**** BEGIN LOGGING AT Wed Nov 05 03:00:00 2014 Nov 05 07:57:17 GranPC: ping Nov 05 08:06:35 morning! Nov 05 08:26:57 morning Nov 05 08:27:34 morphis: Any suggestions for debugging Enyo 1 apps that seem OK in journalctl but don't appear in web inspector and don't launch? Nov 05 08:30:52 Herrie|Veer: yeah Nov 05 08:31:05 need to check with --verbose for LunaWebAppManager Nov 05 08:31:18 then there is something wrong with the load mechanism Nov 05 08:31:25 Herrie|Veer: which apps? Nov 05 08:32:53 Communities Enyo 1 app Nov 05 08:43:36 Herrie: where can I get the ipk for it? Nov 05 09:22:13 Has anyone success running webos-virtualbox-emulator which is build with dizzy branch ? I build one with dizzy-2014-11-03 and find I can't run enter in the rootfs. It seems like vmdk image bug in dizzy branch, since I have build core-image-minimal vmdk image in dazzy and the virtualbox works fine. Nov 05 09:25:31 error msg as here: VFS: Cannot open root device "sda2" or unknown-block(0,0): error -6 Nov 05 09:33:23 Find the solution, Change disk type from 'IDE' to 'SATA' in virtualbox. I think we need to modify the ovf file. Nov 05 09:46:40 coldnew: possible, we're still working on getting the dizzy branch to life Nov 05 09:46:49 only mako is tested and known to work so far Nov 05 09:46:58 if you want something usable use daisy Nov 05 10:13:27 morphis: pong Nov 05 10:16:18 I've gotta update the kernel config verify script Nov 05 10:16:32 it's missing a few lxc thingies Nov 05 14:58:17 morphis, coldnew: I've built qemu yesterday, and was stuck at boot time Nov 05 15:02:26 Tofe: Can you try to see if you can replicate the issue with Enyo 1/2 not launching after first use on 1st boot? Reboot solves it here, but seems LNC might need some tweak to fix it. Nov 05 15:02:35 According to morphis this was fixed but seems broken for me still. **** BEGIN LOGGING AT Wed Nov 05 15:23:06 2014 Nov 05 15:36:56 Herrie|2: on daisy ? Nov 05 15:38:22 I had such bug yesterday on N4 yes; I didn't know if the daisy build included the fix though, so I didn't look into it too much. I'll have a closer look tonight Nov 05 15:39:05 Frankly, I'd be surprised that the bug isn't solved, as the fix today is simply to force LNC to be restarted... Nov 05 15:39:58 Herrie, can you please update http://issues.webos-ports.org/issues/721 with your experiences? Simon apparantly can’t reproduce it. Nov 05 16:05:45 Dougreeder: added Nov 05 16:05:51 Thanks Nov 05 16:06:36 Tofe: Not sure according to morphis fix should be in Black Eye release, but isn't for me.. Either it isn't or not working properly... Nov 05 16:13:31 Herrie , do you know if the wallpapers directory is *supposed* to be in /media/internal ? Nov 05 16:14:15 I have a vague recollection of one of the folders that MediaIndexer know about was somewhere else entirely. Nov 05 16:16:37 Not 100% sure but that's where it was in legacy and would be the most logical place imho Nov 05 16:18:52 Maybe it's not there Nov 05 16:19:00 Seems like we have a single image Nov 05 16:19:11 Still would make sense to have it there though... Nov 05 16:20:03 Anyone got a dev environment setup that could compile a custom luna-webappmanager for me? Nov 05 16:21:45 Herrie: I do but it's in the middle of a build at the moment :p Nov 05 16:29:52 so happy to see another webos user in the office here. he has TWO touchpads in his office. one has a bad charging circuit though. :( Nov 05 16:34:50 thrrgilag: No rush :P Nov 05 16:46:44 Any ETA on your build? Nov 05 16:58:04 Looks like it's going down in flames Nov 05 16:58:07 :/ Nov 05 17:00:09 