**** BEGIN LOGGING AT Tue May 26 02:59:57 2020 May 26 06:29:30 Good morning! Do calls already work with the Xiaomi Mi A1? Asked this question a long time ago, maybe things have changed... Thanks May 26 06:47:33 windknecht: Almost... Seems some other projects got it working with some Xiaomi devices, so we should be able to get it to work also, but so far not yet May 26 06:47:46 We were busy with some more deep underlaying upgrades on the system recently May 26 07:58:20 Morning! May 26 08:00:22 Herrie: you're still talking about qml, right ? May 26 08:00:47 I didn't think we had that many LS2 calls in qml yet May 26 08:01:18 Tofe: Well it seems the side effect of moving away from the LsReqisterPubPriv to LSRegister is that also the JS bits throughout the system that make calls to com,palm.systemservice won't work anymore... May 26 08:02:42 Ah, ok, that's annoying yes May 26 08:03:16 So I have about 40 files all scattered around the rootfs both QML, JS and binaries that now stopped working as a result ;) May 26 08:03:41 Are there a lot of called services to take into account ? or mostly systemservice ? May 26 08:04:13 because it'll be a hardcoded list I think; we don't have access to the new names in our qml adapter May 26 08:04:39 (ls-hubd can afford to try multiple variants, but we can't) May 26 08:04:47 I mean, we should be registering them similarly as the PubPriv did with multiple variants May 26 08:04:56 Or maybe I misunderstood May 26 08:05:12 I thought the LsRegisterPubPriv registered it with multiple names, while LSRegister doesn't? May 26 08:27:20 morning May 26 08:32:07 Morning May 26 08:33:54 Tofe: Some housekeeping PR's: https://github.com/webOS-ports/webos-keyboard/pull/91, https://github.com/webOS-ports/org.webosports.app.firstuse/pull/40, https://github.com/webOS-ports/luneos-components/pull/97 May 26 08:40:55 Herrie: no, PubPriv only registers one name; but the code of ls-hubd (pointed to yesterday) does some legacy migration help with trying different new names when old names are called. So if com.palm.toto is called from an app which registered itself with PubPriv, then ls-hubd will try com.webos.service.toto, org.webosports.service.toto, etc May 26 08:48:10 But what's strange is that JS bits are broken by this luneos-component PubPriv change... do you have an example of JS code that now triggers an error? May 26 08:52:16 Herrie: is your build of qemu of yesterday noon including the modified luneos-component ? May 26 09:01:40 Tofe: It would basically be everything that calls com,palm.systemservice for example: https://github.com/webOS-ports/core-apps/blob/webOS-ports/master/com.palm.app.email/data/EmailAccount.js#L491 May 26 09:02:15 but I don't see how this would involve luneos-components May 26 09:02:37 Isn't systemservice exposed by luna-sysservice ? May 26 09:03:41 It would make more sense if this was related to luna-webappmanager for instance May 26 09:04:44 I reverted the luneos-components in my last build let me put it back again May 26 09:09:30 luna-webappmanager, on its side, is still doing a PubPriv registration May 26 09:09:41 so all our webapps are legacy May 26 09:10:19 (note: migrating that one will be tricky !) May 26 09:10:30 Seems LG is still busy with that at their end May 26 09:10:41 So yes May 26 09:12:00 Ideally, we would have ls-hubd handle a migration table (somehow) and on our side all our code would do LSRegister May 26 09:12:22 Tofe: Yeah that would be great May 26 09:14:13 Anyway. I still don't understand how luneos-components broke a luna-webappmanager->luna-sysservice call May 26 09:17:20 Let me just try again with my latest image May 26 09:25:15 Tofe: It just seems that calls to com.palm.systemservice fail, while the ones to com.webos.service.systemservice succeed May 26 09:25:50 I updated FirstUse