**** BEGIN LOGGING AT Wed Oct 04 03:00:02 2017 Oct 04 06:13:55 JaMa: klog on Touchpad with unstable: https://bpaste.net/show/5b7c74e3132e Oct 04 06:14:03 Seems systemd acting up again Oct 04 06:51:15 Morning! Oct 04 07:23:15 bshah: Seems the ubports guys are checking the patches for Mako :) https://github.com/ubports/android_kernel_google_msm/pull/2 Oct 04 07:23:28 Tofe: You're happy with those 2 PR's ? Oct 04 07:27:24 Herrie|Laptop: I just don't understand why you needed to add the webViewItem property Oct 04 07:28:27 Tofe: You might be right Oct 04 07:28:31 Let me try without Oct 04 07:28:59 You're right without Oct 04 07:29:03 Should be fine Oct 04 07:29:05 Let me update that Oct 04 07:29:15 also there might be a more universal way to do it: if we add onUrlChanged in the webview, and update the useragent there, it might do all the scenarios Oct 04 07:30:17 Tofe: Probably, however for entered urls we do some logics we don't need to do for the bookmark & history urls Oct 04 07:30:27 Might rework that though when I add the https-everywhere bits Oct 04 07:30:36 ok, well, as is it's fine too Oct 04 07:31:04 merged Oct 04 07:32:24 Ah I was cleaning up commits Oct 04 07:32:30 But well ;) Oct 04 07:32:47 ah, oops :) Oct 04 07:32:59 Well we now have 2 instead of 1 Oct 04 07:33:04 Not end of the world Oct 04 07:33:08 Let me clean it up quickly Oct 04 07:33:12 that's bearable Oct 04 07:35:41 Yup LOL Oct 04 07:37:57 Tofe: You reckon the rotation on Hammerhead will be easy to fix or should just push release out? Oct 04 07:39:13 Herrie|Laptop: frankly I don't really know; but I don't think it's a release-blocking issue Oct 04 07:40:50 Tofe: OK Oct 04 07:40:59 Let me build the release images today then Oct 04 07:41:09 Tofe: I came across this yesterday: https://doc.qt.io/qt-5/qtwebengine-platform-notes.html Oct 04 07:41:26 "On Linux, the kernel has to support the anonymous namespaces feature (kernel version >= 3.8) and seccomp-bpf feature (kernel version >= 3.5). Setuid sandboxes are not supported and are thus disabled. To explicitly disable sandboxing, the QTWEBENGINE_DISABLE_SANDBOX environment variable can be set to 1 or alternatively the --no-sandbox command line argument can be passed to the user application executable. Oct 04 07:41:35 Not sure you were aware of that Oct 04 07:41:46 But I guess we're OK now with the patches that we have already probably? Oct 04 07:47:00 Herrie|Laptop: we use more detailed switches, we don't entirely disable the sandboxing Oct 04 07:47:28 Tofe: Yup I'm aware of that Oct 04 07:47:36 I guess it's better to have some sandboxing then none ;) Oct 04 07:52:45 yep :) Oct 04 08:13:54 Herrie: what was the name of previous release and the name of current one? Oct 04 08:14:08 Herrie: I'll add tags to webos-ports-setup Oct 04 08:17:19 ah it's already building, nevermind then Oct 04 08:23:43 JaMa: I'm happy to tag if you tell me how to ;) Oct 04 08:31:46 git tag -a -m "release name" "release name" git-sha Oct 04 08:31:54 git push origin "release name" Oct 04 08:36:28 is Decaf current release name? Oct 04 08:36:43 I'll do it, let me fix one of staging scripts first Oct 04 08:38:44 JaMa: Yes Decaf Oct 04 09:10:42 ok, tags for old releases pushed: Oct 04 09:10:44 2014-12-05-030-blacktie Oct 04 09:10:45 2015-01-06-031-breve Oct 04 09:10:45 2015-02-03-032-americano Oct 04 09:10:45 2015-04-13-033-aulait Oct 04 09:10:47 2015-05-08-034-bombon Oct 04 09:10:49 2015-06-06-035-cubano Oct 04 09:10:52 2015-08-17-036-crema Oct 04 09:10:54 2015-09-07-037-olla Oct 04 09:10:57 2015-10-06-038-latte Oct 04 09:10:59 