**** BEGIN LOGGING AT Fri Jul 15 02:59:57 2011 Jul 15 03:56:22 elpollodiablo-aw, when you get back try the wiki. Jul 15 10:05:10 * rwhitby is about to push XServer and XTerm to the testing feed Jul 15 10:07:17 rwhitby, does that mean we got the keyboard mappings fixed up? Jul 15 10:07:38 Fixed up well enough. Jul 15 10:22:48 cool Jul 15 11:11:11 Xecutah, XServer and XTerm in testing feed now. Jul 15 11:18:24 nice Jul 15 11:18:39 rwhitby, are those current as of "now" Jul 15 11:18:44 yes Jul 15 11:18:55 cool, so they are more up to date than what I have Jul 15 11:19:04 but I'm doing an update to xserver to force portrait on phones Jul 15 11:19:20 I am assuming that Xecutah either has or will have a framework similar to tweaks, so that things can add things into the menu? Jul 15 11:37:39 rwhitby, on the touchpad... when Preware prompts for a restart it prompts for a Phone restart... it should probably be changed to Device rather than phone Jul 15 11:38:00 rather than coming up with some device specific nonsense, just use a nice neutral word Jul 15 11:48:15 Fixed in 0.7.2 in the testing feed now. Jul 15 11:48:22 s/0.7.2/1.7.2 Jul 15 13:35:37 [dtz]: the control works for functions like tab = gesture(ctrl)+i other then that I dont know if any of the other control based key mappings that work. Last time I trying doing any mapping with control/gesture it didnt work, but I will have to try that again. Jul 15 13:41:52 So, Xecutah is not working properly on 1.4.5, cause the shell script are causing defunct processes which are stopping Xecutah being able to run the second command. Jul 15 13:43:38 why did it work in our testing? or did we never test it on 1.4.5? Jul 15 13:45:16 it always failed like that on 1.4.5 Jul 15 13:45:23 I only just worked out the reason why. Jul 15 13:48:00 ka6sox: Thanks for fixing access to the wiki for me. Jul 15 13:53:24 ok, so I need a way for a C program to execute a shell script, and for the shell script to run the X server and then exit, and for the c program not to wait for the X server to exit. Jul 15 13:53:44 fork? Jul 15 13:53:47 the X server must continue running. Jul 15 13:54:09 yeah, maybe I do have to fork the service Jul 15 13:57:56 can you just fork the X server? Jul 15 13:58:32 so replace /line/that/runs/startx with ( /line/that/runs/startx; )& Jul 15 13:58:44 assuming that you are calling startx, but easily changeable Jul 15 14:10:25 fork doesn't fix it either Jul 15 14:10:42 and the line is running Xsdl, not startx Jul 15 14:12:47 yeah, I figured startx was wrong, but knew that you would know to adapt it Jul 15 14:13:06 so forking off the X server didn't fix it... Jul 15 14:13:15 no Jul 15 14:13:31 maybe running the entire shell script in a fork? Jul 15 14:13:44 so instead of forking in the shell script... fork in the service? Jul 15 14:13:46 that's what I've tried Jul 15 14:13:48 oh Jul 15 14:13:51 well damn.. Jul 15 14:14:03 forking in the service, forking in the script, everything. Jul 15 14:14:14 weird... Jul 15 14:14:38 and it's only on 1.4.5 - on 2.0 and later exactly the same binaries work fine Jul 15 14:15:07 what changed between 1.4.5 and 2.0 with regards to services... it has to be something with the jail config or something... Jul 15 14:15:11 or shell versions Jul 15 14:15:42 * cryptk doesn't have any 1.4.5 devices to test on Jul 15 14:15:56 I don't even have any devices around that are capable of running 1.4.5 Jul 15 14:17:19 cryptk: you just have TouchPads and Pre2 and Veer? ;) (I envy you) Jul 15 14:18:12 I don't have a veer Jul 15 14:18:17 I have a TP and a Pre2 Jul 15 14:18:45 I have an old Pre- somewhere around here in a box, but it is one that I haven't unpacked yet from my move... and I have no clue what box it is Jul 15 14:21:49 cryptk: it's nothing to do with services or jails Jul 15 14:22:05 it's a C program running as root outside of any jails executing a shell script Jul 15 14:24:05 Well, I'm done with