to call to com.webos.service.systemservice, but there are many other places that still call to com,palm.systemservice May 26 09:27:44 Herrie: btw, is your meta-webos-ports pushed somewhere ? I could reproduce this on my side and debug ls-hubd to see what it does (or doesn't do) in GetServiceRedirectionVariants May 26 09:28:42 https://github.com/webOS-ports/meta-webos-ports/commits/herrie/zeus-acg-qt5.15-large May 26 09:28:50 I could also share my qemu image May 26 09:29:03 For the QML apps i'm not surprised, and I can migrate them or put a migration table in luneos-components. It's the web apps that are a mystery May 26 09:29:20 Well let's do both, I'll have to build that someday anyway May 26 09:30:26 I've rebased your branch on top of zeus, just for the sake of it May 26 09:30:55 what branches should I take elsewhere ? May 26 09:31:19 for meta-qt5 I guess it's master, or dunfell/something? May 26 09:34:27 It's recent testing, so yeah just what we have in testing basically May 26 09:34:53 I haven't updated testing in a long time :) May 26 09:34:59 let me do that May 26 09:35:58 Ah ok May 26 09:36:09 I took JaMa's from after Qt 5.15 update May 26 09:38:53 ok, build started May 26 09:39:25 Herrie: can you also provide me your luna-service2-dbg ipk package ? That way I'll be able to debug ls-hubd right away on qemu May 26 09:39:33 Ah ok May 26 09:39:34 Yes May 26 09:47:42 Tofe: Both on FTP May 26 09:51:41 looks like recent Thunar has some regressions with FTP access; but with ncftp it works well May 26 09:51:49 Actually that might be the wrong ones.... May 26 09:51:54 Let me put correct ones... May 26 09:52:21 :) ok May 26 09:55:23 It's corret actually May 26 09:55:29 My local .vmdk wasn't updated... May 26 09:55:32 Hence I was seeing funny stuff May 26 09:57:08 ok, fetching May 26 10:04:44 Tofe: Now it seems to behave better, just it seems I'm stuck in FirstUse mode somehow May 26 10:05:04 I.e. I don't get the gesture bar after FirstUse is finished May 26 10:05:15 So it seems that in cardshell it somehow doesn't switch May 26 10:17:44 wow, it's a mess yes May 26 10:25:47 Herrie: I'll propose an update for the luna-service2 patch May 26 10:26:25 Task (/home/hextreme/luneos/mi-env/webos-ports/meta-webos-ports/meta-luneos/recipes-upstreamable/nodejs/node-gyp-native_git.bb:do_install) failed with exit code '1' May 26 10:26:35 * BT40 says hello to all May 26 10:26:38 Herrie: basically, add "ret.push_back(std::string("org.webosports.service.") + name.get());" in the Forward compatibility section of hub.cpp, line 155 May 26 10:29:16 https://pastebin.com/V8fLG7ZB May 26 10:29:47 terminal is full of bloat. Cant scroll back much. How to access log? May 26 10:30:17 some bloat example is shown in above pastebin link May 26 10:31:44 Herrie: can you try https://github.com/webOS-ports/luna-service2/pull/9 ? not tested at all yet May 26 10:32:27 BT40: "cp: write error" May 26 10:32:49 means do i need tu run it as sudo? May 26 10:33:45 you have also full logs in tmp-glibc/work/x86-64/node-gyp/xxxxx/temp/log.do_install (path my vary a bit) May 26 10:33:51 may* May 26 10:34:09 no, sudo is never needed here May 26 10:34:21 you still have space left, right? May 26 10:35:03 73gb free in 165gb drive May 26 10:35:09 ok, mmh May 26 10:35:27 drive=partition. 