**** BEGIN LOGGING AT Tue Jan 31 02:59:58 2012 Jan 31 03:23:41 PuffTheMagic: just pushed a patch to fix opengl error handling in special weird cases :) Jan 31 03:24:36 and gn8 :) Jan 31 03:24:39 night Jan 31 03:26:12 was the problem with phones that key events didn't fire/bubble when the plugin had focus or something? Jan 31 03:27:05 the problem was the events received in the C++ code were all 0's Jan 31 03:28:14 did it still work with bt keyboards or was that true of bt keyboards + phone? Jan 31 03:28:28 i didnt test bt on the phone Jan 31 03:28:46 I would expect it would suffer from the same problem Jan 31 03:49:05 http://ompldr.org/vY2pubg/wterm_2012-30-01_194752.png Jan 31 04:17:36 PuffTheMagic. when you did dispatch event and you got all 0s, did you make sure to check that the event was ok before dispatching it/function parameters? Jan 31 04:18:01 check that it was ok? Jan 31 04:18:09 onkeydown can sometimes be (context, keyevent) or just (keyevent) Jan 31 04:18:27 so if you were instead passing the context object to dispatchEvent... Jan 31 04:18:37 that would be too good to be true though :( Jan 31 04:18:52 i doubt that was it Jan 31 04:19:54 rwhitby, i just pushed an alpha with phone support Jan 31 04:20:03 well, anything running webos 2.2+ and enyo Jan 31 04:22:17 i just enabled allow allowKeyboardFocus Jan 31 04:22:22 and how I get some funkyness Jan 31 04:22:52 yeah, I think arrow keys and stuff won't work then/we've changed things Jan 31 04:22:55 it's hard to keep up :( Jan 31 04:23:50 arrow keys work currently Jan 31 04:42:06 Brybry, still messing with it Jan 31 04:44:16 I'm thinking if I can think of a better way to handle it Jan 31 04:45:00 i wish you had a phone to test on too Jan 31 04:48:42 it should be relatively the same as long as I assume that dispatchEvent/plugin translating the event is broken Jan 31 04:54:13 options I've been thinking of: a testDispatch() at the start to set behavior, quick and dirty stop propegation/handle shift/alt/ctrl modifiers, catch when a bt keyboard is connected and set behavior that way, hmm Jan 31 04:55:31 Brybry: in recognition of your contributions to wTerm Jan 31 04:55:39 lol, thanks Jan 31 06:34:15 Brybry, i just added mini vkb layout for the TP when using bt that provides function keys and escape Jan 31 06:37:09 Brybry, im thinking i might be able to get dispatching working again if I create a clean event and populate wit fields from event that blows up the stack Jan 31 06:39:08 see that's just weird Jan 31 06:40:40 well in webos3.x there might be something lower level clears up the even Jan 31 06:40:48 i will paste here in a min what happens on the pre3 Jan 31 06:53:33 Brybry, the double events are because of how i named some functions :( Jan 31 06:54:08 mmm? because it was catching for the plugin and then bubbling up to document.blah? Jan 31 06:54:33 basically Jan 31 07:10:16 mmm preferences text inputs Jan 31 07:16:10 things are getting even weirder now wrt input on pre3 Jan 31 07:16:18 now that i have renamed things Jan 31 07:17:04 [20120131-02:16:51.347351] error: Uncaught RangeError: Maximum call stack size exceeded., app.html:1 Jan 31 07:17:05 yay Jan 31 07:21:53 I think we'll have to move away from document.onkeydown/onkeyup and do it for the wTermApp kind and I guess plugin or else we're going to mirror input from preferences text inputs to the terminal Jan 31 07:22:39 or we could check what has focus ;) Jan 31 07:23:09 but i actuall think i have a way to not use document.