**** BEGIN LOGGING AT Mon Jul 18 02:59:57 2011 Jul 18 03:00:23 work out the script, then just put it into a C service with feedback to the user. Jul 18 03:00:36 make each step idempotent and fail-safe Jul 18 03:00:47 have each step be a service call Jul 18 03:00:55 wow, that would rock so hard. Jul 18 03:00:56 have a UI front-end to manage it Jul 18 03:01:00 ka6sox: before we get to that stage, send it to me so i can use it? :3 Jul 18 03:01:08 but everything rwhitby says is on the money :) Jul 18 03:01:23 dtzWill, sure Jul 18 03:01:34 jacques: that's the next step to wide-spread chroot deployment, since no end-users will doctor Jul 18 03:01:35 let me dig it up and put it up on the git server Jul 18 03:01:56 ka6sox: you could put it in the bootstrap repo Jul 18 03:02:19 rwhitby, okay Jul 18 03:03:25 rather than use a TP though...I'll use the pre- I have for development Jul 18 03:03:27 rwhitby, so a C service which links to libparted.so ? Jul 18 03:03:33 no Jul 18 03:03:39 its an lvm Jul 18 03:03:39 no, just does the bits of the script Jul 18 03:03:50 that would require kparted Jul 18 03:03:51 lvm resizes, vfat resizes, etc Jul 18 03:03:57 oh, lvm is beyond me Jul 18 03:04:09 jacques, thats really simple stuff Jul 18 03:04:14 I have avoided them in the past Jul 18 03:04:28 its easier than partitioning... Jul 18 03:04:46 trenchcoat does those things now when you doctor, to resize the usb partition Jul 18 03:05:16 so it sounds like the hard part is the UI / testing ? Jul 18 03:05:23 hard partS Jul 18 03:05:23 yes, Jul 18 03:05:44 the app actually will need to do a luna restart, and take it offline Jul 18 03:06:00 so that things aren't messed up. Jul 18 03:06:13 why? Jul 18 03:06:34 the cryptofs has to be offline Jul 18 03:06:48 orly? that's not going to happen then Jul 18 03:07:03 why? Jul 18 03:07:31 resizing a running partition is a BAD idea Jul 18 03:08:08 I don't see how an app is going to be able to take cryptofs offline Jul 18 03:08:29 script does it Jul 18 03:08:50 I will call the cryptofs off event and then do the resize Jul 18 03:08:59 then turn it back on. Jul 18 03:09:15 just like the usb event. Jul 18 03:11:19 good luck doing that reliably and safely Jul 18 03:14:16 I've done it like 30+times Jul 18 03:14:50 I do it exactly like the usb gadget does Jul 18 03:15:03 and if it hangs in the middle, how does the user who does not have a command line fix it? Jul 18 03:15:10 but I don't connect the usb to the flash array Jul 18 03:15:49 I purposely hung it like 4X...and recovered by rebooting. Jul 18 03:23:22 I love being able to get a chroot started with two clicks, and to just swipe it away when done. Jul 18 03:23:47 i love it not requiring my PC to run :D Jul 18 03:28:19 two click chroot? sweet. Jul 18 03:28:33 I love being able to get to xapps directly :D Jul 18 03:29:22 * halfhalo misses having a chroot. he needs to rectify that when he gets to a pc Jul 18 03:36:34 I have testing feed enabled, why don;t I see ubuntu chroot? Jul 18 03:37:14 jacques, i see it on my TP Jul 18 03:37:30 that's nice :-\ Jul 18 03:37:52 jacques, what device r u on? Jul 18 03:38:11 tp Jul 18 03:38:20 idk thn Jul 18 03:38:25 u see xecutah? Jul 18 03:38:30 device doesn't matter Jul 18 03:38:41 yeah I have xecutah/xserver/xterm Jul 18 03:39:01 jacques, it should be there then Jul 18 03:39:03 rwhitby, something I am missing? Jul 18 03:39:10 nope Jul 18 03:39:27 I see things updated today, Mendeley, Navit Maps, etc. Jul 18 03:39:31 http://ipkg.preware.org/feeds/webos-internals/testing/armv7/org.webosinternals.ubuntu-natty-chroot_11.04-2_arm.ipk Jul 18 03:39:45 you have the armv7 testing feed enabled? Jul 18 03:39:52 hmm Jul 18 03:40:09 I don't see that feed Jul 18 03:40:20 oh, wasit yes Jul 18 03:40:26 I have it Jul 18 03:40:45 ipkg.preware.org/feeds/webos-internals/testing/armv7 Jul 18 03:41:02 so it's a mystery then? Jul 18 03:41:18 that's a bit concerning Jul 18 03:41:38 is your feed name unique? Jul 18 03:42:08 rwhitby, there are two URIs under the heading webos-internals-testing Jul 18 03:42:26 as there are under webos-internals and optware-testing Jul 18 03:42:50 if what I said doesn't answer that question, I don't understand the question Jul 18 03:43:26 jacques: dunno - next step is to check the list files themselves Jul 18 03:43:38 pastebin your conf file first Jul 18 03:46:25 I can't find it Jul 18 03:51:06 its in /media/cryptofs/apps/etc/ipkg/ Jul 18 03:54:02 rwhitby, why no webos user in the production chroot? Jul 18 03:55:28 rwhitby, http://pastebin.com/3PDk7QBc Jul 18 04:00:44 ka6sox: no way to set a password on package install. Jul 18 04:01:09 not going to encode a security vulnerability into the page Jul 18 04:01:10 package Jul 18 04:01:23 jacques: looks correct Jul 18 04:01:26 rwhitby, you see anything strange in my config file above? only thing I see is ubuntu chroot is arm not arm7 Jul 18 04:01:39 frm looking at http://ipkg.preware.org/feeds/webos-internals/testing/armv7/Packages Jul 18 04:01:39 jacques: any other conf files with the same feed names (second field after src/gz) ? Jul 18 04:01:42 rwhitby, how is a non-priv user in a chroot going to do that? Jul 18 04:02:09 rwhitby, no, grep shows that as only one Jul 18 04:02:26 ka6sox: they install ssh, and suddenly the device has a know webos/webos exploit Jul 18 04:03:06 how is that any diff than xterm Jul 18 04:03:32 ka6sox: xterm can only be controlled by the user, not a remote network attack Jul 18 04:03:52 so it's physical security vs network Jul 18 04:04:08 who installs ssh? Jul 18 04:04:22 and where? Jul 18 04:04:23 (and yes, we need to deal with the "give the tablet to your mate" level of physical security) Jul 18 04:04:55 your non-priv users does "sudo install ssh" in the chroot. oops, they just gave out remote root. Jul 18 04:05:07 sudo apt-get install ssh Jul 18 04:05:34 my non priv user has no access as I give it a sudoers file and place that in the chroot when I make it. Jul 18 04:06:06 and I bet you set a password that I don't know Jul 18 04:06:16 no password Jul 18 04:06:24 none at all Jul 18 04:06:35 so how do you install a package when you're in the chroot? Jul 18 04:06:39 just a non-priv user that isn't part of sudoers Jul 18 04:06:44 xterm can do that Jul 18 04:06:50 since its priv Jul 18 04:07:09 ok, I must not understand what you are asking for here, Jul 18 04:07:32 currently you get a user-interactive chrooted xterm, in which you can say "adduser" to create a new user and then use that. Jul 18 04:07:57 we do not auto-create such a user in the tarball, cause