Hmmz ;) Nov 05 17:00:32 Doh! forgot to update luna-next Nov 05 17:00:32 I guess I need to setup my own build environment sometime too :P Nov 05 17:00:56 Just normally don't touch c++ stuff but recently seem to run into it too often:P Nov 05 17:01:03 i was still sourcing a my copy pre dizzy Nov 05 17:02:09 just waiting for qtwebkit to finish and then i can reset and try your modifications Nov 05 17:02:50 Yeah just need a new arm binary for my N4 to test Nov 05 17:02:55 It's a 1 line change :P Nov 05 17:10:52 Herrie|Veer: i can try to build too Nov 05 17:11:58 For N4 ? Nov 05 17:12:13 yes Nov 05 17:14:40 Great :) Just pushed my code to http://github.com/herrie82/luna-webappmanager Nov 05 17:15:17 ok Nov 05 17:15:50 What I'm trying to do is to get Communities app working.. It's headless (noWindow set to true) but currently not included in our list of allowed headless apps. So would like to have this binary to test if the app works then or there are other issues still Nov 05 17:15:50 :P Nov 05 17:16:04 So code is nasty but should work at least for now Nov 05 17:18:04 I the restriction on noWindos apps new to LuneOS? Nov 05 17:18:20 DougReeder: Yes Nov 05 17:18:37 We have a small list in webappmanager.cpp that are currently allowed to run Nov 05 17:18:50 Rest is basically blocked. Sure morphis can give the rational Nov 05 17:19:15 I think it's memory usage but not sure Nov 05 17:19:40 There are valid reasons for an ordinary apps to be noWindow - generally, either for a scheduled activity like a sync, or an action using launchParam. Nov 05 17:20:10 DougReeder: Yeah I guess so. Not sure how many use it though Nov 05 17:20:19 Would need to check my IPK collection Nov 05 17:20:40 We can extend the list in general or just allow them but that's a design decision Nov 05 17:20:55 I don't remember the reason why we don't currently Nov 05 17:21:06 it may be required to allow dockMode to work as intended. Nov 05 17:21:29 We don't have a dockmode currently either ;) Nov 05 17:21:34 Heh. Nov 05 17:21:38 Would need implementation at some time :P Nov 05 17:21:55 It was part of old LunaSysMgr and not in LunaNext yet afaik Nov 05 17:23:19 Herrie|Veer: hm, something wrong with my arm compiler. need some time to understand what Nov 05 17:23:47 OK ;) Nov 05 17:24:20 DougReeder: We basically went away from BrowserServer/BrowserAdapter model for webkit1 due to move to webkit2 Nov 05 17:24:42 I’m not familiar with that. Nov 05 17:24:44 However our webkit2 implementation might need some rework due to the memory usage issues we have Nov 05 17:25:47 That might have been more of an issue for Mojo apps. Enyo 1 has a different relationship between cards and the JavaScript window global. Nov 05 17:26:10 See isis-project.org Nov 05 17:26:28 And then the part about isis browser and b2g compare Nov 05 17:26:37 It has some background Nov 05 17:28:59 Basically owo isis = qt 4.8 = webkit 1. Nov 05 17:29:30 We're using QT 5+ which has webkit2 so different rendering model Nov 05 17:30:20 And lots less customization ;) Nov 05 17:30:34 So easier to upgrade stuff :) Nov 05 17:35:13 That’s a good thing. Nov 05 17:35:18 does anyone have luneos running on their nexus 4 here? Nov 05 17:35:28 would be real happy if they could check something for me Nov 05 17:35:48 Got Black EYe, sure Nov 05 17:36:06 DougReeder: can you try running /system/bin/getprop Nov 05 17:36:12 does it segfault or does it spew the android properties? Nov 05 17:36:34 returns nothing, doesn’t segfault Nov 05 17:36:47 try /system/bin/getprop ro.build.fingerprint Nov 05 17:36:49 i think