**** BEGIN LOGGING AT Fri Sep 16 02:59:56 2011 Sep 16 11:23:13 halfhalo: does node not utilize the openssl store of certs? Sep 16 11:23:29 I already have an app out to remove diginotar from webos Sep 16 15:38:53 jhojho thats a good question. I have to go look that up. More importantly, however, is to see if palm's node is the same or different in that regard. Sep 16 15:51:49 halfhalo, i've started a node term app :D Sep 16 15:51:57 sweet! Sep 16 15:52:12 halfhalo, u have any interest in working on it? Sep 16 15:52:20 you should make it so that it does not use the system node Sep 16 15:52:25 PuffTheMagic sure, why not Sep 16 15:53:03 how would it not use system node? Sep 16 15:53:16 isnt 3.0.3 coming out soon Sep 16 15:53:21 wont that have new node? Sep 16 15:55:29 yeah, but not using the system node let's it be run on older devices, and gives the ability to update the node it is using much easier Sep 16 15:58:44 well the ui i was planning on having as enyo at first Sep 16 15:58:50 maybe mojo compat later Sep 16 16:07:00 halfhalo, gonna have to take some ideas from Ace Sep 16 16:07:13 from the editor area Sep 16 16:07:23 nice Sep 16 16:07:25 using individual dom nodes for each char Sep 16 16:07:59 i think that would be the way to do it Sep 16 16:09:50 yup Sep 16 16:18:25 halfhalo, got a question Sep 16 16:18:33 yesterday you said I cold just do this.... Sep 16 16:18:48 var tty = require('tty'); Sep 16 16:18:48 tty.setRawMode(true); Sep 16 16:18:48 b = tty.open('/bin/bash'); Sep 16 16:18:48 b[0].on("data",function(data){console.log(data)}); Sep 16 16:19:06 how do I know if that is from stdout or stderr Sep 16 16:26:32 looking at the source now... it looks like both stderr and stdout are mapped to the same fd. Sep 16 16:27:16 hmm Sep 16 16:28:08 halfhalo, here is my current service http://dpaste.com/616048/ Sep 16 16:28:12 im worried about 2 things Sep 16 16:28:14 most of the code for the tty module is just keycode handling, the actual opening of a process is done with child_process Sep 16 16:28:26 what happens when run gets to the end of the function Sep 16 16:28:34 and how could I handle multiple instances Sep 16 16:29:49 PuffTheMagic and now you know why I completely and utterly hate the way services are implemented. Sep 16 16:30:29 ya i dont even really understand why i need to use a service for all of this Sep 16 16:31:20 ha Sep 16 16:31:27 0.2.3 dont even have the tty module :D Sep 16 16:32:28 ka6sox, do we have newer versions of node building on the autobuilder at all? Sep 16 16:32:42 would be nice to have a update node ipk Sep 16 16:33:09 PuffTheMagic, no I don't unless halfhalo submitted it Sep 16 16:33:55 It can build them, but they cannot replace the current shipping node Sep 16 16:34:18 okay so that means it belongs in /opt Sep 16 16:34:21 halfhalo, new node versions break api compat? Sep 16 16:34:36 just like ruby/rails Sep 16 16:34:44 ka6sox, /opt doesnt help if i cant control where a webos app looks for its node Sep 16 16:34:53 sure you can Sep 16 16:34:58 you can specify where it looks. Sep 16 16:35:01 :D Sep 16 16:35:06 how Sep 16 16:35:06 (unless its in a jail) Sep 16 16:35:22 if its in a jail then that makes it tough Sep 16 16:35:42 yes and no. it breaks the abi compatability with palm's node addons in the 0.2.x series. the 0.4.x should be much more flexible in that regard Sep 16 16:36:01 thats all halfhalo questions. Sep 16 16:36:04 sorry. Sep 16 16:36:34 there is already a shim in place so that most services targeted at 0.2 work on 0.4 Sep 