then that auto-created user would have a known password. Jul 18 04:08:15 what is it that you are requesting be different? Jul 18 04:09:10 just have the chroot come with a non priv user that doesnt' have access to sudoers Jul 18 04:09:30 if not I will just add it to my installer... Jul 18 04:15:13 ka6sox: what use is that non-priv user with no sudo access to the person who installs this chroot? they can't do anything with it out of the box Jul 18 04:15:35 the chroot has no apps, so they can't run anything Jul 18 04:15:44 the user has no sudo, so they can't install anything Jul 18 04:16:31 what is the work-flow for the end-user who does not know what 'adduser' is? (since anyone who does know what adduser is can just use that) Jul 18 04:16:42 nm...I'll just handle it a different way. Jul 18 04:16:58 its just 1 more step to do. Jul 18 04:17:11 ok, just checking whether this was a one-off for your particular situation, or something generally useful for the Preware package. Jul 18 04:17:32 cause I don't see a way to have a non-priv user without a known password in the Preware package Jul 18 04:18:25 the chroot tarball has it...not for the base os. Jul 18 04:18:31 only the chroot tarball. Jul 18 04:35:56 ka6sox: apologies, I misunderstood your request Jul 18 04:36:34 I now realise you're asking for a non-priv user *without* a password, which you can have root sudo to to run stuff. Jul 18 04:36:36 xapps don't like being run as root most of the time. Jul 18 04:36:46 no problem at all with that. Jul 18 04:36:48 and its dangerous. Jul 18 04:37:17 how do you suggest we add that user? can rootstrap do it without adding a password, or do we have the Makefile hack it in later? Jul 18 04:37:59 since you can't chroot on the source device its a postinst Jul 18 04:38:13 chroot useradd foo Jul 18 04:39:32 ok. do you want to test such a change? Jul 18 04:39:38 * rwhitby needs to get lunch Jul 18 04:39:44 already done Jul 18 04:40:10 oh, you should have just pastebin'd me a patch then and I might have understood quicker ;) Jul 18 04:40:28 kk Jul 18 04:40:39 * rwhitby bbiab Jul 18 05:00:56 rwhitby, I used the browser function to install 2 optware apps, and until I restarted preware they dind Jul 18 05:01:00 didn't show up. Jul 18 05:02:16 I used "Install Package" and Browse, found the packages in /media/internal and installed. Jul 18 05:02:40 correct Jul 18 05:02:48 but they didn't show up in the installed list till after I rebooted preware. Jul 18 05:03:08 is that correct? Jul 18 05:03:13 yep, until you update feeds Jul 18 05:03:17 doesnt' do that for normal apps. Jul 18 05:03:23 oh Jul 18 05:03:26 based upon feeds. Jul 18 05:03:33 hmmm Jul 18 05:09:32 it's possible that Preware could take the information extracted from the loaded package and put it into the installed packages database. But in general stuff related to Install Package is quite separate from the feeds handling, so may not be easy. Jul 18 05:11:49 okay thats understandable. Jul 18 05:12:02 its not like its never showing up Jul 18 05:12:16 and so can't be removed without getting ugly Jul 18 05:45:48 okay I don't understand why the postinstall is complaining about /etc when the installer puts termcap in there. Jul 18 05:46:24 /opt/etc Jul 18 06:12:44 cross-compile/packages/media/libvorbis is empty for me - no Makefile Jul 18 06:13:16 oh, I see apparently it was moved to nonworking Jul 18 06:16:14 feel free to move it back if you fix it Jul 18 06:38:12 rwhitby, libvorbis appears to have the same issue as sdl-gfx, and builds when I do http://www.webos-internals.org/wiki/Widk#X--tag.3DCC:_command_not_found Jul 18 06:38:54 "export echo=echo" Jul 18 06:39:19 jacques: we should put that in the Makefile then Jul 18 06:40:33 it seems like a hacky fix, though I admit I don't understand what's going on there Jul 18 06:46:43 if it works it works Jul 18 06:55:40 I guess "a hacky fix to libtool" is redundant anyway Jul 18 06:55:54 lol Jul 18 06:58:16 cryptofs is mounted in /media/internal by /etc/init.d/mountall.sh Jul 18 06:58:31 then turned on and off by the 2 storage.d scripts Jul 18 06:59:09 so to resize /media/internal you at least need to stop cryptofs temporarily Jul 18 07:01:20 then umount /media/internal and resize that...then remount and restart cryptofs. Jul 18 07:01:49 at that point everything else is not "dangerous" Jul 18 07:11:06 pkill -SIGUSR1 cryptofs Jul 18 07:11:29 pkill -SIGUSR2 cryptofs Jul 18 07:21:22 so the script that does that must not be in cryptofs Jul 18 07:21:32 correct Jul 18 07:23:09 ok, so we'd need to put that in /var/usr/sbin/ or something. Jul 18 07:23:31 * rwhitby creates a repo for Resizah Jul 18 07:23:41 already ther.. Jul 18 07:23:43 just rename Jul 18 07:24:18 Resizah? Jul 18 07:24:19 really? Jul 18 07:24:22 :p Jul 18 07:24:25 can it be done in a service method so that it can report status back to the front-end UI? Jul 18 07:24:39 Xuzz: it's a running joke, just to annoy oil and destinal Jul 18 07:24:54 and me too, apparently Jul 18 07:25:02 too many -ah names Jul 18 07:25:18 rwhitby, it can log and the service pick that up. Jul 18 07:25:28 call them from the service Jul 18 07:25:38 but I am opposed to services in general... Jul 18 07:25:47 they stay around Forevah Jul 18 07:26:04 even after they should be LONG gone. Jul 18 07:26:13 ka6sox: I'm open to other ideas, but something like this needs to give progress feedback to the user Jul 18 07:26:22 it can't just go dark for minutes while it is resizing Jul 18 07:26:44 you need continuous warm fuzzies for a tool like this Jul 18 07:26:56 Xuzz, oil: open to other ideas on naming :) Jul 18 07:27:09 rwhitby: I'm assuming this is to resize partitions? Jul 18 07:27:21 LVM partitions resizing and creation Jul 18 07:27:23 any moreso than with with the device going blank and it being mountable on a host? Jul 18 07:27:49 ka6sox: yes, cause in that case you're not resizing partitions Jul 18 07:29:05 rwhitby, unless you can tell me how the app will continue running with cryptofs suspended then no. Jul 18 07:29:42 ka6sox: I assume JS is cached when read, so we stay on the same scene and report updates there Jul 18 07:30:11 okay so that means this can't be dependent on another app in cryptofs Jul 18 07:30:20 such things need to be tested. but a tool that goes into the feeds for this needs to have a front-end with frequent progress reporting Jul 18 07:30:23 it has to be totallhy standalone Jul 18 07:31:14 but unless we can continue running an app with cryptofs suspended then I'm not sure. Jul 18 07:31:19 this app is so