it. Xecutah, XServer and XTerm marked as webOS 2.0 minimum is my solution to this problem. Jul 15 14:28:19 * rwhitby bbt Jul 15 14:38:11 well...I guess I don't have to work on that anymore.... Jul 15 14:38:15 he did it Jul 15 14:54:45 one of the reasons I didn't want to do a C thing for monitoring when x-server dies is that it leaves a running c service after that till you reboot. Jul 15 15:09:36 cryptk: do you have the preware version of xecutah Xterm and Xserver installed or one you built? Jul 15 15:09:57 preware on the Pre2 and the TP at the moment Jul 15 15:10:14 and xecutah, xserver and xterm work on both Jul 15 15:10:34 the Pre2 is on 2.1.0 Jul 15 15:10:53 I dont think we ever tried it on 2.0.1 Jul 15 15:11:11 I don't have a 2.0.1 to work with Jul 15 15:11:17 yeah me neither Jul 15 15:12:28 sleep time for me, nn all Jul 15 15:12:58 nn cryptk Jul 15 15:19:40 it seems like Xterm does not work on 2.0.1 either, I dont know if it is the same reason it does not work on 1.4.5 or not but it sounds similar to it. Jul 15 15:43:25 hey folks, everyone still sleeping? Jul 15 15:45:58 no Jul 15 15:47:36 I was just catching up on the backlog, any idea what the plans for xterm rotation are? Jul 15 15:47:57 "rotation"? Jul 15 15:48:25 like multiple screens? Jul 15 15:48:30 is it currently in landscape for TP use? Jul 15 15:48:42 Just wondering why landscape vs portrait Jul 15 15:50:22 <[dtz]> so it's pushed, eh? :) Jul 15 15:50:46 <[dtz]> bring on the complaints regarding issues we know aren't fixed yet/are still working on Jul 15 15:50:52 * [dtz] grumbles it's a testing feed Jul 15 15:50:56 is it pushed? Jul 15 15:51:33 [dtz]: How did you determine the keycode for the gesture area in the original pre mapping? Jul 15 15:51:34 [dtz], ya, its starting. Jul 15 15:51:51 <[dtz]> rwhitby: fwiw xserver absolutely *can* run withuot xecutah just fine, that's something time was spent on :) Jul 15 15:52:08 <[dtz]> but maybe i'm bad at striking balance between what end-users will want to understand and the truth Jul 15 15:53:39 <[dtz]> like xecutah not actually doing anything other than exec'ing xserver/xterm. presenting it as xecutah doing everything ("allows you to run Xserver in a card", etc) strikes me as dishonest, but then again is probably simplest for end-user experience. Jul 15 15:54:01 <[dtz]> (if they need xecutah to run the only x client that exists, and perhaps the most important one, xterm, why would they want to run xserver by itself? etc) Jul 15 15:54:07 <[dtz]> ka6sox-farfarawa: hehe Jul 15 15:54:16 <[dtz]> Error454: trial and error, actually. Jul 15 15:54:27 <[dtz]> Error454: but it's also documented too ^.^ Jul 15 15:54:57 [dtz]: I saw that key mapping javascript in the older OS but not in 2.1 Jul 15 15:55:55 [dtz]: I was trying to spit out raw keystrokes through c, but the gesture area doesn't modify any bits Jul 15 15:56:39 <[dtz]> cool that we can just point to dev beta xD, that solves that. Jul 15 15:57:05 [dtz]: the pre2 is different, so I'm attempting to track it down Jul 15 15:57:38 <[dtz]> Error454: if you uncomment #define DEBUG in sdl.c, it'll log all the keystrokes Jul 15 15:57:46 <[dtz]> (and rebuild, etc, obviously) Jul 15 15:58:25 [dtz]: to messages? Jul 15 15:58:52 <[dtz]> Error454: what are messages? Jul 15 15:59:26 as in /var/log/messages Jul 15 15:59:58 cryptk|offline, rwhitby: keymappings are still where they were when i left them/few days ago, etc, right? Jul 15 16:00:04 I'll figure it out Jul 15 16:00:32 Error454: rotation--presently we force portrait on the phone, since landscape is kinda silly Jul 15 16:00:43 dtzWill: I was thinking the same thing about why would you need Xserver to run by its self if Xecutah is the only one to have the ability to run X progams. that might help solve the issue with 1.4.5 and 2.01 not being able to execute a second command to run Xterm. Jul 15 16:00:57 Error454: but on the tp we detect orientation via accelerometer