2015-12-09-039-marocchino Oct 04 09:11:02 2016-01-13-040-medici Oct 04 09:11:04 2016-02-07-041-miel Oct 04 09:11:07 2016-03-03-042-milk Oct 04 09:11:09 2016-05-12-043-mocha Oct 04 09:11:12 2016-06-16-044-tobio Oct 04 09:11:14 2016-08-09-045-touba Oct 04 09:11:17 2016-09-15-046-zorro Oct 04 09:11:19 2016-10-13-047-ca Oct 04 09:11:22 2016-12-25-048-cappuccino Oct 04 09:11:24 2017-02-21-049-chailatte Oct 04 09:11:27 2017-05-01-050-coldbrew Oct 04 09:11:29 Herrie|Laptop: from now on please tag after the release is closed (the staging feed closed and merged to public) Oct 04 09:11:32 and create 2 tags, one with just release name Oct 04 09:11:53 and 2nd with the date of "merged to public feed" as shown on 2nd to last line in info.XXX file Oct 04 09:12:20 followed with -XXX number of the staging feed and then release name as shown in examples above Oct 04 09:13:15 I'm not completely sure about the closing date of americano and ca releases, because they probably weren't closed correctly as the info file didn't have the lines from closing command Oct 04 09:33:32 JaMa: Thnx Oct 04 09:46:56 JaMa: I guess we miss Cortado still? Oct 04 09:47:44 Should be info.051 Oct 04 09:47:46 strange I've missed it http://build.webos-ports.org/releases/cortado/info.051 Oct 04 09:48:08 ah I've created cortado, but forgot to create the 2nd tag for it Oct 04 09:48:34 JaMa: Yeah looks like it Oct 04 09:48:37 fixed Oct 04 09:49:41 will drop stable-2015-09-10 tag as it is the same as olla release Oct 04 09:50:56 btw: touba release was incorrectly tagged, so I've updated tag for it as well Oct 04 10:01:37 JaMa: Thnx for all the help Oct 04 10:01:44 You seen my log for the unstable? Oct 04 10:01:50 Seems it hangs right away with systemd Oct 04 10:01:56 Could be we need further kernel patches Oct 04 10:40:23 I've seen the link but haven't read it yet Oct 04 10:40:34 now testing qemux86 and rpi3-64 from local unstable build Oct 04 10:40:46 rpi3-64 graphics should work now with latest meta-raspberrypi Oct 04 10:42:00 JaMa: That's nice on the rpi :) Oct 04 10:42:14 qemux86 failed for me, probably mesa stuff Oct 04 10:42:23 Got stuck quickly when I tried last week Oct 04 11:01:39 JaMa: Oct 04 12:59:18 qemux86 luna-next[507]: FATAL: 12:59:18.709: Could not initialize egl display Oct 04 11:01:48 Failed by a Oct 04 12:59:18 qemux86 LunaWebAppManager[505]: CRITICAL: 12:59:18.717: Failed to create display (No such file or directory) Oct 04 11:01:52 Then luna-next crashes Oct 04 11:04:40 for the hammerhead log this looks more suspicious to me <27>systemd[1]: Failed to determine whether /sys is a mount point: Bad file descriptor Oct 04 11:05:16 but dunno why, can you please paste current pyro log to compare it with this one? Oct 04 11:06:05 we might drop https://github.com/webOS-ports/meta-webos-ports/commit/6a8f2dac8ddac44df4d7e170f56e623df98ee552 now as it was fixed upstream Oct 04 11:11:39 JaMa: That log was from Tenderloin Oct 04 11:11:49 Since I could easily get it with klog Oct 04 11:11:52 From webOS side Oct 04 11:12:09 But I suspect Hammerhead will have something similar, just harder to debug Oct 04 11:13:19 for qemux86 and rpi UI I've tried to enable kms and gbm in qtbase, but haven't verified if it helps other than checking log.do_configure to make sure it was really enabled Oct 04 11:14:11 Herrie|Laptop: and btw you again started stable builds while testing pipeline was still adding new jobs, so might run out of ram again Oct 04 11:14:34 JaMa: Ah sorry forgot about that Oct 04 11:41:16 JaMa: Stable built unusually quick for so many changes. Is it reusing