it should return something Nov 05 17:37:04 returns a blank line Nov 05 17:37:11 eh Nov 05 17:37:13 whatever, at least it doesn't segfault Nov 05 17:37:20 gotta look into that :P Nov 05 17:37:53 DougReeder: I think generally if an app has to do any background activity it should write a service for that and synchronize with the app when it's active Nov 05 17:39:12 GranPC: /system/bin/getprop will not work outside of the lxc android container Nov 05 17:39:25 is it meant to segfault tho? Nov 05 17:39:40 why are you using it outside the container? Nov 05 17:39:44 use /usr/bin/getprop instead Nov 05 17:39:50 was trying to check if the property service was running Nov 05 17:40:07 yep ok that seems right Nov 05 17:40:22 GranPC: /usr/bin/getprop comes from libhybris Nov 05 17:40:46 it first tries to connect with the property service and if that doesn't work it falls back to accessing the plain buildprop files Nov 05 17:41:16 if /usr/bin/setprop works and you can see the result with getprop then you're fine Nov 05 17:41:24 i think i've found the issue Nov 05 17:41:29 bash-4.3# /usr/bin/test_hwcomposer Nov 05 17:41:29 test_hwcomposer: test_hwcomposer.cpp:157: int main(int, char**): Assertion `err == 0' failed. Nov 05 17:41:34 ala Nov 05 17:41:35 bash-4.3# setprop test 1234 Nov 05 17:41:35 bash-4.3# getprop test Nov 05 17:41:35 1234 Nov 05 17:41:49 GranPC: that doesn't mean much Nov 05 17:42:01 luna-next is crashing out because of hwcomposer tho (i think?) Nov 05 17:42:12 you know where exactly it is crashing? Nov 05 17:42:12 Nov 05 17:33:44 flo luna-next[1505]: FATAL: 17:33:44.165: QPA-HWC: hw_get_module(HWC_HARDWARE_MODULE_ID, (const hw_module_t **)(&hwc_module)) in create returned -2 Nov 05 17:42:33 i'm seeing some google hits for this from the people working on sailfish Nov 05 17:42:59 on the flo? Nov 05 17:43:09 yeah Nov 05 17:43:13 well not for flo Nov 05 17:43:34 morphis that is certainly the preferred architecture for apps designed for webos 2 and later, where services are available Nov 05 17:44:29 GranPC: what does /system/bin/logcat say? Nov 05 17:44:44 GranPC: and is the Android container running correctly? Nov 05 17:45:01 depending on the hwcomposer implementation there might be a service needed running to get it working Nov 05 17:45:11 http://paste2.org/DYJ2Z4Oh Nov 05 17:45:22 DougReeder: hm, but what is the advantage of the nowindow card if you can do such things in a service? Nov 05 17:45:36 yeah i'm pretty sure it's running properly, what's a good way to check? Nov 05 17:45:47 Apps designed for Firefox OS that do periodic sync have to be launched periodicallly. Allowing a similar mechanism under LuneOS allow those Firefox OS apps to be ported more easily. Nov 05 17:46:34 GranPC: that hardly depends on the device Nov 05 17:46:35 It’s probably fine to restrict noWindow apps for now, we just may need to revisit that in the future. Nov 05 17:47:12 DougReeder: for periodic sync a service should be responsible Nov 05 17:47:39 service doesn't require that much memory and doesn't run all the same where a nowindow card does Nov 05 17:48:00 Oh, certainly a service will give better performance. Nov 05 17:49:24 that is my concern Nov 05 17:49:42 and the user does not have any control over the nowindow cards Nov 05 17:49:54 And there isn’t a similar noWindow setting on Firefox OS, so they may not expect it. Nov 05 17:50:36 but one thing we can improve is printing out error messages when a nowindow app is launched but not allowed to Nov 05 17:51:23 I think it’s probably fine to stick with restricting noWindow apps