* now Jan 31 07:47:54 damit Jan 31 07:48:08 that stupid cross-hair is showing up and i cant remember how to make it go away Jan 31 08:00:16 verified that you can do a (at least with the bt keyboard), wTermApp:ApplicationEvents.onKeyup/onKeydown handler and a Terminal:keydownHandler/keyupHandler (I guess you could do application events for that too...) Jan 31 08:00:53 dont think we need to do anything wTermApp Jan 31 08:01:37 to catch non-plugin-focused input but I could be wrong Jan 31 08:02:11 i just pushed some changes Jan 31 08:02:16 give them a shot Jan 31 08:02:28 i get this on the TP now Jan 31 08:02:29 [20120131-03:00:56.674641] error: Uncaught RangeError: Maximum call stack size exceeded., app.html:1 Jan 31 08:02:43 which is what i was getting on the Pre3 Jan 31 08:02:53 but input from BT keyboard works Jan 31 08:02:57 does does vkb Jan 31 08:03:00 and slider kb Jan 31 08:03:03 and there are no repeats Jan 31 08:03:15 so the only difference is that im not using applicationEvents at all Jan 31 08:03:44 i must be modifying the events some Jan 31 08:05:25 Brybry: i might have also figured out the stuck key issue Jan 31 08:06:02 need to have onmouseout send keyup Jan 31 08:10:40 give focus to the vkb and then press a key on your bt keyboard and it shouldn't be caught Jan 31 08:10:48 (aka press a vkb key) Jan 31 08:11:48 ? Jan 31 08:12:26 like with current head if I type on the virtual keyboard then bluetooth input no longer happens Jan 31 08:13:42 it should work Jan 31 08:14:02 plugin.keyup sends focus to trm Jan 31 08:14:25 so it should leave bt in usable state Jan 31 08:16:30 oooh, hmm Jan 31 08:16:42 might be a != typo in keydown Jan 31 08:17:17 possibly im getting tired Jan 31 08:17:22 it's all good :) Jan 31 08:18:49 hmm, no Jan 31 08:18:51 that didn't fix it Jan 31 08:18:51 damn Jan 31 08:21:40 the dispatchEvent is kind of redundant there as well/it's not needed Jan 31 08:22:40 thought u introduced it Jan 31 08:23:52 here's how things originally worked: bt input when the plugin area had focus was fine but when the vkb keys had focus it was broken Jan 31 08:24:30 so I caught non-plugin area key events (wTermApp), passed those to the plugin area, and set focus on the plugin area (the last bit may have been redundant as well as the event being redirected might have set focus) Jan 31 08:26:12 so if you typed on the vkb and then on a bt keyboard it caught 1 bt key event, passed that on, and then everything was handled transparently by luna/webkit Jan 31 08:32:42 hmm Jan 31 08:39:00 I think I know how to do it Jan 31 08:58:38 grrr arrow keys Jan 31 08:58:40 why do you hate me so Jan 31 09:00:43 ya there are like 3 sets of sym codes for arrow keys Jan 31 09:05:02 I'm looking to see if we moved stuff from handleKeyboardEvent to fakeKeyEvent Jan 31 09:08:49 lol, yeah that last commit :D Jan 31 09:22:15 I think this works: https://github.com/Brybry/wTerm/commit/26badcd39680eb5ed89cfd3b235dfc008266a876 Jan 31 09:23:24 and no more call stack size error stuff Jan 31 09:29:08 moin Jan 31 09:29:26 Brybry: why name it btKeydown/btKeyup? Jan 31 09:29:59 because it's 3am and I will get into a 15 minute 'what is the best name for this function' debate with myself Jan 31 09:30:13 and so I named it the first thing that came to mind that wasn't 'foo' or 'bar' Jan 31 09:30:22 :D Jan 31 09:30:32 and it