and set it accordingly Jul 15 16:01:16 note that if the device is flat, that won't work, but it's best we can do. Jul 15 16:01:28 down the line we probably want to/will support dynamic rotation, etc Jul 15 16:01:34 dtzWill: On my last build, landscape is forced on the pre2. Did they change the rotation values on the pre2? Jul 15 16:01:38 but that's signifiacntly less important than say, having *shift* work on the tp Jul 15 16:01:40 elpollodiablo1, but is there any reason not to allow the app itself to call the X-server Jul 15 16:01:47 Error454: all phones are portrait in latest build, yes. Jul 15 16:02:28 not executah but the app you are launching. Jul 15 16:02:32 ka6sox-farfarawa, no I have no issue with it calling the app itself but why force the end user to do it? Jul 15 16:02:59 to launch Xserver first then launch the app they want Jul 15 16:03:10 right Jul 15 16:03:16 I isnt that waht was going on with the old xterm? Jul 15 16:03:28 so make the xserver an event.d thing and have the app call it Jul 15 16:04:38 I am just throwing out ideas to take dztWill comment and potentially coming up with a fix for 1.4.5 and 2.0.1 issue. Jul 15 16:04:56 not the user first starting the xserver Jul 15 16:05:33 means the app calls the x-server daemon then does its magic to attach. Jul 15 16:05:46 but that may still not work on 1.4.5 Jul 15 16:06:08 i know nothing about the 2.0.1 or 1.4.5 bug, not actually having devices running those handy atm. Jul 15 16:06:10 :( Jul 15 16:06:45 elpollodiablo1: xecutah just does the same thing icons for xserver/xterm used to do, only runs them outside the jail Jul 15 16:07:01 elpollodiablo1: while not auto-strating an xserver with xterm is a design decision, i'm not sure it'd fix anything :/ Jul 15 16:07:21 (or am i misunderstanding what you were suggesting?) Jul 15 16:07:28 dtzWill, understood that wont' fix this bug Jul 15 16:07:33 ka6sox-farfarawa: true, might need to rethink something and terming what is more important. Jul 15 16:08:09 dtzWill, I'm just trying to work out a clean flow for chroots (ie) Jul 15 16:08:20 ka6sox-farfarawa: /me listens Jul 15 16:08:22 terming=determining Jul 15 16:09:05 with a chroot you want to bind mount /sys /dev /proc to the chroot before you chroot to it. Jul 15 16:09:22 then after you are done tear those down so you are OTA safe Jul 15 16:14:15 dtzWill: what was the design decision behind not auto-starting a xserver with xterm? Is the goal to do a 1 xserver to run multiple x apps type of thing? Jul 15 16:14:31 elpollodiablo1: there was none really, afaik. Jul 15 16:14:56 elpollodiablo1: xserver runs fine all by it's lonesome, and xecutah was built to simply solve the needing to break the jail for xterm Jul 15 16:15:09 so rwhitby banged out that service, and made it exec xterm Jul 15 16:15:29 while he was at it, added a separate line for starting the xserver, and that's kinda how that happened. Jul 15 16:15:48 i think as we add other x clients, or chroots, or what-have-you, we'll probalby want to think more carefullya bout how to architect this Jul 15 16:16:10 the problem is that the stuff to make xecutah go gets shoved into the background..kinda like libs and it all then becomes the "app" Jul 15 16:16:18 oh alright, well the issue for 1.4.5 is that the 2nd command (xterm) from xecutah never gets executred Jul 15 16:16:46 atm the big push is "We need a terminal!", with emphasis on "now"--so i think we'll come back to clean up later Jul 15 16:17:03 elpollodiablo1: by 2nd command do you mean 'when i hit the xterm button in xecutah nothign happens'? Jul 15 16:17:12 yep Jul 15 16:17:18 or that there are multiple comamnds in the xecutah.sh script in the xterm project, the 2nd of which never firtes Jul 15 16:17:20 i see. Jul 15 16:17:43 you click Xserver card pops up, tap Xterm nothing loads Jul 15 16:18:52 kk. that's the symptom, someone'll have to debug what exactly is failing at some point. Jul 15 16:19:04 