dangerous that you don't want it depending on anything else Jul 18 07:31:25 right Jul 18 07:31:47 have you tested that you can't have a Mojo scene still updating from a service with cryptofs suspended? Jul 18 07:31:54 I would expect it to work Jul 18 07:32:02 rwhitby: Resizah.. nice :P Jul 18 07:32:14 palm blanks the screen and doesn't let you do anything Jul 18 07:32:23 doesn't mean we can't have our app instead keep reporting Jul 18 07:32:56 palm however suspends input during that time. Jul 18 07:33:00 Xuzz, oil: see? destinal loves the name. Jul 18 07:33:20 maybe if i stop saying how much i hate -ah it will go away Jul 18 07:33:51 they just throw up a png file Jul 18 07:33:59 and lock input Jul 18 07:34:17 ours could just as easily keep reporing Jul 18 07:34:20 reporting Jul 18 07:35:09 ka6sox: do you have a prototype we can look at? Jul 18 07:35:39 I am rewriting it now...its been almost 2yrs and I know a lot more now about how it works. Jul 18 07:35:51 I'll push to the repo as soon as I have a working script. Jul 18 07:36:00 (without reporting for now) Jul 18 07:39:11 that means any preware installed things like any optware things might not work. Jul 18 07:39:29 they would have to be standalone in /opt/* Jul 18 07:39:48 so fatresize from optware, for example, would need to be installed via cli Jul 18 07:40:07 this tool needs to be standalone. Jul 18 07:40:20 no dependencies Jul 18 07:40:20 we can provide the tools Jul 18 07:40:36 touchpad has resizefat on it Jul 18 07:40:50 for others then we need it. Jul 18 07:41:02 maybe veer already has it. Jul 18 07:43:06 so you want to put this in /var/lib/palm? Jul 18 07:43:28 what is /var/lib/palm? Jul 18 07:43:32 nm Jul 18 07:43:39 /var/usr/sbing Jul 18 07:43:42 -g Jul 18 07:44:38 the whole app including the js stuff lives there? Jul 18 07:44:39 it should be /var/usr/sbin/org.webointernals.resizah Jul 18 07:44:47 okay Jul 18 07:44:56 no, the app should be in the normal place Jul 18 07:45:23 (assuming your tests of whether JS is cached while showing a single scene are successful) Jul 18 07:45:28 I'm not sure that you wouldn't hang it. Jul 18 07:45:57 I'm no js expert...so making that bit is beyond me. Jul 18 07:46:02 app can only be executed from cryptofs, you don't have a choice Jul 18 07:46:59 okay wish me luck...I'm about to suspend cyptofs and see if the screen goes blank. (it shouldn't) Jul 18 07:47:47 we should be able to test this using the Hello! app. Jul 18 07:48:08 just have it call the hello service once every second counting up numbers for the name. Jul 18 07:48:37 while it's doing that, suspend cryptofs Jul 18 07:49:21 the service C binary should be /var/usr/sbin/org.webointernals.resizah Jul 18 07:49:29 can't use node.js service, cause that's in cryptofs Jul 18 07:49:44 I was going to put time's exhibition mode up. Jul 18 07:49:55 but thats okay to use hello! too. Jul 18 07:50:01 we need to work out how to do dynamic C services that only run when called Jul 18 07:50:17 yes, because its a MESS. Jul 18 07:50:32 it's a much tidier mess than the jail mounts Jul 18 07:50:53 yep. how to leave messes around that need to be cleaned up. Jul 18 07:51:13 and C services are a mess cause they are the only way to do what we need to do Jul 18 07:51:27 I'm all ears for a better alternative Jul 18 07:52:47 isn't hello! in testing Jul 18 07:56:42 how do you put an app in exhibition? Jul 18 07:59:15 add code to do it Jul 18 07:59:34 okay the countdown timer ran without any issues Jul 18 07:59:38 I can't find hello Jul 18 08:00:52 is there an underlying c service that does something? Jul 18 08:02:56 impostah uses a c service right? Jul 18 08:03:41 hello is in git Jul 18 08:04:01 impostah has a C service Jul 18 08:04:19 okay, changing scenes is a no-go as expected Jul 18 08:06:01 but its not in preware-build right? Jul 18 08:10:34 correct, hello is not released Jul 18 08:15:28 okay, main scene in govnah is updating Jul 18 08:15:39 with cryptofs disabled Jul 18 08:16:16 does switching scenes hang? Jul 18 08:16:57 yes Jul 18 08:17:17 what about going into card view and back into the app again? Jul 18 08:18:00 that still works Jul 18 08:19:03 but I broke govnah by trying to go into preferences...let me restart it Jul 18 08:19:24 ya, that fixed that Jul 18 08:19:39 so we should make this a 1 scene app Jul 18 08:20:03 oil, is there a way to do a progress bar in js/mojo/enyo? Jul 18 08:20:22 yes Jul 18 08:20:26 and yes Jul 18 08:21:15 then we do like trenchcoat does...and make milestone entries Jul 18 08:21:25 when it does this the bar moves to x Jul 18 08:21:53 because we wont' get the level that says 10%, 11% etc from the resizes Jul 18 08:21:56 I would prefer to give extremely detailed status messages about what is happening Jul 18 08:22:08 thats not hard. Jul 18 08:22:13 including percentage progress through the resizes Jul 18 08:22:21 just like preware does with downloads Jul 18 08:22:24 the resizes dont' do that. Jul 18 08:22:37 the underlying apps don't Jul 18 08:22:46 resizefat doesn't Jul 18 08:22:53 lvm resize doesn't Jul 18 08:23:21 lvcreate doesn't Jul 18 08:23:33 mke2fs -j does Jul 18 08:23:44 do you have a wiki page with all these commands on it? Jul 18 08:24:26 this was never a public thing..and because it was dangerous to tell folks to do this I never put one up. Jul 18 08:24:51 put a readme in the git repo and I'll mod that and up load it. Jul 18 08:26:29 is resizah in which section? Jul 18 08:30:03 there is no resizah repo yet - I was waiting for Xuzz and oil to come up with a better name ;) Jul 18 08:30:22 okay well there is resizer and I'm working in there. Jul 18 08:31:00 applications/resizah.git Jul 18 08:31:14 yup..thats where I'm working Jul 18 08:31:40 Presizer Jul 18 08:31:57 does the Veer have fatresize already? Jul 18 08:38:13 tailor? Jul 18 08:46:34 rwhitby, pushed Jul 18 08:47:27 ka6sox: you sure? Jul 18 08:47:36 http://git.webos-internals.org/applications/resizah/ Jul 18 08:47:38 me sure about what? Jul 18 08:47:43 no, not resizah Jul 18 08:47:59 I was working on the old name. Jul 18 08:48:21 * oil votes tailor Jul 18 08:48:47 do I get a vote for my own app? Jul 18 08:49:34 git clone git@git.webos-internals.org:applications/resizah.git Jul 18 08:49:36 well, if its your app, its not really voting Jul 18 08:49:46 you just decide :) Jul 18 08:49:49 fatal: no matching remote head Jul 18 08:50:25 try now Jul 18 08:50:50 * oil just thinks anything without *ah is better then anything with *ah Jul 18 08:51:08 "Withoutah" - excellent suggestion Jul 18 08:51:33 totally missed