for now. But to welcome web app developers, we may need more flexibility at some point. Nov 05 17:53:06 One reason Palm & HP failed to attract more developers was a lack of support for apps using frameworks other that Mojo and Enyo. Nov 05 18:38:02 morphis: will do a PR for that Nov 05 18:38:12 GOt that ready: re better error logging Nov 05 18:38:35 Just need someone to compile it for me since i don't have an environment so I can test on N4 ;) Nov 05 18:41:56 morphis: got a tad further now Nov 05 18:42:01 wayland is dying tho Nov 05 18:42:13 how could I debug that? Nov 05 18:47:07 Herrie: just got my build system cleared up... did you want me to do a build with your webappmanager change? Nov 05 18:54:28 thrrgilag: yes would be great! Nov 05 18:54:53 Need to setup my own env sometime though, downloading ubuntu already ;-) Nov 05 18:56:07 mako build right? Nov 05 18:56:29 yes pls Nov 05 18:56:56 So I can test some legacy apps on Mako and propose changes where needed to the luna-webappmanager ;-) Nov 05 19:12:23 Herrie: build in progress, i'll let you know when it's done Nov 05 19:12:33 :) Nov 05 19:23:49 Herrie: a PR for the error message? Nov 05 19:24:30 DougReeder: so what would you vote for? Nov 05 19:24:33 enable noWindow apps? Nov 05 19:25:01 basically I don't want to block anything but we need to care about the system staying in a healthy state Nov 05 19:26:50 DougReeder: for #721, you expect the wallpapers dir to be present? Nov 05 19:29:12 morphis: I'll add some debugging Nov 05 19:29:19 Setting up my build env now Nov 05 19:30:02 morphis: Shouldn't we expect it? Nov 05 19:30:05 I would ;) Nov 05 19:30:11 GranPC: what do you mean by "wayland is dying"? Nov 05 19:30:13 But well we only have one currently ;) Nov 05 19:30:21 morphis: journalctl says connection to wayland lost Nov 05 19:30:26 managed to debug it but I have no idea where to go from here Nov 05 19:30:29 http://paste2.org/ZkK7HfXc Nov 05 19:31:07 so the client is dying? Nov 05 19:31:09 or the compositor? Nov 05 19:31:22 I'm really not sure, journalctl said it lost connection to the compositor Nov 05 19:31:34 but testing stuff I can see the client dies too Nov 05 19:31:51 ok Nov 05 19:31:53 Nov 05 18:38:29 flo maliit-server[2201]: WARNING: The Wayland connection broke. Did the Wayland compositor die? Nov 05 19:31:55 start with the compositor first Nov 05 19:32:02 how can I debug that? Nov 05 19:32:13 that comes because the compositor dies Nov 05 19:32:17 or stops Nov 05 19:32:44 yeah I guessed that much Nov 05 19:32:54 is there any way to attach gdb to the compositor right after it gets started Nov 05 19:33:04 (or maybe just run it under gdb directly) Nov 05 19:35:21 it is Nov 05 19:35:25 systemctl stop luna-next Nov 05 19:35:39 cp /etc/luna-next/environment.conf ~/env.conf Nov 05 19:35:58 then edit env.conf and put a export for each line and put the value for LUNA_NEXT_OPTIONS into " Nov 05 19:36:03 then Nov 05 19:36:08 . ~/env.conf Nov 05 19:36:15 export XDG_RUNTIME_DIR=/tmp/luna-session Nov 05 19:36:26 luna-next $LUNA_NEXT_OPTIONS Nov 05 19:36:57 from that point you can also start it with gdb Nov 05 19:37:58 aha awesome Nov 05 19:38:39 morphis: Started: http://webos-ports.org/wiki/Pre_and_Post_Release_To_Do Nov 05 19:40:07 morphis: welp. http://paste2.org/dhHD980I Nov 05 19:45:18 GranPC: that one is fine Nov 05 19:45:20 just press c Nov 05 19:45:24 and hit enter Nov 05 19:45:33 gdb stops there as openssl is performing some arch detection Nov 05 19:45:38 oh Nov 05 19:45:41 which issues unknown asm instructions Nov 05 19:45:45 that looks