ka6sox and rod would have more of the technical explaination for it. Jul 15 16:19:08 i should have a 1.4.5 device somewhere around here that can at least boot to play wiht it. Jul 15 16:20:38 I think rod already did the debugging but didnt think it was a big issue and just made it webos 2.x and greater before we found out it doesnt seem to work for 2.0.1 Jul 15 16:23:26 you might want to read the log before you came on, rwhitby and cryptk had a descussion about the issue and some things they tried to fix it with out any luck. Jul 15 16:24:18 elpollodiablo1: haha i _did_ read the backlog, hehe Jul 15 16:24:37 i'm still ping'd on all mentions of xterm/xserver from when this was my baby a year ago haha Jul 15 16:24:46 so even if i dind't wanna, it'd at least make my awaylog Jul 15 16:25:08 anyway yeah i should go re-read taht, my impression was rwhitby diagnosed it but wasn't able to fix it atm and in the spirit of pushing things along said skip 1.4.5 for now Jul 15 16:25:15 oh I wasnt aware of that Jul 15 16:25:47 yeah that is my impression as well Jul 15 16:25:54 mine too... Jul 15 16:26:08 he put a minimum version of 2.x is what I saw. Jul 15 16:29:47 Sorry, just went through a tunnel on my morning commute. So with meta-tap on Pre2, x, c and v are the only 3 keys that do not work. Jul 15 16:30:51 This is from the SDL layer spitting out keycodes Jul 15 16:31:24 Does meta-c work on the Pre+? Jul 15 16:33:46 off to work, have a good one Jul 15 17:39:44 someone tell Error454 when he gets back that gesture+{c,v,x} don't work b/c the OS om nom's those key presses for paste/etc Jul 15 17:41:42 no ctrl-x ctrl-c ? but ... emacs :-( Jul 15 17:42:20 linuxjacques_, ctrl-a x Jul 15 17:43:38 is that a screen thing? Jul 15 17:44:10 yes, but we could have the X-server or the apps themselves escape things. Jul 15 17:44:15 annoying for sure. Jul 15 17:44:27 but maybe necessary Jul 15 17:45:15 vim works fine as far as I can see for what I do. Jul 15 17:45:35 seems like an app should be able to say "I'm not interested in system cut/paste, gimme ma gesture-[cvx]" Jul 15 17:46:01 tell luna yhat Jul 15 17:46:20 yeah, I mean there should be a way to tell luna that Jul 15 17:49:20 hehe vim works great ^.^ Jul 15 17:49:31 linuxjacques_: agreed, but last time i checked (1.4.5) you couldn't do that :/ Jul 15 18:16:07 dtzWill: what else need to be included in xserver to get the keyboard to handle sticky keys and some of the other functions implemented in the keyborad config stuff? Jul 15 18:43:26 elpollodiablo1: xkbset is already shipped, it just needs to be invoked. Jul 15 18:43:39 that makes sticky keys work on phones iirc Jul 15 18:43:56 i actually dno't know how the bt keybord works atm, it's paired with my media center for now :P Jul 15 18:44:10 the problem with shift/etc on the virtula keygboard, is it doesn't send 'shift' events Jul 15 18:44:22 hopefully sdl does see the modifiers somehow, haven't poked at that yet. Jul 15 18:44:24 (A vs a, etc) Jul 15 18:44:35 (i'm pretty sure the SDL_Event for keyboards does have modifier information) Jul 15 18:44:45 so anyway we don't send an explicit shift event to X, which is why that doesn't work on the tp Jul 15 18:44:56 (tangential from sticky modifiers) Jul 15 18:45:58 so we just need to add some magic invocations there :) Jul 15 18:46:13 since the touchpad virtual keyboard won't hav any modifiers, we can probalby just always enable them Jul 15 18:46:16 not sure re:bt touchpad yet Jul 15 18:49:24 yeah the BT keyboard vs onscreen might need different mappings or some "magic invocations" Jul 15 18:49:24 oh damn i never pushed my original X changes for the new foramt Jul 15 18:49:25 >.