sstate from testing as well? Oct 04 12:12:19 yes Oct 04 12:12:36 it's actually the same sstate-cache directory symlinked from both Oct 04 12:12:46 JaMa: OK :) Oct 04 12:34:43 rpi3-64 luna-next start seems to get stuck on reading settings: Oct 04 12:34:45 #0 0x0000007fa538c740 in __GI___libc_read (fd=3, buf=0x2a670688, nbytes=16384) Oct 04 12:34:48 at /usr/src/debug/glibc/2.26-r0/git/sysdeps/unix/sysv/linux/read.c:26 Oct 04 12:34:51 #1 0x0000007fa580a0d0 in ?? () from /usr/lib/libQt5Core.so.5 Oct 04 12:34:53 #2 0x0000007fa582915c in ?? () from /usr/lib/libQt5Core.so.5 Oct 04 12:34:56 #3 0x0000007fa57b93ac in QFileDevice::readData(char*, long long) () from /usr/lib/libQt5Core.so.5 Oct 04 12:34:59 #4 0x0000007fa57c2348 in QIODevicePrivate::read(char*, long long, bool) () from /usr/lib/libQt5Core.so.5 Oct 04 12:35:02 #5 0x0000007fa57c2a7c in QIODevice::readLineData(char*, long long) () from /usr/lib/libQt5Core.so.5 Oct 04 12:35:05 #6 0x0000007fa57b9ff8 in QFileDevice::readLineData(char*, long long) () from /usr/lib/libQt5Core.so.5 Oct 04 12:35:08 #7 0x0000007fa57c3170 in QIODevice::readLine(char*, long long) () from /usr/lib/libQt5Core.so.5 Oct 04 12:35:11 #8 0x0000007fa57c3560 in QIODevice::readLine(long long) () from /usr/lib/libQt5Core.so.5 Oct 04 12:35:14 #9 0x0000007fa5f4443c in Settings::identifyHardware() () from /usr/lib/libLunaSysMgrCommon.so.3 Oct 04 12:35:17 #10 0x0000007fa5f454d8 in Settings::Settings() () from /usr/lib/libLunaSysMgrCommon.so.3 Oct 04 12:35:20 #11 0x00000000004039c0 in _start () Oct 04 12:35:25 you do know pastebin exists right? Oct 04 12:36:30 yes Oct 04 12:36:46 it might be missing rpi3-64 config like rpi3 has Oct 04 12:36:48 ./meta-luneos/recipes-webos/luna-sysmgr-conf/luna-sysmgr-conf/raspberrypi3/luna-platform.conf Oct 04 12:42:24 nope, it's not missing because raspberrypi3 is included in MACHINE_OVERRIDES Oct 04 13:13:32 JaMa: Luna-next config maybe? Oct 04 13:14:29 I've found some missing dtb changes in meta-rpi-luneos, trying it now Oct 04 13:14:33 then I'll re-check luna-next Oct 04 13:15:37 yes, the luna-next-conf looks correct Oct 04 14:36:21 Herrie, I'm curious: why do we need to change the user agent for the bookmarks pane of the browser? Oct 04 14:48:50 DougReederPhone: In general not, but we do need it for history items and these are populated in the same list when autocompleting. We could exclude it for bookmarks but I guess if a user visits a site, bookmarks it, expects it to work the next time too Oct 04 15:15:29 So it's not something in our code that needs to be updated, it's because websites haven't been updated? Oct 04 15:42:10 DougReederPhone: Well it's a bit different Oct 04 15:42:38 We want to "trick" certain sites by sending a different user agent to make them behave in a better way. Oct 04 15:43:14 I.e. Netflix doesn't like our useragent and wouldn't work. So we impersonate being on ChromeOS for example Oct 04 15:43:39 Same with web.whatsapp.com, it doesn't like tablet/mobile in useragent because it will tell you to download whatsapp app Oct 04 15:43:56 If you send Chrome + Desktop you can use web.whatsapp.com :P Oct 04 15:46:11 So we have a default useragent which will work on most sites, but it some cases for some sites we want to send something else to "trick" them into getting the results we want. Not ideal but well Oct 04 15:46:39 I.e. we want to force a site to be displayed like we're on Android or iOS or on a Desktop Oct 04 15:47:21 We can specify this once per site and this will then be used every time this site is loaded (i.e. entered, history or