should really only fire from bt key presses Jan 31 09:30:33 well, i have no idea whether it works Jan 31 09:30:38 but it looks totally wrong^^ Jan 31 09:30:58 I just dunno if the phone stuff still works then Jan 31 09:31:07 you forward the keyEvent to the plugin node when versionMajor == 3 Jan 31 09:31:21 but the plugin node only uses it if versionMajor != 3 Jan 31 09:31:31 yeah, that's on purpose Jan 31 09:32:06 != 3 is PuffTheMagic's pre3 keyboard input code Jan 31 09:32:08 i just don't get why that would work :) Jan 31 09:32:35 because bt input automagically works when the plugin has focus Jan 31 09:32:41 you just have to catch that 1 key input when the plugin doesn't have focus and forward Jan 31 09:32:45 aka after typing on the virtual keyboard Jan 31 09:32:57 so shouldn't you focus first and then forward? Jan 31 09:33:10 maybe Jan 31 09:33:41 yeah Jan 31 09:33:43 that's how I did it the first time :D Jan 31 09:33:47 and i still doubt that forward will do anything; i don't think the key event can be routed back to the plugin the "native" way from js Jan 31 09:33:48 https://github.com/Brybry/wTerm/commit/d487176a3c312c642c9e97b8b13c46e6e44d3f1c Jan 31 09:34:12 the rerouting definitely works Jan 31 09:34:29 that's how I fixed the focus issues originally a few weeks ago Jan 31 09:34:40 k Jan 31 09:35:45 ah, and i guess the event handlers should return true or something like that to stop further processing (fixing duplicate key events) Jan 31 09:36:29 in this case they don't need to but I don't really know why Jan 31 09:40:24 when I did logging plugin:keydownHandler() didn't seem to bubble to wTermApp:btKeydown Jan 31 09:42:50 i think the jailer doesn't start the plugins Jan 31 09:42:54 on my pre- 2.1 Jan 31 09:54:30 and strace can't follow the jailer, damn -.- Jan 31 09:54:48 child_stack=0x9e89e9b0, flags=|SIGCHLD) = 2776 Jan 31 09:54:48 syscall: unknown syscall trap 0x0f000000 Jan 31 09:54:48 [pid 2776] ????( Jan 31 09:54:48 Process 2776 detached Jan 31 10:07:09 ok. so i did workaround the jailer (replaced it with a shell wrapper) Jan 31 10:07:10 [20120131-11:06:12.229125] error: Uncaught plugin not ready, /usr/palm/frameworks/enyo/1.0/framework/build/enyo-build.js:10,064 **** BEGIN LOGGING AT Tue Jan 31 17:23:32 2012 Jan 31 17:54:14 Brybry, you here? Jan 31 17:54:22 yes sir Jan 31 17:56:09 good morning Jan 31 17:57:18 see PM Jan 31 17:57:22 kk Jan 31 18:59:45 stbuehler, so you are saying i need to return true in these even handlers? Jan 31 19:13:02 stbuehler, why is the jailing running on your device Jan 31 20:27:33 re Jan 31 20:28:03 PuffTheMagic: it might have prevented the duplicate key events (i didn't follow your discussion closely, so i'm just throwing in some ideas) Jan 31 20:28:23 PuffTheMagic: and the jailer gets started everytime, just with newer webos versions it detects the NO_JAIL thing Jan 31 20:28:58 if i run the jailer commad directly in novaterm (to start wterm), it prints an useless error and exits Jan 31 20:29:09 so i guess it doesn't recognize the option Jan 31 20:33:11 stbuehler, try no_jail Jan 31 20:33:17 lower case worked pre webos 3.0 Jan 31 20:34:37 next time :) but i worked around the jailer, so there are other problems to fix too Jan 31 20:36:45 stbuehler, i believe i fixed the stuck key issue :D Jan 31 20:36:55 gotta listen to onmouseout too Jan 31 20:36:59 and set up on that Jan 31 20:37:21 perhaps you should do "real" multi-touch