my actual suggestion :/ Jul 18 08:51:45 second choice was Trimmah Jul 18 08:51:49 I didn't Jul 18 08:52:29 ka6sox: what name have you chosen for your app? Jul 18 08:53:21 still trying to make it work...but I like tailor as it goes with tinker. Jul 18 08:54:39 done Jul 18 08:54:50 oil +1 Jul 18 08:56:05 * rwhitby bbiab (bus) Jul 18 08:56:15 ka6sox: let me know when you've pushed Tailor Jul 18 08:56:21 will do Jul 18 08:58:00 readme pushed Jul 18 08:58:27 yay, no *ah :) Jul 18 08:59:05 * ka6sox renames Impostah to Spy Jul 18 08:59:14 now I only need a Soldier :D Jul 18 08:59:17 wow, a much better name Jul 18 09:01:21 rwhitby, production units don't have fatresize Jul 18 09:03:07 aww no fatresize :( Jul 18 09:03:22 hooray ipkg-opt to the rescue Jul 18 09:03:40 dtzWill, ya, I was able to get that one in to optware (with a LOT of help) Jul 18 09:03:55 that was a result of our experimenting Jul 18 09:04:56 tempted to give the workflow here: http://git.webos-internals.org/applications/tailor/tree/README a shot. Jul 18 09:05:09 it's what i'd expect to do--with goodness around making sure cryptofs isn't running Jul 18 09:05:21 any reason i couldn't go ahead and do this myself atm? :) Jul 18 09:05:30 i already did that ...but please add to it Jul 18 09:05:39 (also if i do, should i reinstall optware or something to get it using the new partition?) Jul 18 09:05:52 not helpful in this case. Jul 18 09:06:10 as the reason for doing this is to make the new partition. Jul 18 09:06:36 errr i thought optware bootstrap would use this partition if it existed Jul 18 09:06:42 allieviating the small /opt issue Jul 18 09:06:52 i think i misunderstood somewhere, sorry. Jul 18 09:06:55 yes, it does Jul 18 09:06:57 sorry Jul 18 09:07:07 otherwise it uses cryptofs. Jul 18 09:07:23 and a loop Jul 18 09:07:32 does fatresize really take a mountpoint? Jul 18 09:07:38 small nitpick if not Jul 18 09:08:18 its probably a partition...but its been 2yrs... Jul 18 09:08:25 I have to install it again. Jul 18 09:08:28 okay well even afer nuking cryptofs Jul 18 09:08:43 /dev/mapper/store-media is mounted everytwhere Jul 18 09:08:46 for jails mostly Jul 18 09:08:54 /dev/mapper/store-media on /var/palm/jail/com.artofbytes.grave-defense-hd/media/internal type vfat (rw,relatime,fmask=0000,dmask=0000,allow_utime=0022,codepage=cp437,iocharset=iso8859-1,shortname=mixed,utf8,errors=remount-ro) Jul 18 09:09:01 so we'll need to clean those up Jul 18 09:09:17 lsof is your friend. Jul 18 09:09:17 "jailer -C" does that trick nicelyu Jul 18 09:09:31 dtzWill, add that :D Jul 18 09:10:43 well i did something wrong ^.^ Jul 18 09:11:06 that pkill line should unmount cryptofs i'd imagine, but i was silly and didn't check...it was still mounted Jul 18 09:11:14 rebooting to poke at that some. Jul 18 09:11:25 it's possible that the jail thing will be a non-issue if cryptofs unmounts itself properly Jul 18 09:11:28 that pkill line disconnects the cryptofs Jul 18 09:11:41 jas. Jul 18 09:12:40 the pkill lets you umount /media/internal Jul 18 09:12:49 * dtzWill creates jails Jul 18 09:12:49 ka6sox: can these things be run from an xterm on-device? Jul 18 09:13:08 probably. Jul 18 09:13:12 but haven't tried Jul 18 09:13:30 ka6sox: cryptofs is still listed as mounted after that pkill command Jul 18 09:13:30 ? Jul 18 09:13:49 dtzWill: that is correct Jul 18 09:13:53 yup Jul 18 09:13:54 suspended, not unmounted Jul 18 09:14:01 ahhh, nifty. okay. Jul 18 09:14:08 sorry, didn't realize it was cool like that :) Jul 18 09:14:23 the umount /media/internal will fail if its running Jul 18 09:14:58 dtzWill: it's a fuse app which drops all it's open filesystem handles when suspended and opens then again when resumed Jul 18 09:15:04 without the calling app knowing Jul 18 09:15:31 hmm fatresize (trying to do a fatresize -i /dev/mapper/store-media) is giving "input/output error" :( Jul 18 09:15:39 it worked *Before*, lame Jul 18 09:15:53 ipkg-opt: resource temporarily unavailable Jul 18 09:15:54 the hell Jul 18 09:16:18 ipkg-opt runs from cryptofs ususlly Jul 18 09:16:20 remember...when cryptofs is broken you can't call up things Jul 18 09:16:27 right Jul 18 09:16:33 oh didn't know from cryptofs Jul 18 09:16:34 both cases thats true Jul 18 09:16:38 yup Jul 18 09:16:50 hrm, so i can't really use the fatresize from optware Jul 18 09:16:55 its worse with loop in cryptofs Jul 18 09:17:02 dtzWill, ypu Jul 18 09:17:04 at least not without some magic, right? Jul 18 09:17:06 okay. Jul 18 09:17:09 correct. I thought I said before that this tool can have no dependencies Jul 18 09:17:18 it has to bring them. Jul 18 09:17:26 and put them in the right places Jul 18 09:18:05 dtzWill: what device are you doing this on? Jul 18 09:18:20 rwhitby: poking at my TP atm Jul 18 09:18:42 oh /opt is loop mount from cryptofs somewhere? interesting Jul 18 09:18:48 shows what i know :) Jul 18 09:18:48 yup Jul 18 09:19:01 with no ext3fs partition thats what it does Jul 18 09:19:07 dtzWill: resizefat is native Jul 18 09:19:15 ah, haha Jul 18 09:19:16 no need for optware stuff Jul 18 09:19:16 great Jul 18 09:19:36 how about on a Veer? Jul 18 09:20:33 doctor has it Jul 18 09:20:41 in the rootfs, so yes. Jul 18 09:20:57 wonder if its in all 2.x Jul 18 09:20:58 as does Pre 2 Jul 18 09:21:35 as does Pre+ 2.1.0 Jul 18 09:21:54 * dtzWill crosses his fingers Jul 18 09:21:55 ^.^ Jul 18 09:22:35 its going to take some 5+minutes for that to happen. Jul 18 09:22:54 * dtzWill nods, but is thankful to have estimate so he doesn't sweat too much Jul 18 09:22:55 :) Jul 18 09:23:10 giving ext3 8G sound good? :) Jul 18 09:23:14 yes Jul 18 09:23:15 what affects that time? what is the min and max? Jul 18 09:23:32 rwhitby, the new version can be verbose Jul 18 09:23:46 so the calling app can listen to it. Jul 18 09:23:53 new version of what? Jul 18 09:23:59 resizefat Jul 18 09:24:10 -v Verbose progress report. Jul 18 09:24:12 * dtzWill wishes he'd run it with the -v flag Jul 18 09:24:14 there's an old version of resizefat? Jul 18 09:24:25 the one in optware Jul 18 09:24:38 that's not resizefar Jul 18 09:24:38 that's fatresize, apparently different app? Jul 18 09:24:42 ypu Jul 18 09:24:50 ok Jul 18 09:24:51 (fwiw resize to 24G just completed) Jul 18 09:24:55 okay. well...we can get verbose reports. Jul 18 09:25:13 dtzWill: wonder if it depends on how full and fragmented the fs is Jul 18 09:25:35 rwhitby: I can't imagine it doesn't Jul 18 09:25:39 the pre- I did a 2G on was SLOW Jul 18 09:25:56 since it's trying to resize it, it needs to guarantee that last 