bookmarks. i guess we should cover via launchParams too Oct 04 16:26:34 So, we have to change our code several places, to deal with direct navigation, bookmarks, etc? Oct 04 16:37:31 If so, that sounds like a great place for some refactoring. Unless it's a 1-line change? Oct 04 16:38:32 1 line change Oct 04 16:38:38 But refactoring planned Oct 04 16:39:19 * DougReederPhone nods Oct 04 16:41:39 Still have plans to incorporate https anywhere Oct 04 16:42:33 Implementation for this is at affected locations Oct 04 16:43:06 Yeah, that you only want to have one set of code for. Oct 04 16:45:21 I gues it's not yet practical to assume a hostname typed into the URL bar is HTTPS. Too many sites that are broken for HTTPS, and it's not easy for software to detect that. Oct 04 16:53:42 Though, it would be a good exercize to see if heuristics help. Oct 04 16:55:24 For example, if connecting to https://hostname.com gets *any* HTTP error, instead of displaying it, fallback to http://hostname.com Oct 04 17:01:33 ... but HTTPS Everywhere is probably the most functional solution. Oct 04 17:03:07 Perhaps it would be useful to have a setting that prohibits unencrypted connections, remapping URLS to HTTPS if HTTPS Everywhere doesn't. Oct 04 17:03:18 ... but that would be a low priority. Oct 04 17:07:12 DougReederPhone: Yeah that's the problem, doesn't work anywhere with http Oct 04 17:07:14 +s Oct 04 17:07:26 So https-everywhere has a whole database which we already convert to sqlite 3 Oct 04 17:07:35 We just need to hook it up, intercept URL and rewrite Oct 04 17:09:12 MartinHov_: Could you give http://build.webos-ports.org/luneos-stable-staging/wip/images/qemux86/ a quick go? Oct 04 17:09:20 See if everything works as expected? Oct 04 17:16:05 Herrie: the whole stable is built ? Oct 04 17:21:36 Herrie: I'm downloading that qemux86 image now .. Oct 04 17:36:08 MartinHov_: Thnx Oct 04 17:36:22 Herrie: stable qemux86 052-279 works as well as the latest testing-0-482 image. HW KB works after reboot. (first you need to focus the input) mouse+ wifi works fine. System Update is up to date! = better than testing image :) all good so far Oct 04 17:36:22 I'll check Hammerhead & Tenderloin Oct 04 17:37:21 Tofe/DougReederPhone: Can any of you test the Mako? Oct 04 17:40:47 Not until tonight, at the earliest Oct 04 17:41:17 DougReederPhone: OK, Tofe might get to it earlier since it's evening for him ;) Oct 04 17:41:26 It's in the "stable" directory? Oct 04 17:41:26 latest qumux86 images have no Search preferences (no Google or alike) - but perhaps this was never available .. Oct 04 17:41:40 MartinHov_: Where exactly? Oct 04 17:42:33 Herrie: no Search preferences in Settings app. Oct 04 17:43:03 so AFAIK - just time will not use (Google etc) to search anything .. Oct 04 17:44:50 MartinHov_: We provide some via luna-init with Google as default AFAIK Oct 04 17:45:16 Herrie: eg. testing-0-479 has Google as default. but not the newest images .. Oct 04 17:45:34 I just verified this on 0-479 Oct 04 17:45:37 Weird Oct 04 17:45:44 Neeed to check that Oct 04 17:46:03 We didn't really touch that anywhere Oct 04 17:46:08 Recently Oct 04 17:46:14 Herrie: I confirm it's already tonight here Oct 04 17:46:44 Tofe: DougReederPhone: everything was built already and it's in http://build.webos-ports.org/luneos-stable-staging/wip/ Oct 04 17:47:41 downloading Oct 04 17:47:51 testing image 0-480 and 0-481 both also have Google as default .. Oct 04 17:49:19 0-482 and 052-279 have "Not set" and UI has no option to change it. Oct 04 17:51:05 BTW - in older testing images once