stuff :D Jan 31 20:37:28 i am Jan 31 20:37:38 i listen to ontouchstart/end Jan 31 20:37:47 oh, nice Jan 31 20:37:55 but here is the problem Jan 31 20:38:09 if you dont use them, and you just use mouseup/odnw Jan 31 20:38:31 then if you are currently pressing A and while that is down you press Z, Z does not register Jan 31 20:38:46 hmmm shit Jan 31 20:38:51 i just broke key combos Jan 31 20:38:53 :( Jan 31 20:39:03 can use on mouseup Jan 31 20:39:14 err Jan 31 20:39:15 mouseout Jan 31 20:39:40 damit Jan 31 21:07:32 hm, screen resizing/changing rotation has some weird effects Jan 31 21:14:54 on phone or TP? Jan 31 21:15:16 stbuehler, i changed the code to not use hardcoded heights Jan 31 21:15:26 and there are def some after affects to that that need to get straigntened out Jan 31 21:15:30 straightened Jan 31 21:18:03 na, it's about the screenbuffer Jan 31 21:18:16 i'm not sure whether to insert new lines or reuse lines from the scrollbuffer Jan 31 21:18:40 it say reuse lines Jan 31 21:18:57 this is new territory here, not like xterm rotates much Jan 31 21:18:58 looking at the console behaviour Jan 31 21:19:07 well, you still can resize it :) Jan 31 21:19:14 true Jan 31 21:19:20 the problem is not the rotation Jan 31 21:19:27 konsole add new empty lines Jan 31 21:19:39 and it remove lines from the bottom until it hits the cursor Jan 31 21:20:12 i'm gonna try that Jan 31 21:22:18 PuffTheMagic: btw, are there some details for #51? Jan 31 21:22:55 vttest Jan 31 21:23:13 menu 2 Jan 31 21:23:17 the scrolling tests Jan 31 21:23:22 compare them to xterm Jan 31 21:23:48 k Jan 31 22:00:47 PuffTheMagic: ok, xterm erases the complete screen, even if smaller margin is set. Jan 31 22:01:25 kde konsole too Jan 31 22:01:35 is that all that was missing? Jan 31 22:02:13 well, i found only one difference; on one screen, wterm showed more lines with text (soft scroll stuff) than xterm Jan 31 22:03:58 the erase function don't mention any behavior with margin Jan 31 22:06:26 are you sure its erase and not remove Jan 31 22:07:28 decstbm(first, last);ed(2); Jan 31 22:07:42 line 404 in vttest/main.c Jan 31 22:08:43 well you said xterm, but now you are pointing to vttest so im confused Jan 31 22:10:46 well, i have no idea whether what xterm does is correct Jan 31 22:11:35 i'd day xterm is our best exemplar Jan 31 22:11:35 but it erases the whole screen on ESC[2J, regardless of margin Jan 31 22:11:44 xterm fails on menu 1 Jan 31 22:11:53 wterm fails menu 1 after menu 2 Jan 31 22:12:11 ya I need to fix that wterm weirdness after menu 2 Jan 31 22:12:20 it has to do with the restoring term state at the end Jan 31 22:13:09 stbuehler, are you using the vttest from wterm dir when you test xterm (the one that sets row,cols automatically) or are you using the orig where you need to set them your self Jan 31 22:13:45 i use the wterm binaries Jan 31 22:13:50 ok Jan 31 22:13:54 although i didn't pull that checkout for some time Jan 31 22:14:28 well at some point I had it figure out the actual term size and use that as defaults if the user didnt set sizes Jan 31 22:14:40 i know many of the tests fail if the rows/cols dont match Jan 31 22:31:02 where is that user comng from Jan 31 22:33:43 uhm Jan 31 22:33:51 if you don't know it? Jan 31 22:34:19 i mean, i know the box it has to be on to have to domain Jan 31 22:34:28 but i havent used irc on it in forever Jan 31 22:34:38 so idk why it just started