8G is unused Jul 18 09:26:10 right Jul 18 09:26:17 go go lvreduce Jul 18 09:26:31 add another 100M or so to the size of the LV Jul 18 09:26:40 does lvreduce take a long time? Jul 18 09:26:46 oh, no it's instant. Jul 18 09:26:47 haha Jul 18 09:26:48 not usually Jul 18 09:26:48 :3 Jul 18 09:26:55 or is resizefat the only fear-inducing long time step? Jul 18 09:27:05 hooray my /media/intenral wasn't clobbered Jul 18 09:27:06 thats teh only one Jul 18 09:27:19 rwhitby: resizefat is the only long step it seems Jul 18 09:27:25 ok, so we have sufficent progress then Jul 18 09:27:31 once cryptofs is remounted we are good. Jul 18 09:27:33 so SIGUSR1 has..cryptofs...toggle it's active state? Jul 18 09:27:39 no Jul 18 09:27:45 SIGUSR2 restores Jul 18 09:27:48 USR2 Jul 18 09:27:50 so the README there probably means USR2 at the end then Jul 18 09:27:54 yes Jul 18 09:28:09 and lvreduce -L24G /dev/mapper/store-emdia Jul 18 09:28:35 not 24G for 8G Jul 18 09:28:38 (fully understand these are semi-obvious notes, but just being explicitly clear about things/corrections) Jul 18 09:28:45 rwhitby: oh? Jul 18 09:28:55 19G Jul 18 09:28:57 i wanted to see size of the original partition, but wasn't unable to do so effectively Jul 18 09:28:57 ah Jul 18 09:29:09 well time to reduce the partitoin again Jul 18 09:29:10 ^.^ Jul 18 09:29:13 round 2.... fight! Jul 18 09:29:16 there are other partitions Jul 18 09:29:18 put -v Jul 18 09:29:25 rwhitby: indeed, i wasn't thinking O:) Jul 18 09:29:41 (excellent way to be while resizing partitions...) Jul 18 09:29:58 heh :) Jul 18 09:30:22 * dtzWill does resizefat to 19G Jul 18 09:30:31 so far -v isn't helpful Jul 18 09:30:31 lol Jul 18 09:30:36 add -v and see how much time it adds Jul 18 09:30:39 might be just doing the fsck atm Jul 18 09:30:44 nvm Jul 18 09:30:48 6 percent complete. Jul 18 09:31:08 prints new line periodically (can't see pattern, prints same 'percent' multiple times, no noticeable tick) Jul 18 09:31:15 so just prints lines like "44 percent complete." Jul 18 09:31:23 i wouldn't imagine -v adds any overhead, tbh. Jul 18 09:31:32 probably not Jul 18 09:31:40 makes rwhitby have warm fuzzies Jul 18 09:31:42 doesn't print information often enough for i/o to be a bottleneck, and computing percent done seems trivial compared to....fs management Jul 18 09:31:55 100 percent complete printed 2x Jul 18 09:31:55 xD Jul 18 09:32:23 so we add echo's to the various steps and have the app listen to the stream Jul 18 09:32:50 well it isn't done yet Jul 18 09:32:54 but says 100 percent complete Jul 18 09:32:56 idk what's going on Jul 18 09:33:05 second FSCK Jul 18 09:33:14 ah Jul 18 09:33:19 could be more verbose aboutt hat -.- Jul 18 09:33:38 done. Jul 18 09:33:55 can you capture all that and post it on a pastebin? Jul 18 09:35:19 the vg is called "store" Jul 18 09:35:36 first thing I'd do is pvscan Jul 18 09:35:46 and find out if there really is 8G free Jul 18 09:35:58 http://pastebin.com/94v8Pnap Jul 18 09:36:19 pvscan says 8.64GB free Jul 18 09:36:36 errrr Jul 18 09:36:47 err? Jul 18 09:36:56 is media internals hozed? Jul 18 09:37:03 because I suspect it is. Jul 18 09:37:06 it's not. Jul 18 09:37:09 kk Jul 18 09:37:13 at least it doen't seem to be Jul 18 09:37:18 ls has my files anyway. Jul 18 09:37:22 not the most extensive fs test ;) Jul 18 09:37:26 errr was b/c of Jul 18 09:37:31 http://pastebin.com/N7tUr6Kj Jul 18 09:37:35 i'm not sure what that means Jul 18 09:37:40 did you restart the cryptofs? Jul 18 09:37:46 ka6sox: i haven't yet, no. Jul 18 09:38:03 * dtzWill does so Jul 18 09:38:15 cryptofs up, and it seems fine Jul 18 09:38:20 before you try to run things that should be reconnected. Jul 18 09:38:24 * dtzWill votes it works, if a cryptographic fs can be read Jul 18 09:38:34 mkay so time to make ext3 goodness Jul 18 09:38:38 okay... Jul 18 09:38:42 ka6sox: no? Jul 18 09:39:08 games running off cryptofs... Jul 18 09:39:26 lvcreate --L 8G --name ext3fs store Jul 18 09:39:41 indeed, ty. Jul 18 09:39:57 with the script you will have to use a force...which is ugly but works. Jul 18 09:39:57 err -L anyway. Jul 18 09:40:09 ...force? Jul 18 09:40:26 for the vgreduce Jul 18 09:40:33 sorry Jul 18 09:40:35 lvreduce Jul 18 09:40:47 oh, i see. to not have it prompt, yes. Jul 18 09:40:52 go go mke2fs Jul 18 09:41:55 /etc/fstab entry: /dev/mapper/store-ext3fs /media/ext3fs ext3 rw,noatime 0 0 Jul 18 09:42:08 mmhmm, /me does so Jul 18 09:42:52 creating the mout point might be nice Jul 18 09:43:02 ya, Jul 18 09:43:08 okay I add to workflow Jul 18 09:43:26 looks good Jul 18 09:43:39 any advice on nuking old /opt and setting up new /opt? Jul 18 09:43:47 not sure why icare, but seems like the thing to do Jul 18 09:43:48 ya Jul 18 09:43:50 and giving optware 8G seems nice Jul 18 09:43:55 well. Jul 18 09:44:12 remove all of preware-optware stuff Jul 18 09:44:21 then reset it up... Jul 18 09:44:38 (mine didn't so I had to hack it up to make it work) Jul 18 09:44:56 trenchcoat forgot to add the entry to fstab Jul 18 09:45:01 so it didn't mount. Jul 18 09:45:35 aww preware won't remove dep's for me Jul 18 09:45:36 hehe Jul 18 09:45:46 instead...it left / rw and wrote the files to /opt as mounted on the realroot Jul 18 09:46:09 so I filled up my / partition Jul 18 09:47:15 okay I'm writing up the 2 scripts Jul 18 09:47:21 hmm removed optware and such, rebooting Jul 18 09:47:27 got immediately silenced errors in the removal Jul 18 09:47:31 not sure it worked, but shrug, idk. Jul 18 09:47:39 * dtzWill reboots before installing bootstrap again Jul 18 09:48:03 it got mad because there are other things you installed with the CLI Jul 18 09:48:09 so the dirs aren't empty Jul 18 09:48:29 and the prerm's aren't happy. Jul 18 09:51:45 well Jul 18 09:51:49 /dev/mapper/store-ext3fs on /opt type ext3 (rw,noatime,errors=continue,barrier=0,data=writeback) Jul 18 09:51:52 seems god Jul 18 09:51:53 *good Jul 18 09:53:52 * dtzWill dl's ubuntu chroot Jul 18 09:53:54 ^.