16 16:36:46 but new node can be put in there and a jailbroken app can use them if called explicitly Sep 16 16:37:08 or use the shim :D Sep 16 23:13:12 dtzWill: see /msg Sep 16 23:32:07 so now that rwhitby accurately reported that you can have more than 1 chroot on a device... Sep 16 23:32:24 we need to tweak the icons so that they are distro specific. Sep 16 23:32:31 hi eric Sep 16 23:32:40 egaudet Sep 16 23:32:43 heyo Sep 16 23:33:02 I have a question about uinput Sep 16 23:34:39 is there a way we could intercept keycodes from an external keyboard and pass them to SDL? Sep 16 23:34:39 k Sep 16 23:34:51 hmm Sep 16 23:34:58 (specifically the X-server) Sep 16 23:35:14 well you can generate all the hard key strokes you want with uinput Sep 16 23:35:34 as long as you have a way to intercept the external entries then yes Sep 16 23:36:27 okay so BT keyboard uses BT_Spp support that passes to keyboard input Sep 16 23:36:27 For the keyboss stuff I intercepted the actual hard keys and then generated whatever I wanted instead Sep 16 23:36:34 okay Sep 16 23:36:39 we don't have those. Sep 16 23:36:45 don't have what? Sep 16 23:36:55 hard keys. Sep 16 23:37:05 because its a bluetooth keyboard. Sep 16 23:37:18 well is it still using the input subsystem? Sep 16 23:37:24 yes Sep 16 23:37:49 I guess I could "tee" them. Sep 16 23:37:50 so the bt keyboard uses the input subsystem, so you can intercept it from that way. Now the question is where does hidd get the virtual keystrokes from Sep 16 23:38:11 from the VKB? Sep 16 23:38:27 i.e. for phone webOS with hard keyboard, hidd was reading the input subsystem to get the keystrokes and then giving them to luna Sep 16 23:38:55 ah, okay Sep 16 23:39:00 so uinput will only be useful to inject in the event that hidd still reads from input Sep 16 23:39:14 otherwise, have to find out if there's another way to get in between hidd and luna Sep 16 23:39:20 right. Sep 16 23:39:20 egaudet: so I'm thinking of supporting patches to /media/cryptofs stuff, but appending the md5sum of the unpatched file to the appdir md5sums record. Thoughts? Sep 16 23:39:29 s/but/by/ Sep 16 23:39:34 assuming it's still similar to that because I haven't so much as looked at webos 3 yet unfortunately :( Sep 16 23:40:09 egaudet, okay I need to find out where hidd hooks into the BT subsystem. Sep 16 23:40:27 I know it probably looks for a notification on LS Bus. Sep 16 23:41:11 rwhitby, so just recording the md5sum of the original files? What is being solved ? Sep 16 23:41:34 ka6sox, i thought you said it hooks into input subsystem, it's the vkb that might be tricky part Sep 16 23:41:47 no Sep 16 23:41:56 the vkb issue has been solved Sep 16 23:42:00 not pretty...but workable Sep 16 23:42:06 egaudet: deletion of original file when reverting is the issue Sep 16 23:42:33 the BT keyboard still doesn't recognize things we need like ctrl and esc Sep 16 23:43:07 rwhitby, I'm not following, who deletes the original? Sep 16 23:43:34 the prerm does at the moment. Sep 16 23:43:40 ka6sox, so the BT keyboard should be coming in at /dev/input/ right? Sep 16 23:44:08 egaudet: the logic says "I can't find an md5sum of the original, so it must have been a new file, so I'll delete it" Sep 16 23:44:44 egaudet, yup, I'll find out here in a little while Sep 16 23:45:02 rwhitby, oh do they not include md5sums of all files anymore? Sep 16 23:45:13 dtzWill, the current X-server isnt' in a Jail right? Sep 16 23:45:22 egaudet: I'm referring to files in /media/cryptofs Sep 16 23:45:28 (of