**** BEGIN LOGGING AT Tue Feb 02 02:59:58 2016 Feb 02 08:07:52 morning Feb 02 08:08:30 Morning! Feb 02 08:34:50 Tofe: Seems we might need you to look into the Jethro update sometime. I get LunaWebAppManager.service constantly crashing on Maguro. Feb 02 08:37:18 Ah wait on Tenderloin it happens too Feb 02 08:38:10 Tofe: Should I send you some log? Feb 02 08:39:17 Seems problem might be in service_registry_impl.cc Feb 02 08:42:12 If you have a log, why not, yet Feb 02 08:42:13 yes Feb 02 08:42:19 doesn't crash on N4? Feb 02 08:42:48 Tofe: Didn't have a chance to test latest unstable yet. Previous one wouldn't even boot Feb 02 08:43:15 Sound doesn't work on maguro and tenderloin with Jethro could be due to webappmgr crashing not sure Feb 02 08:43:24 Or another problem altogether Feb 02 08:44:10 I see no mention of the word pulse or audio in the log at all Feb 02 08:45:32 Ah alsactl restore 0' failed with exit code 99 Feb 02 08:54:38 Sent you logs and some relevant extracts. Feb 02 09:12:45 thanks Feb 02 09:13:56 Not sure you can do anything with it. The lunawebappmgr one seems straight forward but could be a single one in a line of many that would follow... Feb 02 09:14:56 It looks like we have two separate issues here Feb 02 09:17:38 LunaWebAppManager crash is here: https://github.com/webOS-ports/qtwebengine-chromium/blob/webOS-ports/master/chromium/content/common/mojo/service_registry_impl.cc#L29 Feb 02 09:21:21 Seems in newer versions of Chromium that's being dealt with differently Feb 02 09:22:00 Which version does jethro take? Feb 02 09:22:12 Qt 5.6, or still our 5.5.1 ? Feb 02 09:22:27 Tofe: Quite sure still 5.5.1 let me check Feb 02 09:22:46 Would make sense, since it points exactly to that line Feb 02 09:25:05 yes, later version deal really differently with this, interesting... Feb 02 09:25:55 Seems in Chromium 44 they changed the whole !bound_ check Feb 02 09:26:05 Herrie|Veer: it could be that all is related to pulseaudio running in system mode Feb 02 09:26:24 Chromium does use pulseaudio behind the scene Feb 02 09:26:32 Not sure it'll be an easy backport or it might be better to do Jethro and Qt 5.6 at once Feb 02 09:26:40 and if the binding to pulseaudio failed, maybe bound_ isn't set correctly Feb 02 09:26:46 Tofe: We've been doing system mode forever I guess? Feb 02 09:26:54 Herrie|Veer: I don't know :) Feb 02 09:26:59 But yeah might be good to fix anyway I guess Feb 02 09:27:28 But when I see alsa warning, luna-next warning, qtwebengine not binding, it rings a bell Feb 02 09:27:48 Tofe: Yeah that's true Feb 02 09:28:48 It's not necessarily the system mode thing, but it point to pulseaudio nevertheless Feb 02 09:33:02 Could you send me the whole maguro journal? Feb 02 09:34:23 It's attached in the rar file ;) Feb 02 09:36:08 But it's the whole thing ? I don't see the beginning of the boot Feb 02 09:36:19 Tofe: Hmmmz Feb 02 09:36:27 This is what it exported Feb 02 09:36:35 I guess it was looping so much :P Feb 02 09:36:45 I can do another one later I guess Feb 02 09:36:45 oh. Feb 02 09:37:02 yes, just after boot maybe Feb 02 09:37:12 I'll reflash image and do it at firstuse instead Feb 02 09:37:20 This was on 2nd boot already I think Feb 02 09:37:39 As you wish; I don't think it matters much in this case Feb 02 09:38:52 I'll do it when home tonight Feb 02 09:39:00 The 2 minutes it takes extra :P Feb 02 10:09:49 Herrie|Veer: in your country, do you have "area codes" in phone numbers too? Feb 02 10:10:11 usually juste after the national prefix Feb 02 10:13:11 Enyo g11n says yes :p Feb 02 10:18:47 Tofe: Yeah :P Feb 02 10:18:54 6 for mobile Feb 02 10:19:28 (0)20 for Amsterdam, (0)10 for Rotterdam, (0)30 for Utrecht to name a few :P Feb 02 10:19:54 Herrie|Veer: I'd like to check something, if you have some minutes Feb 02 10:21:19 let's say you enter a new contact with the following fictive NL number: 0113234567 Feb 02 10:21:57 and then you look at db8 to know what is its normalized format Feb 02 10:23:08 I guess the normalized format here would be -765432-311--, but I'm