^ Jul 18 09:53:58 alright well that was a win Jul 18 09:54:05 gotta run for the night, enjoy all :) Jul 18 09:54:16 dtzWill, just get the chroot app that works with xecutah Jul 18 09:54:19 thanks for the help, and i'll be happy to help robustify and automate this moving forward Jul 18 09:54:25 ka6sox: that's what i'm doing :D Jul 18 09:54:49 okay, the basic scripts are done...now to start all the checking stuffs to make it safer. Jul 18 09:55:38 if I had 5 minutes more to think of a name I would have called either muck or mucker Jul 18 09:56:28 dtzwill you missed that... Jul 18 09:56:46 it should be /ext3fs/opt Jul 18 10:02:27 I could see us wanting to use this in a re-entrant fashion so I'm making the second script able to know that there is a ext3fs partition and resize that one if necessary. Jul 18 10:25:18 yeah, the scripts have to be absolutely fail-safe and fool-proof Jul 18 10:27:02 rwhitby, I'm thinking about having 3 scripts Jul 18 10:27:09 1 that resizes the fat Jul 18 10:27:19 1 that creates the ext3fs Jul 18 10:27:30 and a 3rd that would resize the ext3fs Jul 18 10:28:24 resizing the ext3fs is a bit more involved. Jul 18 10:37:36 you'll also need scripts which return information to the front-end about the current status and sizes Jul 18 10:37:51 ya, I've added shells for those Jul 18 10:39:23 does not appear as if there is an fsck for vfat present. Jul 18 10:50:34 wish we had fsck.vfat Jul 18 10:50:51 I guess we can bring that to the party Jul 18 10:51:39 what do Palm's scripts do? Jul 18 10:53:42 for? Jul 18 10:53:59 I don't know what trenchcoat does. Jul 18 10:54:03 its a binaray Jul 18 10:54:11 no, mountall.sh Jul 18 10:54:21 it does an fsck on /media/internal Jul 18 10:54:33 fsck.msdos Jul 18 10:54:40 and fsck.vfat too Jul 18 10:54:52 both in /usr/sbin Jul 18 10:55:05 the rest are in /usr/bin Jul 18 10:55:11 yep Jul 18 10:55:17 okay thats fine. Jul 18 10:55:29 let me add those into the script Jul 18 10:55:32 your wish is granted. you have two left. Jul 18 10:55:39 heh Jul 18 10:56:53 # erase /media/internal first. It takes a while, and if user gets Jul 18 10:56:53 # impatient and pulls the battery we need /var to still hold the WIPE Jul 18 10:56:53 # flag file Jul 18 10:57:21 lol Jul 18 10:57:51 mountall is responsible for wiping the device Jul 18 10:58:08 danger! Jul 18 11:00:57 * rwhitby bbt Jul 18 11:01:03 nn Jul 18 11:01:09 I'll take a look at the checked-in scripts when I wake Jul 18 11:01:28 okay, I'll poke a lot more till I get them closer to what I want. Jul 18 18:45:52 'evening all :) Jul 18 18:48:55 howdy dtzWill, it's still morning here Jul 18 18:49:13 dtzWill: if you don't mind my asking, where are you located? Jul 18 18:49:19 actually it's 1:45pm here haha ^.^ Jul 18 18:49:25 not sure where my 'evening' came from :) Jul 18 18:49:26 d'OH Jul 18 18:49:33 especially b/c i woke up about an hour ago.... Jul 18 18:49:34 hehe Jul 18 18:49:39 it's 13:45 here too Jul 18 18:49:49 I am so out of it today Jul 18 18:49:56 got very little sleep last night Jul 18 18:49:57 linuxjacques_: i'm at UIUC (in Illinois) Jul 18 18:50:01 linuxjacques_: awww, :( Jul 18 18:50:17 linuxjacques_: due to work and other responsibilites, or difficulty sleeping? Jul 18 18:50:32 well, it's because I bought an android game and started playing it "just to try it" on my nookcolor Jul 18 18:50:53 got some friends who just ...can't sleep properly (sleep apnea amongst other things), and idk how they don't go entirely mad. Jul 18 18:50:55 linuxjacques_: hahaha nice Jul 18 18:51:48 what game? :) Jul 18 18:54:10 hmm we need to update the headless app deal to support touchpad Jul 18 18:54:28 i have a very strong urge to flat out uninstall any application that pops up that device border thing Jul 18 18:54:53 dtzWill: sorry, someone came into my cubicle Jul 18 18:54:59 the game is Galcon Jul 18 18:55:11 linuxjacques_: oh, yeah. i have galcon on my phone and galcon HD on the tablet Jul 18 18:55:11 lol Jul 18 18:55:14 i know how dangerous that game is Jul 18 18:55:16 I fired it up at 2:30 and before I knew it, it was 5am Jul 18 18:57:47 dtzWill: strange question: how much has double buffer extension been used/tested in our Xsdl? Jul 18 18:58:03 linuxjacques_: not at all. what do you ahve to report? Jul 18 18:58:23 i actually seem to recall that playing some game had weird issues and meant to look into it Jul 18 18:58:59 that said i don't see how any code path could result in double-buffering in the current design Jul 18 18:59:23 we have a block handler (runs whenever x has "nothing better to do") that pushes updated regions to the screen Jul 18 18:59:33 hardly a synchronous thing.... although i don't know how double-buffer is supposed to work. Jul 18 18:59:34 dtzWill: I built an app which uses the extension, and it's flaky (it could be the app, but it always seems to fail in the X double buffer opcode) Jul 18 18:59:56 flaky as in, sometimes it exits immediately with an X DBE error Jul 18 19:00:00 linuxjacques_: is the app available for me to play with, etc? Jul 18 19:00:01 oh weird Jul 18 19:00:09 you sure we even ahve the DBE? Jul 18 19:00:11 sometimes it runs for a while and then exits, and sometimes it runs for hours Jul 18 19:00:15 o_O Jul 18 19:00:28 dtzWill: xdpyinfo tells me we do :-) Jul 18 19:00:37 fair enough Jul 18 19:00:55 dtzWill: and it makes a difference, when the app is run without --double-buffer, it flickers Jul 18 19:01:20 it's drawing on the root window Jul 18 19:02:26 linuxjacques_: atm things will tear, not a whole lot to be done about it Jul 18 19:02:28 dtzWill: just saw your q, I haven't committed anything yet because I went down lots of long paths and just got it to even run last night Jul 18 19:02:43 i can look at the DBE and see what that'd take Jul 18 19:02:53 err there are things to be done, but not sure what the best is Jul 18 19:03:15 I was surprised at the number of extensions which Xsdl does report Jul 18 19:04:00 dtzWill: it probably makes sense to commit this app so you can use it to test Jul 18 19:04:18 shrug, i'm already got a bunch on my plate but certainly is interesting Jul 18 19:04:30 darn, I saved the output of xdpyinfo, but I saved it on my tp, which is not with me Jul 18 19:04:31 plate comment mostly b/c i'd feel bad if you went to the trouble to commit and i dind't have a chance to help, is all Jul 18 19:04:43 linuxjacques_: (do you want me to run it and pastebin the result?) Jul 18 19:05:27 dtzWill: do you have it? I added xdpyinfo to widk in my copy (and libXdamage and libXfixes IIRC) Jul 18 19:05:48 I *really* need to get re-setup so I can commit changes Jul 18 19:06:57 * dtzWill goes crazy with installing packages in his ubuntu chroot Jul 18 19:07:03 dtzWill: the strange thing is, when not being flaky, the double-buffering did remove the flickering, implying that it does _something_ Jul 18 19:07:17 dtzWill: lol I forgot all you guys have the chroot Jul 18 19:07:31 it doesn't show up for me in preware and nobody knows why :-\ Jul 18 19:07:50 :( Jul 18 19:08:01 i just got the chroot last night Jul 18 19:08:15 had the debian chroot for a while, but now using the same xecutah-integrated new-partition thingy everyone else is on Jul 18 19:08:23 zomg i love having 8GB for optware and chroot goodness Jul 18 19:08:48 lets you make your novaterm/ssh experience quite a bit more usable :) Jul 18 19:12:38 wait - is galcon available for the tp?? Jul 18 19:15:56 linuxjacques_: uh, yes. Jul 18 19:15:58 it's great. Jul 18 19:16:10 'galcon fusion' Jul 18 19:16:21 i mean as great as it is elsewhere anyway :) Jul 18 19:17:31 i've been spendign more time than i should looking how to make X work... 