< Jul 15 18:49:36 elpollodiablo1: indeed. ideally we'd support both? Jul 15 18:49:38 shrug. Jul 15 18:49:46 no reason you can't change the mapping on-the-fly :) Jul 15 18:49:58 that is what I was thinking Jul 15 18:49:59 and we can resize the screen to be fullscreen for bt if/when we support dynamic xrandr Jul 15 18:50:38 that would definatly make it more ideal Jul 15 18:51:03 baby steps :) Jul 15 18:51:48 for the sticky key do we know what needs to be invoked? or should I look into it? Jul 15 18:52:20 elpollodiablo1: i have the commands somewhere, gimme a sec :) Jul 15 18:52:52 +# Make the modifiers sticky! Jul 15 18:52:53 +sleep 1 Jul 15 18:52:53 +xkbset exp 60 =sticky =twokey =latchlock Jul 15 18:52:53 +xkbset sticky -twokey latchlock Jul 15 18:53:02 xkbset should be in bin/ Jul 15 18:53:07 it is Jul 15 18:53:12 I was messing with it already Jul 15 18:53:17 run that and lemme know how it goes ^.^ Jul 15 18:53:25 if that works i'll add it... Jul 15 18:53:52 I am guessing you didnt check all the key mappings? Jul 15 18:54:16 elpollodiablo1: 'check'? Jul 15 18:54:23 elpollodiablo1: all the keymappings for palm-pre original should work. Jul 15 18:54:31 I know those work Jul 15 18:54:33 palm-prex should work, but i don't remember how extensively i tested it no. Jul 15 18:54:41 but alot has been added to palm-prex Jul 15 18:54:43 since i never even pushed that to the public testing feed Jul 15 18:54:50 well the keycodes are the same Jul 15 18:54:50 lol Jul 15 18:54:53 so.... Jul 15 18:55:07 yeah the rest changed, does something not work? Jul 15 18:55:52 elpollodiablo1: lmk how it works when you get a chance please :) Jul 15 18:56:15 when I first didnt some test it seemed a little off, like ctrl didnt seem accurate and some of the other modifiers did nothing although the mapping has almost something for any key plus modifier Jul 15 18:56:48 I will have to do it when I get home, in a few hours but I just wanted to make sure we were not duplicating efforts Jul 15 19:16:35 rwhitby: can we update xecutah to only show the touchpad warning, if in fact, we are on a touchpad? :) Jul 15 19:17:10 ^______________^ Jul 15 19:17:16 the performance improvements i made for the touhcpad Jul 15 19:17:28 definitely improve thigns on the pre2 too Jul 15 19:17:30 :D Jul 15 19:17:51 does it? share!!! Jul 15 19:17:51 ;) Jul 15 19:18:02 do them? (sorry) Jul 15 19:18:21 NuttyBunny: in testing feed, did them early on Jul 15 19:18:59 the touchpad's higher resolution made them necessar Jul 15 19:19:00 +y Jul 15 19:19:05 they're in the testing feed with the xserver Jul 15 19:19:18 basically it optimizes the "not redrawing the entire screen" case Jul 15 19:20:09 nice... I should check the testing feeds more often ;) Jul 15 19:20:19 NuttyBunny: wel if you have been running any xserver on the touchpad Jul 15 19:20:21 you've seen it. Jul 15 19:20:37 it was so necessary that i don't know that i even commited code without it Jul 15 19:20:37 lol Jul 15 19:21:37 *X Touchpad envy* Jul 15 19:23:35 The main reason I want to get a touchpad Jul 15 19:24:21 * NuttyBunny has no Touchpad Jul 15 19:24:33 HP/Palm no longer loves Mexico Jul 15 19:25:57 is there an easy way to auto-type a users name in here? Jul 15 19:26:34 kind of like how tab auto-finishes Jul 15 19:28:50 elpollodiablo1: tab usually does it... Jul 15 19:29:36 wow I feel stupid Jul 15 19:30:27 gesture+square in "wire" (wirc) Jul 15 19:31:13 that is helpful to know. I think I need to get wirc now Jul 15 20:10:47 once I do a git pull do I need to do make setup again? or do I just need to make for the packages I want to build? Jul 15 20:11:42 just the ones you wanna build Jul 15 20:11:49 setup is 1 time thing, almost always Jul 15 20:14:20 setup is always a one time thing, except when it's not. Jul 15 20:14:45 really, what dtzWill said :-) Jul 15 20:16:02 thanks Jul 15 20:16:11 elpollodiablo1: sticky keys looks good here Jul 15 20:16:42 nice I was just getting ready to build and install on my phone so once I get home it is ready for testing Jul 15 20:16:54 it looks like setup does toolchain, rootfs, and sb2 mapping file - you should almost never have to re-do those Jul 15 20:17:23 elpollodiablo1: gonna