not sure Feb 02 10:23:53 And furthermore, it's not the format that my little algo would output with phonenumber.js... Feb 02 10:25:06 Let me see what I have on my Veer for that number :P Feb 02 10:25:18 Which db kind? Feb 02 10:25:30 com.palm.person:1 Feb 02 10:28:00 It's not in there Feb 02 10:28:10 While I see it in my contacts app Feb 02 10:28:22 Ah wait Feb 02 10:28:43 should be hidden somewhere in the tons of output :p Feb 02 10:29:11 it's: "-765432-311--" Feb 02 10:29:56 For normalizedValue. Value shows "(0113) 234 567" Feb 02 10:36:40 Hey, I'm beginning to be pretty good at this stuff Feb 02 10:38:19 So, problem is, currently the phone app will compute the following normalized number: -7654323-11--, meanning it won't find the contact entered in Contacts Feb 02 10:38:50 Tofe: Ah Feb 02 10:39:30 This might have to do with our country settings somehow? Feb 02 10:39:43 If we want a reliable compatibility, there are two options here: either reuse Enyo.g11n's prefixes from here: https://github.com/enyojs/g11n/tree/master/source/phone/geo/data/area or migrate Contacts to use phonenumber.js Feb 02 10:39:58 I know I have my Veer set to NL for Phone numbers Feb 02 10:40:17 Could be that LuneOS uses US? Feb 02 10:40:41 no no, it's just not the same definition of the term "leading digits" Feb 02 10:40:44 NL phone numbers are always 10 digits. So 020-1234567 or 0123-456789 Feb 02 10:41:08 Could be the US allows longer phone numbers? Not sure Feb 02 10:41:13 I haven't checked the code Feb 02 10:41:59 It's not because of a country code mismatch, my use case here supposed both used NL Feb 02 10:43:17 it's just that google's data says the leading digits are only 2 digits, and Enyo proposes 3 digits for some areas in NL to have a better geolocation Feb 02 10:43:41 Tofe: Google is WRONG Feb 02 10:44:16 Likewise, if I check google's geolocation data, I'll certainly find up to 4 or 5 leading digits dedicated to geolocation Feb 02 10:45:26 We used to have even 4 digits so you could pinpoint the exact town. This was changed to 3 digits somewhere early/mid '90's if I'm correct to accommodate the 10 digits number countrywide and still have sufficient numbers Feb 02 10:46:29 https://github.com/googlei18n/libphonenumber/blob/master/resources/geocoding/nl/31.txt <-- that's what I was expecting Feb 02 10:47:48 So we used to have 03447-1226 for example this became 0344-571226 Feb 02 10:48:13 Well, that kind of data also exists for France, and is going to be more and more wrong, because it's now possible to keep the same phone number after moving to another place in the counter Feb 02 10:49:34 31344 now says Tiel which is the "bigger" town. We used to have 313447 which would point to the town Buren. Same for 31345 which now is bigger town "Culemborg" while previously 313453 would point to Beusichem for example Feb 02 10:50:38 I think we should do the db8 search based on the whole national number; being bother by area code while looking for a contact doesn't make sense Feb 02 10:50:42 But that's history anyway. 31344 and 31345 is what we should at least support, not 3134 Feb 02 11:43:09 Herrie|Veer: the issue also is the way Enyo.g11n stores international numbers: it either has the country code, or the area code, but never both Feb 02 11:44:13 which means that if you entered "+31113234567" for the phone number above, it would store "-765432--31-+", so there's a part of the national phone number that's missing... Feb 02 12:52:17 Tofe hmmz that's strange. It should do that but could be it only does it for numbers that are NOT in the users country Feb 02 12:52:48 Mine is set to +31 NL in regional settings. Could be that a US number would have a proper value in db8 Feb 02 12:53:03 Will check that in a bit Feb 02 12:53:21 Herrie|Veer: no, if you enter +31 it will normalize it differently, you'll see Feb 02 12:54:13 Tofe: Yeah, I need to test & see different scenarios Feb 02 13:18:30 Tofe: For my international non +31 numbers I do have proper values Feb 02 13:18:59 So I guess because I have my Regional