course not....wouldn't work) Sep 16 23:47:08 ka6sox: well the x-server itself i thought could be in a jail Sep 16 23:47:18 xterm defnitely ont--but either way atm we launch both out of the jail via xecutah Sep 16 23:47:21 rwhitby, oh sorry I'm so behind, files out of the appdir? I think then yeah it might be safest to add/append md5sums file although "proper" solution might be to fix logic. I've been away too long to think about it clearly though so I might be off Sep 16 23:47:47 dtzWill, being in a jail leaves messes in tmpfs Sep 16 23:47:56 egaudet: files outside of what is recorded in /md5sums or in doctor. Sep 16 23:48:34 dtzWill, let me look @ teh btkeyboard input nod and see what I see there when I send various keys. Sep 16 23:48:40 rwhitby, yeah that's a bummer, I was under the impression that every file in webOS would stay as part of a md5sum'd package :( Sep 16 23:48:49 ka6sox: yes, although that should only be an issue if you run it /outside/ the jail (by hand or xecutah) *then* from within a jail Sep 16 23:49:07 otherwise i think it cleans up after itself. it used to--making it jail-friendly was a goal for a bit Sep 16 23:49:19 okay Sep 16 23:49:22 egaudet: for these files, the ipkg is unpacked at run time. so the md5sums do actually exist, but in /media/cryptofs/apps/usr/lib/ipkg/info as separate package md5sums Sep 16 23:49:22 ka6sox, so do you know where the VKB sends keystrokes to? Sep 16 23:49:39 no, but I am going to look tonight Sep 16 23:49:43 dtzWill: did you see that folks found the arrow keys? Sep 16 23:50:22 egaudet, if we can solve the input problems then xterm is a good solution for a terminal. Sep 16 23:50:35 rwhitby, doesn't the aupt logic check individual app md5sum files? i.e. specifically NOT /md5sums Sep 16 23:50:39 right now we have problems with ctrl, arrows, esc. Sep 16 23:50:58 egaudet: not that I can see Sep 16 23:51:14 egaudet, so we have to get them "before" luna sucks them up. Sep 16 23:52:26 rwhitby, I thought I remember avoiding the big md5sums files because it wasn't updated on OTA. It -should- use the package's md5sums file Sep 16 23:52:30 ka6sox, egaudet: control and arrow keys work for bt devices, luna doens't eat them Sep 16 23:52:48 luna *does* like to om nom on escape key Sep 16 23:53:04 and control/arrow keys on vkb are 'broken' b/c vkb doesn't have them Sep 16 23:53:26 so the goal is to "fake" ctrl from vkb ? Sep 16 23:53:38 by injecting whatever the bt keyboard does? Sep 16 23:53:56 egaudet: ah, it pobably only uses /usr/lib/ipkg/info instead of also /media/cryptofs/apps/usr/lib/ipkg/info Sep 16 23:54:09 dtzWill, then I wasn't able to duplicate what they said. Sep 16 23:54:12 people ahve discovered that the keyboard code conversion mapping i do results in sending what gets interpreted as control/etc which is nifty Sep 16 23:54:25 I gotta run in about 10 minutes, but tomorrow I should have some free hack time (finally!) and I'll try to help poke around this vkb/btkb stuff Sep 16 23:54:28 I was not able to get the alt set to come up. Sep 16 23:54:34 egaudet: that'd be great :D Sep 16 23:54:38 egaudet, thanks. Sep 16 23:54:54 ka6sox: what works/doesn't for you? Sep 16 23:54:58 rwhitby, yes that makes sense, I never saw an md5sums in /media/cryptofs/apps/usr/lib/ipkg/info before Sep 16 23:55:10 ka6sox: control, arrow keys, and alt ('sym' on the bt keyhboard) should all work Sep 16 23:55:14 on bt keyboard Sep 16 23:55:20 egaudet: appcat, quickoffice, and other stuff ends up in /media/cryptofs/apps now Sep 16 