push to testing feed shortly, so you can just update :) Jul 15 20:17:39 dtzWill: what are you pushing? Jul 15 20:17:52 linuxjacques_: makes the modifiers sticky Jul 15 20:17:58 only effects phones/physical keyboards Jul 15 20:18:11 dtzWill: still cool though :-) Jul 15 20:18:17 yeah :) Jul 15 20:18:34 dtzWill: taking all the credit. :) Jul 15 20:18:41 elpollodiablo1: hmm? Jul 15 20:18:57 elpollodiablo1: i pulled a commit from months ago and applied to head, b/c it got lost in the shuffle Jul 15 20:19:11 from last night's backlog, is there currently a problem with xecutah/xserver/xterm on 1.4.5 ? Jul 15 20:19:12 but it _was_ originall contributed by someone, i have his email somewhere Jul 15 20:19:28 I know I was joking Jul 15 20:19:31 i asked him if he wanted credit at the time, he said no he just wanted to share with me/everyone else, he's happy having it working for himself Jul 15 20:19:32 oh Jul 15 20:19:33 lol Jul 15 20:19:39 :) Jul 15 20:19:51 take not-giving-credit seriously :) Jul 15 20:19:53 I dunno, we may have to ban you from the wiki again elpollodiablo1 Jul 15 20:19:59 :-P Jul 15 20:20:19 ha ha Jul 15 20:20:43 man I guess I need to watch my step around her Jul 15 20:20:45 here* Jul 15 20:21:45 * dtzWill pushes Jul 15 20:21:59 idk how autobuilder is going these days, presumably it'll be in the feeds in 15min-1hr Jul 15 20:22:15 i used to have to do a force-refresh on the ipkg page to get it to show up though, not sure if that's still req'd Jul 15 20:22:36 linuxjacques_: yes there is a problem with x stuff on 1.4.5 Jul 15 20:23:18 someone ping me if the update goes through and somehow doesn't owrk Jul 15 20:23:19 : Jul 15 20:23:20 :) Jul 15 20:24:20 I guess I'll wait a bit to try xecutah on my 1.4.5 pre then. Jul 15 20:29:56 I am heading out, once I get home I will check out the sticky key changes Jul 15 20:33:31 PSA of sorts, the sticky key stuff might have slightly unintended side-effects when using a bluetooth keyboard Jul 15 20:33:50 (namely making modifiers sticky, i think only 'shift' on bluetooth is handled correctly, and i don't remember testing it as much as i _think_ it _should_ be) Jul 15 20:34:14 heads up of sorts/public announcement so one of us remembers to address that down the road, hehe Jul 15 21:22:22 (it's in the testing feed, elpollodialo-awa and others, lmk how it goes :)) Jul 15 21:30:34 dtzWill: thanks for putting it in the test feed. I was having an issue with making xserver Jul 15 21:30:40 elpollodialo-awa: worky ?:) Jul 15 21:31:37 installing now Jul 15 21:33:41 sticky keys working pretty well on initial test Jul 15 21:38:05 all the keys that are displayed on the keyboard are working correctly Jul 15 21:43:38 hooray! they should :) Jul 15 21:43:46 but tyvm for testing so thoroughly :) Jul 15 21:46:51 I am testing the non visable ones and is there going to be a new wiki page for this or should I just add it to the old Xterm page? Jul 15 21:49:00 there's an xterm page? Jul 15 21:49:29 there's a new wiki? Jul 15 21:49:41 linuxjacques_: :) Jul 15 21:49:53 :-D Jul 15 21:50:14 * ka6sox wonders where the new wiki is...and if He has to deal with that too. Jul 15 21:50:52 ka6sox: it's broken! :( Jul 15 21:51:05 http://www.webos-internals.org/newwiki Jul 15 21:51:25 ( http://www.youtube.com/watch?v=JYc05gZFly0 ) Jul 15 21:52:37 it is a general X page http://www.webos-internals.org/wiki/Application:X Jul 15 21:57:44 (I don't know, i'm no good at such things) Jul 15 21:58:44 morning Jul 15 21:59:08 morning rwhitby Jul 15 22:00:19 dtzWill: the comments about needing xecutah relate the the apps pushed to the testing feed, not the underlying capabilities of the executables which will likely be packaged in various ways in the future. Jul 15 22:01:10 ah, makes sense :) Jul 15 22:04:24 and yes, xecutah is an initial "bare bones, don't do anything magical behind the user's back" app at the moment, and will get improved automated workflows once we