'well' Jul 18 19:17:50 with regards to much better rendering speed/drawing acceleration, and better support of the various extensions Jul 18 19:17:54 waayy more time than i should Jul 18 19:18:05 unfortunately don't have any particularly good ideas atm. Jul 18 19:18:45 best tentaive solution is to take the xgl/xegl work and rewrite glitz to use GLES instead of GL Jul 18 19:19:00 but that sounds like a world of hurt, and i'm not sure i can even do it, with its use of various intense GL stuffs :( Jul 18 19:19:46 I. must. buy. it. Jul 18 19:20:48 dtzWill: yeah I was interested in glitz, but I think it's deprecated? (but then, so is Xsdl :-( Jul 18 19:21:38 the more things we have using Xsdl, the more pressure to make it perform better and add features Jul 18 19:22:28 xsdl is definitely deprecated :/ Jul 18 19:23:02 and yeah xgl/glitz are deprecated Jul 18 19:23:13 but did exactly what i wanted: render accel via gl Jul 18 19:23:36 dtzWill: a few days ago, you asked me what hacks I was referring to that had to be done for old webos sdl/gles which might be affecting performance now. Jul 18 19:23:56 on of them is _not_ using the HW_ACCELERATED SDL visual Jul 18 19:24:07 because it didn't work right / was slower on pre Jul 18 19:24:14 I'm not so sure that's still the case Jul 18 19:24:34 you mean SDL_HWSURFACE ? Jul 18 19:24:41 I know the adreno has dedicated GPU memory, but I can't find out how much Jul 18 19:24:44 yeah I meant that Jul 18 19:24:57 i thought only pixi had the adreno? Jul 18 19:25:02 I think - I'm pretty sure ACCEL was in the define I was looking at Jul 18 19:25:06 tp has adreno Jul 18 19:25:12 very interesting. Jul 18 19:25:18 adreno 220 Jul 18 19:25:28 which qualcomm claims is 4x as fast as 200 Jul 18 19:25:30 a bunch of patches to VBA came to my attention in the last few days, someone finished porting it to the pixi Jul 18 19:25:40 amongst that are misc adreno-specific improvements Jul 18 19:25:56 might need to add these to the tp Xsdl :) Jul 18 19:26:03 indeed. Jul 18 19:26:34 I downloaded the adreno sdk from qualcomm - haven't had a chance to go through it Jul 18 19:26:36 biggest one afaict is a special write-only extension that is basically telling the gpu to optimize for our usecase Jul 18 19:26:39 http://www.khronos.org/registry/gles/extensions/QCOM/QCOM_writeonly_rendering.txt fwiw Jul 18 19:27:06 but there is certainly not any adreno programmers manual I could find :-\ Jul 18 19:27:15 interesting. Jul 18 19:27:19 the guy cited some adreno sdk manual Jul 18 19:27:22 lemme pull up the link Jul 18 19:27:32 :-) Jul 18 19:27:35 i also got ahold of the sgx manual and had a look through that as well Jul 18 19:27:49 like i said, i've been spending entirely way too much time on this :) Jul 18 19:28:04 it's amazing how much times these things take Jul 18 19:29:12 dtzWill: while I remember it - do you know anything about the state of MAME in webos? I know some ppl were trying to get it going. I was too at one point but hit some major blocks. Jul 18 19:29:25 last time I looked at it was early 2010 Jul 18 19:29:56 ah he cited a guide included with the adreno SDK Jul 18 19:30:01 which i think is publicly available Jul 18 19:30:13 don't remember much about it, other than what you just said Jul 18 19:30:21 dtzWill: I have the SDK - signed up at qualcomm's dev site and dl'd it Jul 18 19:30:33 but there are like 10,000 files in there Jul 18 19:30:37 >.< Jul 18 19:30:38 lol Jul 18 19:30:51 and I didn't see one which was obviously a manual Jul 18 19:31:14 that extension certainly does look like it's made for our X use case Jul 18 19:31:31 need to get a good way to benchmark performance Jul 18 19:31:41 if we're focusing on the blitting, might make a test app for that later Jul 18 19:31:42 definitely agree. Jul 18 19:31:55 there are old X perftest apps Jul 18 19:31:57 but that's just pushing what's already rendered into the buffer to the screen Jul 18 19:32:06 we have no accel of the rendering itself Jul 18 19:32:09 dunno how relevant they are - a lot of them are so old they don't use extensions Jul 18 19:32:14 although if x is using pixman nicely, i did enable the NEON extensions there... Jul 18 19:32:35 but idk if the xserver itself uses pixman or what :) Jul 18 19:32:43 and they test things nobody uses anymore like arcs :-\ Jul 18 19:32:49 right :/ Jul 18 19:33:30 is there an offscreen rendering extension in X? is that what RENDER is? Jul 18 19:39:19 dtzWill: did you get xdpyinfo in your chroot yet? :-) Jul 18 19:39:40 linuxjacques_: oh, yeah. it's in optware too Jul 18 19:40:44 linuxjacques_: http://pastebin.com/c6dfsmfz Jul 18 19:42:04 in optware? I totally forgot to look/ Jul 18 19:42:55 dtzWill, do you have link to cryptk|offline's upstart fix? Jul 18 19:43:23 dtzWill: see what I mean about the extensions list? it's abit suspect. I mean Composite, RENDER, XVideo ? Jul 18 19:44:25 scoutcamper: i do not, sorry Jul 18 19:44:30 linuxjacques_: composite works ;) Jul 18 19:44:35 transparent terminals :P ^______^ Jul 18 19:44:54 xvideo is support as well, but iirc just doesn't offer anything useful Jul 18 19:45:06 "yep we support xv" "oh cool, do you support FOO?" "no" Jul 18 19:45:07 lol Jul 18 19:46:33 dtzWill: excellent new on composite, but how's the performance? :-) is it hw accelerated? Jul 18 19:46:53 linuxjacques_: haha absolutely not Jul 18 19:46:57 (because the hardware can do it) Jul 18 19:46:58 but it works Jul 18 19:47:05 nod Jul 18 19:47:10 linuxjacques_: right, but we'd have to represent surfaces as textures, etc Jul 18 19:47:17 basically what the various projects like glitz try to do :/ Jul 18 19:47:32 if you think i'm missing something, lmk for sure. definitely more than a little lost in all this hehe Jul 18 20:06:19 yeah surfaces as textures seems to be the way lots of these technologies are going Jul 18 20:07:08 dtzWill: I don't think you're missing anything. I'm learning this as I go. reading lots of stuff Jul 18 20:07:23 I find it all very interesting. Jul 18 20:14:38 dtzWill: re-reading our