23:55:34 dtzWill, was that an update in the X-server? Sep 16 23:55:45 if not then I am confused. Sep 16 23:55:48 dtzWill: BTW, any chance we can get Sym key support in XServer on phones? Sep 16 23:55:55 rwhitby, interesting, is that so they can update via app cat? Sep 16 23:56:00 egaudet: yes Sep 16 23:56:11 ka6sox: it was quite some time ago, yes Sep 16 23:56:16 hmmm Sep 16 23:56:17 okay Sep 16 23:56:21 so we just need esc Sep 16 23:56:33 ctrl-[ Sep 16 23:56:54 which currently somehow gets through from the virtual keyboard Sep 16 23:57:11 rwhitby: it doesn't get through, the vkb doesn't "Emit" escape or anyting else Sep 16 23:57:14 people just say it does Sep 16 23:57:19 it emits the key it says it emits Sep 16 23:57:24 and as a hack to do the mapping Sep 16 23:57:39 x takes the vkb key, tells sdl to /convert it to unicode/ Sep 16 23:58:00 xkb mappings then just say 'keycode for unicode character X is... X' Sep 16 23:58:04 cool cool, well I gotta run for tonight. Be back tomorrow to get my hands dirty :D hopefully bip doesn't crash on me Sep 16 23:58:17 okay egaudet good to see you. Sep 16 23:58:23 with special cases for control/arrow keys/etc that are mapped to 255, 254, 253, 252, 251, etc Sep 16 23:58:40 since those as unicode didn't seem interesting at the time, and because it worked Sep 16 23:58:48 it's a big hack :[ Sep 16 23:59:19 anyway so that's why luna doesn't eat those events, because they're not escape or anythnig, people just got lucky that the translation pipeline happens to make those values result in whatever keys they're seeing X emit Sep 16 23:59:32 yay for unintentional bugs that are not features Sep 16 23:59:35 *now features Sep 17 00:00:01 dtzWill: so my plan for this weekend is to somehow get people off the testing feed Sep 17 00:00:19 rwhitby: yes, i think that's long overdue. Sep 17 00:00:29 how should it be done? Sep 17 00:00:39 preware update that drops testing feed, and change the testing feed url? Sep 17 00:01:18 I'm thinking for every package which was in testing but has migrated to public, add another pseudo-package in Preware which identifies itself as an update, but tells the person to disable the testing feed to see the real version of the pacakge Sep 17 00:01:37 however that gets annoying for real testers Sep 17 00:02:27 so maybe I do change the URLs and create alpha and beta feeds Sep 17 00:03:14 hmm Sep 17 00:03:43 the psuedo-package sounds like a lot of work and mess, and in particular i think sends a little bit the wrong message Sep 17 00:04:29 more of a "we control this, you want testing feed then diiiee" vs "hey if you don't belong here all that coolness is now in main feed, have a good one! [if you still want the testing feed you know what to do]" Sep 17 00:04:51 alpha/beta works, but i miss just having the one testing feed that was just for testing Sep 17 00:05:21 crazy to have a feed with distinct alpha/beta/prod stages from an 'organization' like this xD Sep 17 00:07:13 dtzWill: in general, we know when things go from alpha to beta Sep 17 00:08:07 hmm, okay. Sep 17 00:08:47 i dind't mean crazy as in crazy wrong, mostly just chuckling at the idea for reasons really not worth expressing, mostly b/c i'm unsure myself :) Sep 17 00:09:15 so then we'd just be adding an alpha feed and call the existing one beta Sep 17 00:10:05 dtzWill: no, somethings from current testing would go into each Sep 17 00:10:20 and we'd do a transition period Sep 17 02:49:07 scoutcamper|away, ping? **** ENDING LOGGING AT Sat Sep 17 02:59:57 2011