telephone settings set to NL +31 the +31 act differently Feb 02 13:21:09 Hmmz it doesn't seem that consistent either :P Feb 02 13:21:19 I do have some with +31 in my db8 Feb 02 13:21:47 But it could be I added those when my phone was still on US setting Feb 02 13:25:35 Proper way imho is to store it with country code so with +31. If user enters it without the country code it should prefix it with +31 based on the country NL that was selected in Regional Settings? Feb 02 13:26:05 I sent you some examples, looks like it's pretty much all over the place on 2.2.x Feb 02 13:26:30 Then again this profile was shared with 3.0.5 and has gone through various migrations of devices Feb 02 13:48:01 Herrie: let me see a bit if I find the inconsistencies I was looking for Feb 02 13:48:23 Maybe it's bad luck and it only happens with the FR profile :p Feb 02 13:49:48 mmmh... maybe it's just the mobile numbers? Feb 02 13:49:59 looks like the "6" gets eaten Feb 02 13:53:45 It's gonna be a bit tough to get this working properly, if legacy's normalize is partially broken since the beginning... Feb 02 14:06:44 Herrie: is Contacts the only Enyo app that can add phone numbers in database ? Feb 02 14:07:12 (in LuneOS) Feb 02 14:46:33 Tofe: I think so. I think any other app like Just Type or phone app would just launch Contacts Feb 02 14:46:56 Though there might be something in luna-systemui not 100% sure or in one of the frameworks Feb 02 14:48:45 Herrie|Veer: because if that is so, then I'd say it's easier to migrate to phonenumber.js right away Feb 02 14:54:52 DougReeder or Garfonso might know more Feb 02 15:29:22 morning Feb 02 15:29:25 Herrie: pong Feb 02 15:29:56 Herrie: somehow my client didn't receive your messages :/ Feb 02 15:51:25 nizovn: No worries, can happem Feb 02 15:51:34 s/m/n Feb 02 15:52:06 We're looking to extend the audio service a bit Feb 02 15:52:14 Is that something you could pick up? Feb 02 15:52:45 yup Feb 02 15:53:29 Ok you want me to write down what we'll need? Feb 02 15:54:36 yes Feb 02 15:55:50 OK I hope to get to that tonight Feb 02 15:56:07 Part we need also for vibration since the prefs are stored in audio service :P Feb 02 15:56:48 ok Feb 02 15:58:56 Already digged out most calls, will also get the replies from legacy so you know what to reply :P Feb 02 16:36:42 nizovn: If you're bored now you could have a stab at the mediaindexer :P Feb 02 16:36:50 That has all details already in it Feb 02 16:37:07 But I know you're limited on time during weekdays Feb 02 16:39:10 yes, i probably can have a look tomorrow Feb 02 16:39:49 Shouldn't be too hard for someone who knows c++ :P Feb 02 16:40:05 Let me know when you need more info. I have everything for that at hand already Feb 02 16:40:11 another nodejs module not compatible with recent nodejs in meta-oe: nodejs-module-webos-dynaload/3.0.1-14+gitAUTOINC+d817785147-r0 Feb 02 16:40:30 DougReeder: was it you interested in newer nodejs in images? Feb 02 16:41:01 JaMa: I think it was Andolamin ;) Feb 02 16:41:09 But it might be DougReeder too Feb 02 16:48:04 JaMa: That was me Feb 02 16:48:34 JaMa: Any logfile that might help? Feb 02 17:00:38 Ended up abandoning my work on updated nodejs - only so much time to work on personal projects. Focusing on getting Raspberry Pi working with wayland-egl. Feb 02 17:01:03 Andolamin: https://bpaste.net/show/8aea7ce0aa8d Feb 02 17:01:12 ok Feb 02 17:02:06 node-dbus, node-sqlite3, node-taglib are already dropped, will drop this one as well Feb 02 17:53:23 Herrie|Veer: seems the only place to migrate would be here: https://github.com/webOS-ports/loadable-frameworks/blob/webOS-ports/master/contacts/javascript/properties/PhoneNumber.js#L162 Feb 02 17:55:02 Tofe: That's good Feb 02 18:29:36 * HoloIRCUser1 waves, being actually DougReeder, but not having figured out this IRC client yet. Feb 02 18:38:20 Herrie: I can pretty easily hack the PhoneNumber.normalizePhoneNumber function inside loadable-frameworks, but what do you suggest, to call our phonenumber.js ? Load it with a hard-coded path ? Feb 02 