conversation, maybe the guy was talking about the adreno optimization guide Jul 18 20:14:48 I _do_ have that (on my machine at home) Jul 18 20:14:53 linuxjacques_: quite possibly, i don't know. Jul 18 20:14:56 but yeah that sounds very likely Jul 18 20:15:04 also, i'm thinking maybe we should just use glitz... Jul 18 20:15:11 I made a note to read it several times, because I only understood about 20% of it Jul 18 20:15:25 dtzWill: we brought Xsdl back from the dead, why not glitz? Jul 18 20:15:32 yep! Jul 18 20:15:38 dtzWill: only thing, it would need to be glitzES Jul 18 20:15:40 well if we get glitz we'll probably want Xgl too Jul 18 20:15:51 linuxjacques_: oh, if it was just bringing it back i'd have done it already Jul 18 20:15:59 heh heh Jul 18 20:16:00 it's converting it to glitzES as yuo put it taht would be the challenge Jul 18 20:16:09 fun Jul 18 20:16:14 probably noit Jul 18 20:16:14 llo Jul 18 20:16:16 :P Jul 18 20:16:18 lol Jul 18 20:17:27 I keep telling myself 2d GL -> GLES must be easier than 3d Jul 18 20:20:20 dtzWill: did we look at this last time? http://sourceforge.net/projects/dogless/ "OpenGL to ES realtime translator" ? Jul 18 20:20:57 http://git.webos-internals.org/libraries/gl-wes-v2/ Jul 18 20:21:04 http://git.webos-internals.org/libraries/nanoGL/ Jul 18 20:21:05 haha Jul 18 20:21:07 we tried :( Jul 18 20:21:20 i remember looking at dogless, but... hmm Jul 18 20:21:32 don't remember why that didn't work, curious Jul 18 20:23:03 was using those things to try to get things like quake, etc working :) Jul 18 20:23:08 http://developer.qualcomm.com/file/607/adreno200performanceoptimizationopenglestipsandtricksmarch10.pdf Jul 18 20:23:30 note that's for the 200, I haven't found any docs for the 220 (in the tp) Jul 18 20:23:58 yeah I remember gl-wes-v2 Jul 18 20:25:03 also, http://developer.qualcomm.com/file/603/adrenosdk2101.zip Jul 18 20:25:15 these are on http://developer.qualcomm.com/tools Jul 18 20:56:55 i managed to break my virtual keyboard? Jul 18 20:56:56 lol Jul 18 20:56:56 xD Jul 18 20:56:57 x.x Jul 18 20:57:03 it isn't showing up Jul 18 20:57:05 and bt is off Jul 18 20:57:06 boo Jul 18 20:57:25 hmm Jul 18 21:04:03 nothing a restart can't fix... :( Jul 18 21:04:11 was having fun with lxde and oodles of ubuntu packages Jul 18 21:04:20 really wanted to play with kubuntu-mobile and similar Jul 18 21:04:28 but couldn't figure out how to start them Jul 18 21:07:52 adreno used to be called imageon Jul 18 21:08:22 ATI did Imageon, then were bought by AMD, who sold it to Qualcomm who renamed it Adreno Jul 18 21:10:08 I love how qualcomm's site says Adreno 220 is "Coming in 2011" Jul 18 21:21:54 guess i should just add xrandr support and call it a day :( Jul 18 21:27:58 "just add xrandr support" ? Jul 18 21:28:35 I thought it was not quick/easy to do Jul 18 21:30:00 i think it'll be not so bad Jul 18 21:30:13 mostly boilerplate stuff and handling misc state stuff during rotations Jul 18 21:30:33 but shouldn't be a big deal--mostly need to claim we support xrandr, and then advertise the orientations/sizes we support Jul 18 21:30:51 although all of this has lots of painful boilerplate :/ Jul 18 21:31:11 what about dynamic resolution changes? that's a sub-case of rotation? Jul 18 21:34:02 after xrandr, add xinerama, and then we can dynamically add screens :-D Jul 18 21:34:55 I still can;t think of a way to have an app's windows in different cards :-( Jul 18 21:34:56 linuxjacques_: yes, xrandr includes resolution changes Jul 18 21:35:07 The X Resize, Rotate and Reflect Extension, called RandR for short, (via http://cgit.freedesktop.org/xorg/proto/randrproto/tree/randrproto.txt , FYI) Jul 18 21:35:32 linuxjacques_: integrate the rootless X stuff (the cygwin and xquartz things i think use this) is best bet i'd imgaine Jul 18 21:35:57 hmmmmm Jul 18 21:36:07 I'll look at that Jul 18 21:36:19 and we can hook up the accelerometer to randr ^.^ Jul 18 21:36:27 nice Jul 18 21:36:30 ..with some kind of rotation lock >_> Jul 18 21:36:39 soo glad the device has that xD Jul 18 21:36:54 can we hook up the compass? I like my windows to always point north Jul 18 21:38:06 lolol Jul 18 21:38:09 linuxjacques_: <3 Jul 18 21:47:00 I guess the correct term is magnetometer :-) Jul 18 21:52:14 haha your humor is now famous on twitter! :) Jul 18 21:56:02 morning Jul 18 21:56:20 rwhitby: morning :), thought that was you. Jul 18 21:56:21 linuxjacques_: have you done a search for ubuntu in Preware? Jul 18 21:56:30 rwhitby: dose anyone else control the webosinternals twitter? Jul 18 21:57:34 rwhitby: no, it didn't occur to me. I'll try it as soon as I am reunited with my tp Jul 18 21:59:37 dtzWill: no Jul 18 21:59:40 haha, awww 'reunited' Jul 18 21:59:55 linuxjacques_: cause it's in Linux Apps, not Applications Jul 18 21:59:56 I miss it. Jul 18 22:00:12 i know, i love my tp :) Jul 18 22:00:16 dtzWill: are you enjoying the two click chroot convenience? Jul 18 22:00:37 rwhitby: hmm, I thought I was looking at everything in the webos-internals-testing feed Jul 18 22:00:40 rwhitby: hehe yeah, i am :). is it smart about not re-mounting, etc? Jul 18 22:00:48 no Jul 18 22:00:49 but yeah i absolutely am, and having 8GB to back it with is niiiiiice Jul 18 22:02:07 rwhitby: hmm I see it on my pre now. I guess I was doing it wrong :-\ Jul 18 22:04:07 since i got my tp, i pretty much only use my tp for phone calls, heh Jul 18 23:28:37 bleh... Jul 18 23:29:22 ka6sox: stay up too late last night? Jul 18 23:29:42 no, laptop died. Jul 18 23:29:48 its in hospital now Jul 18 23:29:49 * rwhitby spots a touchpad keyboard mapping Jul 18 23:30:10 need one... Jul 18 23:32:15 * linuxjacques_ looks around, "where? where?" Jul 18 23:32:40 ka6sox: the new dell one? Jul 18 23:34:00 no, mb Jul 18 23:34:35 ouch Jul 18 23:35:06 but...its going to be fixed right (I hope) and they said $free so thats a good number. Jul 18 23:38:26 ouch Jul 18 23:38:53 I wasn't going to feel too bad for the dell, but the mb I feel for Jul 18 23:48:52 okay, now back to our regularly scheduled programming...already in progress. Jul 18 23:53:40 bbl Jul 19 00:00:32 luckily the scripts are on my desktop here... Jul 19 00:00:53 let me push what I have (which is not ready to work but contains some outline bits.) Jul 19 01:47:34 yuck...the version of /bin/sh on the device is the busybox version.... Jul 19 01:47:39 no wonder the scripts are failing Jul 19 02:37:10 whats the likelyhood that the TP version of busybox sh will be anywhere near the one for Veer? **** ENDING LOGGING AT Tue Jul 19 02:59:56 2011