link now works for the kernel source
Jul 22 23:50:51 <ka6sox> going to get it and see what it really is
Jul 22 23:51:01 <linuxjacques_> ka6sox: wow, so they are fixing things right now
Jul 22 23:51:15 <ka6sox> linuxjacques, yes, they often do @ 10pm too.
Jul 22 23:51:19 <linuxjacques_> ka6sox: but I know where to get stock 2.6.35, I want the patch :-P
Jul 22 23:51:30 <ka6sox> patience grasshopper :D
Jul 22 23:51:41 <linuxjacques_> OK
Jul 22 23:51:44 <ka6sox> they eventually will get it fixed....
Jul 22 23:52:30 <linuxjacques_> only two hacking weekends left before I go to the 8-5 palm dev day
Jul 22 23:52:47 <dtzWill> linuxjacques_: yeah, and given decoding overhead, it could be quite a bit faster
Jul 22 23:53:05 <linuxjacques_> dtzWill: yeah, good point.
Jul 22 23:53:19 <dtzWill> some misc mpeg2, shrug, got it from the mplayer samples repo
Jul 22 23:54:16 <linuxjacques_> I wanna know what you did
Jul 22 23:55:34 <ka6sox> gonna try chromium to see if that can do ssl cert access browsing
Jul 22 23:56:56 <dtzWill> haha this is great. some details to work out an polish
Jul 22 23:57:16 <dtzWill> linuxjacques_: uhhh instead of giving kdrive a shadow buffer to render to
Jul 22 23:57:29 <dtzWill> i mmap the framebuffer and have it render straight to that
Jul 22 23:57:45 <ka6sox> okay so direct FB writes
Jul 22 23:57:48 <dtzWill> luckily kdrive's fbdev stuff handles that and misc other things like the fb ioctl stuff
Jul 22 23:57:49 <linuxjacques_> saving two(?) copies ?
Jul 22 23:57:50 <ka6sox> and alternate them.
Jul 22 23:58:09 <ka6sox> render a frame in fb0 then switch
Jul 22 23:58:14 <ka6sox> render in fb1 then switch
Jul 22 23:58:25 <dtzWill> saves having to render to shadow buffer, and then upload as texture to gpu!! and then render as texture x.x
Jul 22 23:58:31 <dtzWill> ka6sox: nope, just /dev/fb1
Jul 22 23:58:35 <linuxjacques_> nice
Jul 22 23:58:37 <ka6sox> nice
Jul 22 23:58:40 <dtzWill> /dev/fb0 is used by luna for general rendering
Jul 22 23:58:46 <ka6sox> kk.
Jul 22 23:58:46 <dtzWill> fb1 is composited into the card
Jul 22 23:58:50 <dtzWill> so vkb/etc/etc all work
Jul 22 23:59:00 <dtzWill> small issue about not drawing on /dev/fb1 when other apps are using it :3
Jul 22 23:59:02 <ka6sox> ya, we used vkb.
Jul 22 23:59:14 <dtzWill> but we can catch the SDL events indicating we aren't active app and just pause rendering
Jul 22 23:59:22 <dtzWill> (and force redraw when we're active)
Jul 22 23:59:40 <linuxjacques_> :-)
Jul 22 23:59:47 <dtzWill> spent the last forever trying to find a faster way to get a 2d surface onto the screen
Jul 22 23:59:56 <dtzWill> answer is always fb, but since we wanted card goodness i never accepted that
Jul 23 00:00:02 <dtzWill> looking at SDL stuff iwas like oooooo
Jul 23 00:00:03 <dtzWill> hahaha
Jul 23 00:00:10 <dtzWill> it's seriously faster
Jul 23 00:00:46 <dtzWill> x11perf maybe?
Jul 23 00:00:49 <dtzWill> rec's on which test? lol
Jul 23 00:00:49 <linuxjacques_> this is very exciting
Jul 23 00:01:54 <dtzWill> someone run x11perf on something on the existing x server :PP
Jul 23 00:02:09 <dtzWill> pleeeeaaseeee
Jul 23 00:02:10 <dtzWill> haha
Jul 23 00:02:10 <dtzWill> :3
Jul 23 00:02:28 <linuxjacques_> I'm heading home right now.
Jul 23 00:02:38 <dtzWill> haha
Jul 23 00:02:38 <dtzWill> xD
Jul 23 00:02:40 <linuxjacques_> where my tp is
Jul 23 00:02:44 <dtzWill> fuck people are supposed to be here in 4 minutes
Jul 23 00:02:45 <dtzWill> >><
Jul 23 00:02:48 <dtzWill> xD
Jul 23 00:02:51 <linuxjacques_> lol
Jul 23 00:02:57 <dtzWill> plenty to time to socialize after i poke at this more....
Jul 23 00:02:58 <dtzWill> xD
Jul 23 00:03:19 <linuxjacques_> biab
Jul 23 00:03:19 <dtzWill> no one has x and a chroot to quickly run x11perf?