500gb drive May 26 10:36:02 tmp-glibc/work/x86_64-linux/node-gyp-native/4.0.0+gitAUTOINC+7245415296-r0/temp/run.do_install.2194006:1 exit 1 from 'cp -R --no-dereference --preserve=mode,links -v /home/hextreme/luneos/mi-env/webos-ports/tmp- (rest is cut) May 26 10:36:29 I guess your partition is a classic ext4 fs ? May 26 10:36:39 yes May 26 10:36:46 ok, so it's not a symlink issue May 26 10:39:11 '/home/hextreme/luneos/mi-env/webos-ports/tmp-glibc/work/x86_64-linux/node-gyp-native/4.0.0+gitAUTOINC+7245415296-r0/image//home/hextreme/luneos/mi-env/webos-ports/tmp-glibc/work/x86_64-linux/node-gyp-native/4.0.0+gitAUTOINC+7245415296-r0/recipe-sysroot-native/usr/lib/node_modules/node-gyp/package.json' the "/image//home/" bit looks odd May 26 10:39:16 let me see what's in mine May 26 10:41:41 BT40: here's mine https://paste.ubuntu.com/p/YkrX3VXJVR/ the "image//home" seems to be normal after all May 26 10:42:55 BT40: can you try to just do "bb node-gyp-native" ? May 26 10:44:04 there are 2441 lines in log file. pastebin seems not optimal. here is file https://www.dropbox.com/s/avvopqj3k3f6c0t/log.do_install-nodejs-native?dl=0 May 26 10:44:38 right :) May 26 10:44:46 log file shared on dropbox. May 26 10:45:00 BT40: you can also try to execute cp manually, to see full error message May 26 10:45:33 thanks nizovn , will try just now May 26 10:45:36 Tofe: meta-qt5/dunfell is 5.14, 5.15 is only for master and it was released today :) May 26 10:47:57 Tofe: yes image/home/path-to-tmpdir is normal for native recipes May 26 10:50:31 cp Documents/irc-log Downloads/ May 26 10:50:36 Tofe: in about an hour I can try I guess May 26 10:50:49 above command works fine in terminal (new window opened for that) May 26 10:51:41 BT40: i mean to copy that long command that is failing to terminal May 26 10:52:04 Tofe: should I prepare final 5.15.0 for you to build (instead of another upgrade in couple days)? May 26 10:52:06 i have created/mounted separate home partition for extra space. Hope that might not be the issue May 26 10:52:52 JaMa: no hurry, I can rebuild it later on May 26 10:53:20 ok, will continue fighting firefox on dunfell May 26 10:53:24 :) May 26 10:53:29 Tofe: Do i need to specify land? Or just write directly this command as you specified May 26 10:54:39 BT40: it should be fine to just do the "cp -R --no-dereference --preserve=mode,links -v /home/hextreme/luneos/..........." command, all paths seem to be absolute paths May 26 10:59:39 cp -R --no-dereference --preserve=mode,links -v /home/hextreme/luneos/mi-env/webos-ports/tmp-glibc/work/x86_64-linux/node-gyp-native/4.0.0+gitAUTOINC+7245415296-r0/temp/run.do_install.2194006:1 May 26 10:59:53 have i typed correct command? May 26 11:00:07 seeking approval before running May 26 11:01:02 nope, wait May 26 11:02:36 the command (from your paste) is the following: https://pastebin.com/QRknm5uf May 26 11:02:40 oh. i didn't defined destination in above command. So dumb question by me :) May 26 11:04:24 Tofe, tunned in terminal. No error thrown. Although due to bload cant scroll back to see full May 26 11:04:31 tunned=runned May 26 11:09:52 Machine=land just started command MACHINE=land bb node-gyp-native May 26 11:09:58 will see what happens now May 26 11:10:47 Tasks Summary: Attempted 130 tasks of which 126 didn't need to be rerun and all succeeded. May 26 11:11:22 Tofe, does that means error has been resolved? May 26 11:11:56 shall i proceed with regular build command? May 26 11:15:29 yes, probably was random disk error May 26 11:15:32 BT40: Yes problem solved May 26 11:15:38 Tofe: Building now May 26 11:17:21 nizovn, same error occured previous time also while building. sorry i am newbie. can u explain a bit more so that i can learn May 26 11:17:23 Tofe: IPK's on server May 26 11:17:40 BT40: Sometimes random errors happen (i.e. memory corruption, disk error etc) May 26 11:17:43 what caused this error twice and how you guys sorted May 26 11:17:53 Usually a sign that there's some hardware issue May 26 11:18:09 I was having "internal compiler erorrs" and it turned out I had a faulty memory module May 26 11:18:19 Causing random build failures May 26 11:18:35 I ran memtest86+ and found out. Then started to narrow it down by testing memory in pairs May 26 11:18:47 And found that 1/8 modules had errors, other 7 were fine May 26 11:19:20 