18:49:23 Tofe, Herrie, any app can create contact records, but only the contact linker can create the person records displayed by the Contacts app (and I presume other apps). Does that help? Feb 02 18:51:43 HoloIRCUser1: and as the contact linker also uses ContactsLib, in loadable-framework, I think we're in the right place to change the way the normalized numbers are generated Feb 02 19:59:59 ok, seems like that would sort of work, if I knew how to supply the region code correctly. But the main worry is how to organize the code and the dependencies to have a nice reuse without having too much hardcoded paths Feb 02 20:23:42 Tofe: I'm not a big fan of hardcoded paths, but even core-apps use it for Enyo, so I think we're OK with one more Feb 02 20:27:09 Herrie: you sure ? It would be a pretty ugly hard-coding :) Like "../../../../../../lib/qt5/qml/LuneOS/Telephony/phonenumber.js" Feb 02 20:27:51 I'm sure we can find something at least slightly better Feb 02 20:30:24 Tofe: just do a /usr/palm/etc Feb 02 20:30:36 Instead of using the dirty ../../../ variant? Feb 02 20:31:07 So at least it's clearer where it's coming from... I'm not a big fan of the ../../../../ variant ;)_ Feb 02 20:31:15 nope, doesn't work in a manifest Feb 02 20:31:36 it also was the version I tried first Feb 02 20:32:07 https://github.com/webOS-ports/loadable-frameworks/blob/master/contacts/manifest.json <-- this manifest Feb 02 20:32:55 Tofe: Ah Feb 02 20:33:01 Yeah manifest might be different Feb 02 20:33:04 It works in html Feb 02 20:33:28 https://github.com/webOS-ports/core-apps/blob/master/com.palm.app.accounts/index.html#l22 :S Feb 02 20:33:59 we could maybe split the problem in two: put phonenumber.js/* in /usr/palm/frameworks, and have a more reasonable relative path in the manifest, and include it in Qt with an absolute path Feb 02 20:35:29 Or, we can modify https://github.com/webOS-ports/loadable-frameworks/blob/master/contacts/index.js so that it does a special treatment for our phonenumber.js Feb 02 20:36:22 yes, I like the latter Feb 02 20:37:15 we can add a dedicated array in manifest.json, like "depends", which would be expressed in absolute paths, and evaluate that in index.js Feb 02 20:37:58 :) Feb 02 20:52:08 voilĂ : https://github.com/webOS-ports/loadable-frameworks/compare/webOS-ports/master...Tofee:webOS-ports/master?expand=1 Feb 02 20:52:34 The PhoneNumber modification is still a bit buggy I think Feb 02 20:56:43 Maybe someone here knows how to retrieve the region in the correct way ? Garfonso maybe ? I mean something like "US" or "NL" etc Feb 02 20:58:35 Tofe: this.oldPhoneRegion = (payload.locale && payload.locale.phoneRegion) || this.currentRegion || {countryCode: "us", countryName: "US, Canada, Caribbean"}; Feb 02 20:59:12 luna-send -n 1 palm://com.palm.systemservice/getPreferenceValues '{"key":"locale"}' Feb 02 20:59:25 Then you should have a locale.phoneRegion Feb 02 20:59:31 You can try this on your Pre 3 Feb 02 20:59:54 Or I can do my Veer now if you want :P Feb 02 21:00:21 but I mean, in the context of the contacts library up here, don't we have something more straightforward ? I found Globalization.Locale.getCurrentPhoneRegion(), but I suspect it doesn't give the wanted result Feb 02 21:01:14 Maybe a simple console.log would help :p Feb 02 21:01:34 Tofe: Quite sure the Globalization.Locale.getCurrentPhoneRegion() will give the same results as my query above :P Feb 02 21:02:37 Tofe: https://bpaste.net/show/6a872ca8cf4f Feb 02 21:04:00 :) ok good, so maybe my solution wasn't so far from working Feb 02 21:07:42 Tofe: This is the full 3.0.5 code for this: https://bpaste.net/show/6a3d45015f52 Feb 02 21:08:52 That's from Regional Settings app ;) Feb 02 21:09:49 There's a function to convert all numbers Feb 02 21:09:56 For the rest nothing too special Feb 02 21:10:03 yup Feb 02 21:10:28 But I'll debug that tomorrow; I'm a bit tired right now :p Feb 02 21:16:36 Tomorrow morning hospital with little one, so will be online around lunch I guess **** ENDING LOGGING AT Wed Feb 03 02:59:58 2016