like one shitty way to do arch detection Nov 05 19:45:46 lol Nov 05 19:46:06 oh man, no info Nov 05 19:46:17 http://paste2.org/t2307dg2 Nov 05 19:46:58 GranPC: next thing you can do is a Nov 05 19:47:02 export LUNA_NEXT_DEBUG=1 Nov 05 19:47:45 not a lot of extra info Nov 05 19:47:49 but hey, at least it's picking up the resolution Nov 05 19:47:55 good Nov 05 19:48:06 GranPC: next step will be interesting: Nov 05 19:48:09 find out where it crashes Nov 05 19:48:15 when gdb is running Nov 05 19:48:18 in another terminal Nov 05 19:48:31 cat /proc/`pidof luna-next`/maps Nov 05 19:48:41 then see in which library the crash happens Nov 05 19:48:55 if it is in one of ours, we have luck Nov 05 19:49:15 if it is in the one in /system it will be harder to solve the problem Nov 05 19:49:46 what if I told you it's in /system Nov 05 19:49:47 and it's closed source Nov 05 19:49:50 /system/lib/hw/hwcomposer.msm8960.so Nov 05 19:50:06 ah Nov 05 19:50:10 no it's not :) Nov 05 19:50:13 it's not? Nov 05 19:50:20 awesome Nov 05 19:50:24 thankfully Nov 05 19:50:30 praise qcom for that .) Nov 05 19:50:44 https://android.googlesource.com/platform/hardware/qcom/msm8960/ Nov 05 19:51:02 ah no Nov 05 19:51:09 isn't that just the headers? Nov 05 19:51:14 https://android.googlesource.com/platform/hardware/qcom/display/+/android-4.4.4_r2.0.1/msm8960/ Nov 05 19:51:18 that is the right one Nov 05 19:51:26 oh man Nov 05 19:51:27 that's awesome Nov 05 19:51:28 just need to find your right android version tag Nov 05 19:51:34 now comes the more awesome part Nov 05 19:51:41 we can load the android symbols in gdb :) Nov 05 19:52:16 GranPC: http://paste2.org/DyLLnKb9 Nov 05 19:52:26 paste all the generated lines into gdb Nov 05 19:52:33 but before you have to install the symbols Nov 05 19:53:09 where do I get them from? Nov 05 19:53:21 from your cm-wop-11.0 build Nov 05 19:53:39 remember "tar cjf cm-wop-11.0-`date +%Y%m%d`-0-.tar.gz system symbols android-ramdisk.img filesystem_config.txt" Nov 05 19:53:49 aha Nov 05 19:53:50 the symbols directory is what you need Nov 05 19:54:02 it's in another package so not installed by default Nov 05 19:54:28 install android-system-image-mako-dbg from tmp-eglibc/deploy/ipk/flo/ Nov 05 19:54:37 then /system/symbols should appear Nov 05 19:54:55 you have to load the symbols before you run luna-next in gdb Nov 05 19:55:09 GranPC: another way would be to add log statements in the hwcomposer module code Nov 05 19:55:25 GranPC: hope that are enough pointers for the moment :) Nov 05 19:55:32 yep, awesome Nov 05 19:55:36 I had no idea hwcomposer was open source actually Nov 05 19:55:42 that'll be huge Nov 05 19:55:44 thanks!!! Nov 05 19:56:01 GranPC: someone really needs to document those things .. Nov 05 19:56:15 i really want to update a lot of things in the wiki after i figure out what i'm doing Nov 05 19:56:38 i'm pretty clueless though Nov 05 19:56:46 i've mucked around with android a whole lot, but never got a traditional linux setup on it Nov 05 19:56:52 :P Nov 05 19:56:59 hehe Nov 05 19:57:10 it's a interesting ride for each target again Nov 05 19:57:20 will never just-work Nov 05 19:57:31 yeah I expected it'd be tricky Nov 05 19:57:39 considering the sailfish people still haven't got their OS running on flo Nov 05 19:57:48 yeah Nov 05 19:58:01 maybe once you get LuneOS running they will have Sailfish running too :) Nov 05 19:58:14 GranPC: the ubuntu guys have? Nov 05 19:58:19 yup Nov 05 19:58:25 i actually tried ubuntu touch on my tablet Nov 05 19:58:30 pretty sure everything worked Nov 05 19:59:02 GranPC: maybe