I change the Search to other selection I can not undo and change back to Google (it is not in the list) Oct 04 17:53:46 I just rebooted 052-279 image for 3 or 4 time and now the Search Preference is Google and there is selection. (like with other images ). This is not a showstopper. Just -reboot a few times. :) Oct 04 17:54:26 Who wants to test Whatsapp with me ? Oct 04 17:54:29 That might a little issue with configurator Oct 04 17:54:32 PM a number :P Oct 04 17:54:47 Herrie: I don't have that app :p Oct 04 17:55:05 Whatsapp on LuneOS ??? Oct 04 17:55:13 do we have that ? Oct 04 17:55:24 MartinHov_: web.whatsapp.com ;) Oct 04 17:55:42 Same like on desktop ;) Oct 04 17:55:59 ok - I have WA on Nexus 4 Android .. and Win 10 in browser .. Oct 04 17:56:16 Herrie: ah but wait, I think I still have a sim in that N4 I'm flashing... Oct 04 17:56:33 I just don't remember what number it was Oct 04 17:56:59 Herrie: we can try WA .. ? Oct 04 17:57:01 Tofe: LOL Oct 04 17:57:21 MartinHov_: PM me number Oct 04 17:58:06 I guess it uses webRTC ? Oct 04 17:59:03 Tofe: not sure Oct 04 18:00:34 mmh on N4 at 1st boot we have sensorfwd at 100% CPU for quite some time Oct 04 18:00:44 Tofe: hmmz Oct 04 18:01:04 I have flickering N5 screen in browser Oct 04 18:01:21 Black, display, black, display Oct 04 18:01:29 oh no, we still have the /system/bin/ files that don't have +x Oct 04 18:01:40 Randomly not always Oct 04 18:03:06 Herrie: you see my private chat ? Oct 04 18:03:07 ok, well, that's not a great stable here, we'll have to make some adjustements Oct 04 18:03:35 Tofe: Yeah weird about Mako Oct 04 18:11:58 ah: "run.do_install.6288: [[: not found" Oct 04 18:15:06 ok, now I'm sure it'll work Oct 04 18:16:39 JaMa: https://github.com/shr-distribution/meta-smartphone/pull/36 ? Oct 04 18:17:59 Herrie: we also get the sensorfwd issue on mako Oct 04 18:23:38 Tofe: Do you have rotation? Because I do have working rotation on TP last time I checked Oct 04 18:24:07 BTW web.whatsapp.com works in LuneOS browser, I can record voice too :-P Oct 04 18:24:29 I'll flash this image shortly to test Oct 04 18:24:38 ah, great Oct 04 18:43:45 Tofe: Will flash TP now Oct 04 18:43:53 To check sensorfwd en sensor_test Oct 04 19:14:57 Tofe: working rotation on Tenderloin: test_sensors: https://bpaste.net/show/80f050414135 Oct 04 19:26:20 BTW - and I have been able to hear the audio from Herrie which he sent from WA on LuneOS to my Android WA. Audio recording works in the browser ! :) Oct 04 19:40:57 JaMa: ^ Can you merge the meta-smartphone one? Oct 04 19:46:32 Tofe: And full Tenderloin: https://bpaste.net/show/141ba41c6c94 Oct 04 19:48:41 Herrie: thanks, it'll help knowing which errors are the critical ones Oct 04 19:51:13 THere seem to be less libhybris one Oct 04 19:51:29 +s Oct 04 20:09:22 merging it in a sec Oct 04 20:10:32 btw you don't need to rewrite it to if/then/fi, just using single brackets in dash works fine Oct 04 20:12:14 JaMa: well, I didn't know that :) Oct 04 20:13:04 merged with the modification Oct 04 20:14:35 triggering new stable build for mako Oct 04 20:15:48 JaMa: Thnx Oct 04 21:31:29 DougReederPhone: If you could test this one would be great! http://build.webos-ports.org/luneos-stable-staging/wip/images/mako/luneos-dev-package-mako-20171004201603-stable-052-112.zip Oct 04 23:24:55 BTW - For me in LuneOS WA the audio listening is broken using the browser in latest stable qemux86 image. But well I found that on youtube videos the audio is getting better after 1-2 minutes of video .. strange .. **** ENDING LOGGING AT Thu Oct 05 03:00:01 2017