showing up recently Jan 31 22:35:38 :D Jan 31 22:35:51 oO Jan 31 22:36:28 i cant believe its almost Feb and I its like mid 50s out and i have by back door open Jan 31 22:36:39 how did you generate vttest/config.h? Jan 31 22:36:54 it's freezing here Jan 31 22:37:14 whatever this crazy fahrenheit value is Jan 31 22:46:08 stbuehler, i just ran configure on my desktop Jan 31 22:46:17 and used that Jan 31 22:51:00 i guess it could be a vttest bug (menu 1 failing after menu 2) Jan 31 22:51:05 it doesn't reset the modes Jan 31 23:03:30 ah, tab stop issue Jan 31 23:17:05 Brybry, ping Jan 31 23:20:14 PuffTheMagic: can i push some stuff? Jan 31 23:21:44 dont have to ask me that Jan 31 23:22:07 go for it Jan 31 23:23:46 äöö working Jan 31 23:25:13 PuffTheMagic: perhaps you can reset #51 now, to see whether i missed something Jan 31 23:27:07 PuffTheMagic: we have two resizes at startup now (just so you know). i don't think it is really slow, so shouldn't be a real problem Jan 31 23:31:06 stbuehler, u cant close tickets as a contributor? Jan 31 23:34:01 i can Jan 31 23:34:09 i just wanted to know whether it really is fixed :) Jan 31 23:34:25 oh Jan 31 23:34:32 1 sec i will test Jan 31 23:34:55 stbuehler, does scrolling look right in irssi with TERM=xterm Jan 31 23:37:39 hm Jan 31 23:37:57 no Jan 31 23:48:49 stbuehler, u fixed menu 1 after menu 2, Jan 31 23:48:56 but the tab test in menu 2 is always wrong now Jan 31 23:49:20 in particular tab resetting is now broke Jan 31 23:56:51 hm Feb 01 00:03:12 ok, i think irssi scrolling works now here Feb 01 00:05:07 git diff Feb 01 00:05:09 ups Feb 01 00:07:20 :D Feb 01 00:31:14 hm Feb 01 00:31:21 linux eats the tabstop Feb 01 00:31:29 so we only see spaces Feb 01 00:31:51 or echo converts them Feb 01 00:33:37 "linux" what? it shouldn't. Feb 01 00:50:00 I can believe that Luna eats the tabstop Feb 01 01:03:02 oh god Feb 01 01:03:41 stbuehler, scrolling in irssi looks good, mc is broken horribly Feb 01 01:04:13 only on xterm though Feb 01 01:04:21 under rxvt it looks good Feb 01 01:11:06 tab stops should be fixed now Feb 01 01:11:24 what is wrong in mc? works here - scrolling through /var Feb 01 01:11:56 the line drawing chars go away Feb 01 01:12:16 and some of the sgr stuff doesnt get processed Feb 01 01:12:43 weird Feb 01 01:12:45 works this time Feb 01 01:12:50 what did I do that made it look like shit Feb 01 01:13:32 hm, but i remember i hade some "poor man" style graphics some time ago too Feb 01 01:14:45 i also dont get color in emacs sometimes Feb 01 01:15:51 hm Feb 01 01:16:21 do you get segfaults sometimes too, like cmatrix crashing from wild rotating? Feb 01 01:17:00 umm, havent noticed Feb 01 01:17:08 maybe once cmatrix crashed in exhibition mode Feb 01 01:17:17 it restarted after starting Feb 01 01:17:23 not sure it segfaulted Feb 01 01:18:03 the main problem is that you don't get back to the shell somehow Feb 01 01:18:17 the shell does start reading the input again, but it doesn't do anything Feb 01 01:18:29 in exhibition mode? no shell is run Feb 01 01:18:35 it launches cmatrix directly Feb 01 01:18:44 no, in normal mode Feb 01 01:18:55 just kill -SEGV $(pidof cmatrix) from novaterm Feb 01 01:24:25 gn8 :) Feb 01 01:24:44 night Feb 01 01:45:42 it segfaults if you SEGV it =] Feb 01 01:51:20 https://github.com/PuffTheMagic/wTerm/issues/84 Feb 01 01:58:04 Brybry, where art thou **** ENDING LOGGING AT Wed Feb 01 02:59:57 2012