ok. How you guys sorted it? mine case i am referring...... Through experience? Or there was some clue in log? May 26 11:19:57 i mean, how Tofe hit idea to run cp manually? logic behind it? May 26 11:20:08 was it to see detailed error message? May 26 11:20:15 or some other logic? May 26 11:20:46 In this case I wouldn't blame disk yet, node-gyp (or nodejs built recipes in general) are failing with "cp: write error" randomly for long time (on otherwise reliable HW) May 26 11:21:31 BT40: it was nizovn who had the idea, but the logic is that you don't get all the errors output when looking at bitbake's log, so you can try the command directly and get more hints May 26 11:21:44 No pampering, just showing respect...... THIS IS WHY U GUYS DESERVE AND HAVE 5000 DOLLARS COMPUTERS. May 26 11:22:54 while a kid (not by age, by experience in front of u i am kid) is having 500 dollar one May 26 11:23:02 Herrie: looks like the little patch helped May 26 11:24:01 * BT40 says thanks to nizovn , JaMa , and of course Tofe & Herrie May 26 11:25:02 Tofe: yeah I have other issues now but know what to do May 26 11:25:09 Regarding webappmgr May 26 11:25:30 https://github.com/webosose/meta-webosose/commit/d2405678f6def2767129422ea022b00db20e44e6 May 26 11:26:59 porting had completed 55% (3500 tarks approx out of 6705)... Now it has resumed from 44% around 2500 packages? Do rest of work is lost? May 26 11:27:38 although it had shown during error that waiting for other tasks to complete. so why run again? May 26 11:29:19 i mean if bb completed rest of the tasks, why it fell back to 44%? May 26 11:30:35 are you using -k? May 26 11:30:35 BT40: regarding the error, this helps often: https://i.pinimg.com/236x/b3/0a/da/b30ada6cf007ad0eb2da9fa6ce594991.jpg May 26 11:37:41 JaMa, are u using -k.......I hope this isn't for me, seems discussion is among you guys May 26 11:38:39 * BT40 still wold like to know why it (bb) resumed from 44% instead of 55% May 26 11:42:07 BT40: that was for you :) -k is "keep going" for bitbake so it doesn't end after first error, but builds as much as possible before it quits May 26 11:43:33 no, i am not using -k. Simple command only, no other parameter. May 26 11:44:01 Still after error, bb didn't stopped abruptly. It completed 4 other tasks which were running parallel. May 26 11:44:21 It specificly said waiting for other tasks to conclude. May 26 11:44:32 yes, that's expected it will finish ongoing tasks but won't start more May 26 11:44:45 Thats why i wonder why it lost that info and started from back May 26 11:45:53 44 to 55% will cost at least an hour to me. Slow rig :) May 26 11:46:24 avg compile is 1% in 5-6 minutes. 11% difference will get 65 mins of delay May 26 11:47:45 it doesn't work that way, some tasks take 1s some take 2 hours May 26 11:48:20 yes, just an average i mean. May 26 11:56:52 BT40: usually it doesn't redo tasks just for annoying the user, there probably is a good reason somewhere May 26 12:02:45 BT40: JaMa has been building with Yocto for 10+ years, he knows his stuff :) May 26 13:16:29 Tofe: Seems that the call to luna:///com.palm.service.accounts/createLocalAccount '{}' doesn't complete May 26 13:17:53 ls-monitor gives me: https://paste.ubuntu.com/p/GjGxff4K4r/ May 26 13:18:07 That's why it never seems to finish first use, because it doesn't succeed in creating the local account May 26 13:21:56 var/log/messages tells me it fails on: luna://com.webos.service.connectionmanager/getStatus May 26 13:22:13 May 26 15:16:01 qemux86-64 user.warn ls-hubd: [] [pmlog] ls-hubd LSHUB_NO_SERVICE {} _LSHubSendQueryNameReplyMessage: Failed Connecting to Service err_code: -3, service_name: "com.webos.service.connectionmanager", unique_name: "(null)", static, fd -1 May 26 13:24:54 OK I think I might have a lead here... May 26 13:32:16 Ehm