actually have working flows for all the devices that we can automate. Jul 15 22:05:07 dtzWill: for 1.4.5, we need a way to start Xsdl from a C service which does not create a zombie process that the C service hangs on waiting for it to finish. Jul 15 22:06:21 either that, or some new execution architecture which works on all devices. Jul 15 22:08:10 rwhitby: oh, is that part of why things don't work on 1.4.5? Jul 15 22:08:50 that is the sole and complete reason why 1.4.5 doesn't work Jul 15 22:08:55 you can start xterm from the command line fine Jul 15 22:09:04 the only reason we don't just fork it off entirely (my scripts did an 'exec' on Xsdl instead of 'fork') is that when running from launcher, a forked process couldn't communicate with the system (wrong pid it would appear, and their ipc stuff relies ont hat) Jul 15 22:09:39 dtzWill: I tried forking the call to the script inside the service. that didn't work either. Jul 15 22:09:43 rwhitby: oh you /want/ to fork it? Jul 15 22:09:52 (running via 'forker' didn't do it?) Jul 15 22:09:57 correct Jul 15 22:10:02 all i know is that 'forker' was needed for 1.4.5 previously Jul 15 22:10:08 ...didn't xterm run just fine previously? Jul 15 22:10:13 I want a clean detached Xsdl process which does not hang the service. Jul 15 22:10:26 your service can't just fork it? Jul 15 22:10:41 i think i'm not understanding, and i'm sorry. Jul 15 22:10:42 tried that, didn't work. Jul 15 22:10:56 problem is the service is hanging waiting for spawned process. Jul 15 22:11:05 correct Jul 15 22:11:14 fork+exec doesn't work (at whatever location) doesn't work? Jul 15 22:11:22 all 'detaching' means is fork+exec, no? Jul 15 22:11:25 popen, system, fork, all hung the service such that it does not respond back to javascrpt Jul 15 22:11:41 did you try doing fork outside of the service? Jul 15 22:11:43 I'm happy for someone else to look at it. Jul 15 22:11:47 & doesn't work? Jul 15 22:12:02 or a fork of any child process hangs things? Jul 15 22:12:13 dtzWill: I tried forker on all lines in the xecutah.sh script Jul 15 22:12:14 linuxjacques_: unless i'm mistaken, that's just gonna cause the shell to do the same fork() Jul 15 22:12:35 rwhitby: i believe you, i'm just trying to understand. mostly b/c if we can't ever call fork then we clearly have an issue. Jul 15 22:12:36 :) Jul 15 22:12:49 i *do* remember somehow i couldn't fork, and that's why 'forker' was written. i don't remember the details. Jul 15 22:13:11 dtzWill: oh, it's quite possible I've just done something wrong. I really do want someone else to look at it and find my error (or not). Jul 15 22:13:39 rwhitby: nope, you're 100% responsible for testing on all devices and versions! Jul 15 22:13:40 ;) Jul 15 22:14:03 preferably I'd like the service to see the initial output of the xecutah.sh script, so it can report errors to the user, but I'm happy to trade that off for working on all devices. Jul 15 22:14:46 dtzWill: I'm a forking newbie, so it is quite likely I've stuffed up. Jul 15 22:14:46 ah yes, see here:http://redmine.webos-internals.org/projects/xlauncher/repository/revisions/cfecb29564b4e3c85cb117ed8f1eeffcd323582b Jul 15 22:15:06 lol Jul 15 22:15:17 apparently i ran into fork()'ing issues that a shell ran into (using '&' in the most clear way), which is why i wrote 'forker' Jul 15 22:15:21 which....somehow fixes taht. Jul 15 22:15:45 dtzWill: did you test that on 1.4.5 ? Jul 15 22:15:46 that quote needs to go somewhere "rwhitby says "I'm a forking newbie" Jul 15 22:16:02 rwhitby: that was the only version out at the time, as i recall. Jul 15 22:16:12 linuxjacques_: I chose my words carefully :) Jul 15 22:16:19 which why i keep talkign about it--that fixed some fork-related issue i had on 1.4.5 Jul 15 22:16:24 shrug, maybe it was different. Jul 15 22:16:44 http://git.webos-internals.org/preware/cross-compile/commit/?id=c2b16e250a08c423059c6f6734270ad10b1e79db Jul 15 22:16:59 