Jul 23 00:03:20 <dtzWill> :)
Jul 23 00:03:29 <dtzWill> alright i'll recompile and perf test myself :(
Jul 23 00:04:00 <ka6sox> okay the kernelsource provided is correct
Jul 23 00:04:08 <ka6sox> lets see if they fixed the patch
Jul 23 00:04:09 <dtzWill> fwiw this should work on all devices, pretty sure
Jul 23 00:06:54 <ka6sox> linuxjacques, the patch is wrong still
Jul 23 00:07:02 <ka6sox> still the 2.6.29 one
Jul 23 00:07:22 <dtzWill> bah x11perf isn't capturing how much better this is
Jul 23 00:07:24 * dtzWill punts x11perf
Jul 23 00:07:24 <dtzWill> lol
Jul 23 00:07:32 <dtzWill> at least not on the tests i checked
Jul 23 00:07:41 <dtzWill> it's clearly muuch better though
Jul 23 00:12:50 <dtzWill> anyone use bt keyboard wiht X?
Jul 23 00:12:56 <dtzWill> does it fill the screen with the kbyaord attached?
Jul 23 00:15:30 <dtzWill> nope :(
Jul 23 00:15:35 <dtzWill> bummer
Jul 23 00:22:23 <dtzWill> linuxjacques: furthermore, depending on various other things
Jul 23 00:22:26 <dtzWill> we might be able to use a full xorg
Jul 23 00:22:46 <dtzWill> with the fbdev (or omapfb?) video driver, hacked to do the needed SDL_Init and SDL_SetVideo to get a card
Jul 23 00:23:01 <dtzWill> and then all we'd have to do is right the input driver....
Jul 23 00:23:05 <dtzWill> *write, even.
Jul 23 00:29:51 <rwhitby> dtzWill: nice!
Jul 23 00:31:37 <dtzWill> rwhitby: :D
Jul 23 00:32:06 <dtzWill> it's seriously way faster, and i think we can fix the remaining issues
Jul 23 00:32:20 <dtzWill> rotated is slower b/c we have ot draw to a shadowbuffer then do the awkward rotation memcpy
Jul 23 00:32:27 <dtzWill> might be faster with a special neon-tastic memcpy, but idk
Jul 23 00:32:47 <dtzWill> also we probably want to backup the framebuffer when we're put in card mode
Jul 23 00:32:50 <dtzWill> and restore it
Jul 23 00:32:56 <dtzWill> or rendre to shadowbuffer or w/e in the interim
Jul 23 00:33:16 <dtzWill> hmm
Jul 23 00:33:27 <dtzWill> maybe we could get away with just telling all of X t redraw itself
Jul 23 00:33:29 <dtzWill> (all clients, etc)
Jul 23 00:33:36 <dtzWill> that seems a lot easier and way less error prone
Jul 23 00:34:12 <rwhitby> yep
Jul 23 00:34:31 <rwhitby> or maybe both
Jul 23 00:34:47 <rwhitby> flash a static image, then redraw.  like OSX does for Lion resume
Jul 23 00:35:05 <rwhitby> make it look like it's instantaneous resume
Jul 23 00:35:36 <dtzWill> rwhitby: ah, yes. bet we can do that just fine :)
Jul 23 00:36:16 <linuxjacques> dtzWill, that msm driver takes a fb device as an arg
Jul 23 00:36:28 <dtzWill> wasn't there some pdk "show this as my pause screen" thing?
Jul 23 00:36:38 <dtzWill> ah
Jul 23 00:36:53 <rwhitby> dtzWill: what would be even cooler is a slower refresh active card
Jul 23 00:36:54 <dtzWill> https://developer.palm.com/content/api/reference/pdk/pdl/pdl-custom-pause-ui-enable.html or so
Jul 23 00:37:10 <dtzWill> rwhitby: sounds like an awful lot of work for ...reason taht aren't clear
Jul 23 00:37:14 <dtzWill> but certainly do-able
Jul 23 00:37:38 <rwhitby> dtzWill: for when we do the side-by-side multi-tasking patch for the TouchPad ;)
Jul 23 00:37:55 <dtzWill> ^.^
Jul 23 00:38:26 <dtzWill> alright, framebuffer branch pushed
Jul 23 00:38:32 <dtzWill> i gotta call it a night w.r.t this stuff
Jul 23 00:38:32 <dtzWill> xD
Jul 23 00:39:00 <dtzWill> combined with the updated keymap... things are good
Jul 23 00:39:11 <linuxjacques> dtzWill, have a good evening
Jul 23 00:40:17 <dtzWill> linuxjacques: ty! :)
Jul 23 00:41:25 <rwhitby> Broadcom Dongle Host Driver (DHD), common DHD core. <- in 3.0.0 patches
Jul 23 00:42:16 <halfhalo> shiny
Jul 23 00:45:04 <linuxjacques> what's that?
Jul 23 00:47:21 <rwhitby> looks like the wifi driver for touchpad
Jul 23 00:47:31 <linuxjacques> ah, cool
Jul 23 00:48:07 <halfhalo> hmm.  I could have sworn the wifi was marvell.
Jul 23 00:48:55 <ka6sox> ys, its broadcom
Jul 23 00:49:00 <ka6sox> says so in /sys
Jul 23 00:49:32 <linuxjacques> my powerd patch is gone :-(
**** ENDING LOGGING AT Sat Jul 23 02:59:57 2011