https://code-review.phablet.ubuntu.com/gitweb?p=aosp/device/asus/flo.git;a=shortlog;h=refs/heads/phablet-4.4.2_r1 has some hints Nov 05 20:00:29 I'm pretty sure they got Sailfish running, but maybe without the phone/radio functionalities Nov 05 20:01:02 last I checked they didn't have display at all or something Nov 05 20:01:18 oh that works now, nice Nov 05 20:01:49 ( https://wiki.merproject.org/wiki/Adaptations/libhybris ) Nov 05 20:02:23 basically they have the screen and the usb port :) Nov 05 20:02:41 well that's one more thing than I have for now :P Nov 05 20:09:38 Morphis, for #721, I expect sideloading the OS to re-create /media/internal/wallpapers if it doesn’t exist and copy the standard wallpapers to it. Nov 05 20:13:04 On noWindow apps, there were several use cases - exhibition mode, syncing and “desk accessories” were the main ones, IIRC. I think it’s fine to block them for now, and wait to see if one of those use cases re-emerges. Nov 05 20:16:49 Herrie: webappmanager failed on the build, gimmie a sec looks like it should be a quick fix Nov 05 20:31:05 Herrie: webappmanager is ready, how can i send it? Nov 05 20:37:27 Herrie, looks like something went bad with my build... good thing nizovn got it for you. :) Nov 05 20:40:22 thrrgilag: i also had issues with compiling, error: cannot open crt1.o: No such file or directory... Nov 05 20:43:16 DougReeder: why? Nov 05 20:43:27 DougReeder: was that on legacy the same? Nov 05 20:43:53 Sorry, #721 or noWindow? Nov 05 20:44:28 721 Nov 05 20:45:56 How else are the standard wallpapers to reach their home, if sideloading doesn’t re-create them? Nov 05 20:46:41 they are stored in /usr/share/wallpapers Nov 05 20:46:43 morphis: btw 673 and 577 can be closed Nov 05 20:47:20 Then, the media indexer should be told about /usr/share/wallpapers Nov 05 20:47:31 DougReeder: it does it from it :) Nov 05 20:49:38 The core problem I’ve experienced, is that I sideload, go to Setting. try to change the wallpaper, and the file selection has no idea there are wallpapers in /usr/share/wallpapers Nov 05 20:50:14 If the wallpapers are in /usr/share/wallpapers, and the Media Indexer knows that, all is well. Nov 05 20:51:19 In webOS, the standard wallpapers are copied into /media/internal/wallpapers when you doctor, IIRC. Nov 05 20:57:32 DougReeder: I think keeping them in a central place is the better option Nov 05 20:57:34 DougReeder: you should see them in the media inde Nov 05 20:57:36 s/inde/index/ Nov 05 20:58:44 After my latest sideload, they appear to be there, and the Media Indexer appears to know of them (but not have thumbnails). Nov 05 20:59:16 hm, thumbnails were there too Nov 05 20:59:19 On earlier sideloads, the Media Indexer had no clue they existed, so on could not select them as wallpapers Nov 05 21:24:29 FWIW, the argument for copying the standard wallpapers to /media/internal/wallpapers is that there they can be a living collection that the user can add to in USB mode. If the standard wallpapers are in /usr/share/wallpapers, the user can’t modify that, and so might create another directory in /media/internal. Nov 05 21:25:02 But I’m cool with either location, so long as the Media Indexer knows about it right away. Nov 05 21:33:19 nizovn: Thnx any easy way to install this ipk? Nov 05 21:33:55 Herrie: opkg install ** Nov 05 21:34:18 I would normally extract the /usr/sbin/LunaWebAppManager and adb push it ;) Nov 05 21:34:22 Ah :) Nov 05 21:36:20 :) Nov 05 21:37:45 Thnx Nov 05 21:37:50 Will give it a go shortly :) **** ENDING LOGGING AT Thu Nov 06 03:00:00 2014