nope... May 26 13:40:09 Ok some other thoughts, trying May 26 13:53:36 OK got it working now... Seems that this is the culprit somehow: https://github.com/webOS-ports/luna-sysservice/commit/be50246857ae72ab570ee3b04577e5165b66af4a May 26 13:53:53 removed com.palm.systemservice from the role.json and .service and stuff started to magically work May 26 13:53:58 Just still don't know why :S May 26 13:55:26 I can indeed confirm that emails now show up right away instead of after a reboot, so that's an improvement May 26 14:23:41 Herrie: the reason might be that ls-hubd checks for allowedNames, but luna-sysservice doesn't actually register that name May 26 14:24:02 so it won't try to use the other migrated variants May 26 14:32:05 Tofe: we already saw it does that in the code May 26 14:32:19 Service name is hardcoded there, see yesterdays logs May 26 14:32:59 But the com.Palm call works now (I added some variants to your ls2 fix from earlier) not sure that made the difference though May 26 15:46:52 ok May 26 15:56:28 qtwebengine's build just finished, it shouldn't be long now May 26 16:00:14 Herrie: so to summarize: I remove com.palm.systemservice from the role.json, and I apply my patch on luna-service2 May 26 16:14:09 Tofe: yes and from the dbus. Service as well May 26 16:14:24 ok; it seems I'm missing a fix for firstuse, too? May 26 16:15:07 Which one? May 26 16:15:18 ls-hubd LSHUB_ROLE_FILE {} Can not find service "org.webosports.app.firstuse-*" permissions for executable "/usr/sbin/luna-qml-launcher" May 26 16:15:39 Hmm weird May 26 16:16:10 it's strange, I do have "org.webosports.*" for the qml launcher May 26 16:16:33 It could be you need to tweak the perms file for firstuse May 26 16:16:56 Add the variant without - not 100% sure May 26 16:17:13 You didn't have this error before? May 26 16:18:42 maybe I'm using an old qml launcher version, let me check May 26 16:23:27 Tofe: let me check after quick dinner May 26 16:23:32 sure May 26 16:26:41 oh, no... I patched luna-service2, which triggers a rebuild of qtwebengine... May 26 16:30:46 It didn't here May 26 16:30:58 But I just built ls2 itself and took ipk May 26 16:36:21 yes, but I needed to rebuild luna-qml-launcher too May 26 16:36:50 because herrie/ls2 branch doesn't have my acg fix yet May 26 16:38:21 Tofe: So many moving pieces, I try to keep track, but not easy :S May 26 16:38:53 For sure, it's far from easy May 26 16:39:28 There are 40+ recipes involved.... May 26 16:44:34 Tofe: To LS2 I also added: ret.push_back(std::string("com.palm.service.") + name.get()); May 26 16:44:34 ret.push_back(std::string("com.palm.") + name.get()); May 26 16:44:38 Not sure if that was needed or not May 26 16:45:27 Let me try ls2 without those 2 added lines May 26 16:45:49 in the same place as my change ? May 26 16:47:16 Only migrated APIs should go there, as the "if" tests if the client is trying to access an old API (com.palm.*, for instance) or a migrated one, and adds the counterpart May 26 16:48:22 Tofe: yes same place May 26 16:48:26 I was trying various things May 26 16:48:31 I now loaded LS2 without it May 26 16:48:33 Let me see May 26 16:48:44 If I understood things correctly, it should change anything. May 26 16:48:54 * Tofe hopes he understood something about ls2 May 26 16:49:11 it *shouldn't* change anything May 26 16:50:18 ls-monitor -i com,palm.systemservice works May 26 16:51:43 but com.palm.webappmanager not, org.webosport.webappmanager does May 26 16:53:09 mmh May 26 16:53:20 webappmanager isn't migrated yet, so that's surprising May 26 16:53:59 I'll give it a try and see May 26 17:00:03 This should be some of the last minor things to fix May 26 17:01:18 We May 26 17:01:28 We're getting in good shape to do a huge merge ;) May 26 17:02:27 oh, yes :) **** ENDING LOGGING AT Wed May 27 03:04:13 2020