but you really shouldn't have those issues if you're running it as root...? Jul 15 22:16:59 dtzWill: should be easy for you to test right now - just add the call to forker in the xecutah.sh script Jul 15 22:17:16 rwhitby: as soon as i get a 1.4.5 device in a state that i can test with, yes. Jul 15 22:17:18 I believe I tried it and it failed, but I'd really like someone else to replicate that. Jul 15 22:17:26 I have a 1.4.5 device Jul 15 22:17:28 rwhitby: absolutely understood. Jul 15 22:17:42 I can play in about 3 hours. Jul 15 22:17:49 but that was to get fork()'ing to work in a jail, hrrmm... Jul 15 22:17:51 "failed" = service hung due to zombie process Jul 15 22:18:18 is it a feature? Jul 15 22:18:49 maybe that doesn't make sense. Jul 15 22:19:00 errr zombie, huh? Jul 15 22:19:44 http://en.wikipedia.org/wiki/Zombie_process Jul 15 22:19:51 hehe i'm there too ^.^ Jul 15 22:20:09 so anyway, since 1.4.5 users have a working Terminal, I just pushed with 2.0.0 minimum last night. Jul 15 22:20:28 1.4.5 -> Terminal. 2.0.0+ -> Xecutah Jul 15 22:20:39 yeah, good call. Jul 15 22:21:16 so much for working on all webos devices ;) Jul 15 22:21:16 I think 2.0.1 is having the same issue though Jul 15 22:22:05 as issue as 1.4.5 with Xterm not loading Jul 15 22:22:21 apparently needing to call wait() is related to zombie-nss.. Jul 15 22:22:22 hrm Jul 15 22:22:59 where/why is wait called? Jul 15 22:24:05 no, pclose does an implicit wait Jul 15 22:24:35 where/why/when is pclose called? Jul 15 22:24:36 * linuxjacques_ reads "The Linux Programming Interface" Jul 15 22:26:01 fromthe wikipedia page: Jul 15 22:26:05 "On modern UNIX-like systems (that comply with SUSv3 specification in this respect), the following special case applies: if the parent explicitly ignores SIGCHLD by setting its handler to SIG_IGN (rather than simply ignoring the signal by default) or has the SA_NOCLDWAIT flag set, all child exit status information will be discarded and no zombie processes will be left." Jul 15 22:28:10 well forker() doesn't wait, but does just exit main--you'd think that'd make Xsdl an orphan though. Jul 15 22:28:15 * dtzWill is trying to see where the issue is Jul 15 22:29:10 relevant stuff starting on page 553 of tLPI Jul 15 22:29:55 What does "om nom" mean? Jul 15 22:30:11 "gesture+{c,v,x} don't work b/c the OS om nom's those key presses" Jul 15 22:30:17 eat Jul 15 22:30:19 Error454: oh it means 'eat' Jul 15 22:30:21 :/ sorry haha Jul 15 22:30:23 with pleasure Jul 15 22:30:33 ah ok Jul 15 22:31:03 (do we know what process is the zombie?) Jul 15 22:31:09 bah i shuld just find a 1.4.5 device Jul 15 22:31:10 brb Jul 15 22:31:14 So I guess the question is what an alternate mapping for ctrl-c might be. I wanted to get at least c, z and then w s for up/down Jul 15 22:31:51 and tab... Jul 15 22:34:28 control+i does what tab does generally Jul 15 22:41:21 the new sticky mapping has the arrow keys Jul 15 22:42:53 sym+r up ,sym+c down Jul 15 22:44:15 * dtzWill looks for a 1.4.5 device with battery that works Jul 15 22:44:22 (my pixi doesn't seem to wanna turn on... at all) Jul 15 22:47:48 dtzWill: is this problem in the xecutah c-service? Jul 15 22:47:58 I don't even see a fork() or an exec() in there Jul 15 22:48:05 how is it starting the child? Jul 15 22:49:35 linuxjacques_: current service does popen. I tried fork and exec also. Please try them too to replicate or disprove my results. Jul 15 22:49:46 ah I see popen() Jul 15 22:51:02 rwhitby: maybe a signal(SIGCHLD,SIG_IGN); before the popen() is in order? Jul 15 22:51:46 I can test later, but for now I'm still at work. Jul 15 22:53:03 it could be in the shell itself Jul 15 22:53:04 the script Jul 15 22:53:08 or in forker Jul 15 22:53:09 etc Jul 15 22:53:25 seeing _which_ process is the zombie is what i intend on trying once i get a device up Jul 15 22:54:12 dtzWill: BTW, glad you were able to do a feed version update overnight. Jul 15 22:54:27