**** BEGIN LOGGING AT Mon Dec 15 02:59:59 2014 Dec 15 03:34:30 EricBlade , standards for what get retuned for a service bus call were not enforced. Dec 15 03:35:26 returnValue is always returned, but beyond that, you can’t depend on anything. Dec 15 03:35:47 Worst of all, errorCodes were not standardized. Dec 15 03:37:04 http://www.webos-internals.org/wiki/JavaScript_Services lists a few retuned by the OS, rather than individual services. Dec 15 03:37:30 errorCode is nominally an integer, but some of the built-in services return a string! Dec 15 03:38:56 The problem for 719 is that LuneOS currently returns no useful info about *why* the call failed. Dec 15 03:39:54 { "returnValue": false, "errorText": "" } is a valid failure response, but not a useful one. Dec 15 04:38:19 sure, but that doesn't discount that the application should rely on returnValue to determine if a call was successful Dec 15 04:47:10 Oh, definitely. Dec 15 04:48:02 I fixed the app so it knows the call failed, but it doesn’t know what to tell the user. Dec 15 04:50:31 Garfonso , I need to know what “types” should be available for an IM field. Dec 15 04:51:22 libpurple is listed as suppoting: Bonjour (Apple's implementation of Zeroconf), Gadu-Gadu, IRC, Lotus Sametime, MySpaceIM, Mxit, MSNP Dec 15 04:52:20 … Novell GroupWise, OSCAR (AIM/ICQ/MobileMe), SIP/SIMPLE, SILC, XMPP/Jingle (Google Talk, LJ Talk, Gizmo5, Facebook Chat, …), YMSG (YIM only basic chat and file transfers), Zephyr Dec 15 04:54:54 Does “Bonjour” include the current iMessage (using Appl ID) or is that only iChat on the local network? Dec 15 05:04:23 One of the absolutely frustrating things about the old system was involving something about the type .. something along the lines of, it was used in certain areas for some things, but then ignored in other areas, i think.. i really can't remember all of the issues that i ran into Dec 15 05:05:13 but if you're going to implement messaging in a way similar to the old system, then IMO it absolutely must be extendable to support new things that we haven't yet imagined. Dec 15 05:06:22 and I would be more than happy to plug in SynIRC and SynerGV3 to it. Dec 15 05:07:16 although i kind of think perhaps group chat should be left to a different resource than 1-to-1 .. Dec 15 05:07:21 Initially, we have libpurple. Dec 15 05:07:49 …but we should support multiple libraries. Dec 15 05:07:59 i suppose you could probably do IRC using Purple. Dec 15 05:08:36 It supports IRC. I have no clue what lib would provide the best user experience. Dec 15 05:09:31 this reminds me, although it's a year and a half too late for most anyone to care, i recently ran across a setting in contacts called "isImBuddy" or something like that, which might fix much of the contact integration problems that SynerGV 2 ran into Dec 15 05:31:14 DougReeder: there is something about IM types in loadable-frameworks. Dec 15 05:32:23 DougReeder: but actually chatthreader does not care about the type a lot. It only determines if it has to handle a phone number by checking for SMS/MMS. Dec 15 05:33:19 Hmmm, so for an IM name like “bob”, how does it know what server to contact? Dec 15 05:34:18 i seem to recall having to do some weird junk to get messaging to work with GV since it used phone numbers .. Dec 15 06:14:27 Garfonso, what exactly is loadable-frameworks/contacts ? Dec 15 06:15:31 And I can’t find anything matching “type_” in the reposiitory. Dec 15 06:16:30 Wait, found them Dec 15 06:18:32 …and they’re the old list from webOS - it includes .Mac, which two renamings ago! Dec 15 06:24:41 welcome to 2009! :D Dec 15 06:27:33 :-S Dec 15 06:53:12 DougReeder: searching does not work good for forked repositories on github.. better search in the openwebos repro. Dec 15 06:54:39 DougReeder: the loadable frameworks stuff is a bloated library that, in legacy, was used by contacts, messaging, telephpony apps and services. Basically Person is a wrapper around the db8 person:1 kind and Contact is a wrapper around contact:1 Dec 15 06:55:25 then there is a lot of stuff about things... uh.. fav, speed dial, photo cropping. It can also convert to/from vcard (which in fact c+dav uses). Dec 15 07:27:54 what package is loadable frameworks? Dec 15 07:27:58 er, what repo ? Dec 15 08:53:03 EricBlade: https://github.com/webOS-ports/loadable-frameworks Dec 15 08:54:52 JaMa: I am getting a lot fetcher errors with latest dizzy Dec 15 09:03:47 JaMa: fixed those Dec 15 09:11:44 the last commit merged? Dec 15 09:22:33 yes Dec 15 10:12:00 Morning! Dec 15 10:26:01 morphis: tested chatthreader today and got it working with some sms messages I copied from legacy. :) One important thing: conversations needs to be null or undefined. Dec 15 10:32:04 morphis: https://github.com/webOS-ports/org.webosports.messaging/pull/4 (or somebody, seems like I don't have write access for that repo) Dec 15 10:42:39 morning Dec 15 10:42:49 Garfonso: Great stuff :) Dec 15 10:44:22 morphis: looks good, sorry I haven't caught them Dec 15 11:06:18 Tofe: Assigned some stuff to you to look at :P Some stuff is probably minor to fix :) Dec 15 12:16:17 Herrie|Veer: yes, saw that, I'll look at it this week Dec 15 12:21:18 Tofe: morning Dec 15 12:22:58 Garfonso: nice Dec 15 12:23:20 Garfonso: you mean the conversations field on the message object? Dec 15 14:13:49 JaMa: somehow your upgrade broke the whole N4 Dec 15 14:13:56 graphics doesn't work anymore Dec 15 14:28:18 JaMa: ah.. never select android-header in distro space, that is really something which belongs into machine.conf Dec 15 14:28:31 that's why it's with weak assignment Dec 15 14:28:39 machine should override it Dec 15 14:29:23 Hey morphis, how likely is it we will ever see a TV build of LuneOS? Dec 15 14:29:44 JaMa: PREFERRED_PROVIDER_virtual/android-headers = "android-headers" isn't weak .. Dec 15 14:30:13 and when building for mako it selected android-headers Dec 15 14:30:20 GodGinrai: TV? Dec 15 14:30:26 hmm strange Dec 15 14:30:43 JaMa: shouldn't it be ?= instead of = Dec 15 14:30:50 morphis: I was curious if at some point I may be able to replace WebOS on my TV w/ LuneOS Dec 15 14:31:33 morphis: yes, strange that I didn't use ?= Dec 15 14:32:01 JaMa: but if we don't set it, it will be unset, right? Dec 15 14:32:07 and that will cause the build to break Dec 15 14:32:28 which is exactly what it should do if no provider is selected Dec 15 14:32:51 choosing one automatically can be hard to figure out later for a developer Dec 15 14:33:04 so better leave it unset to remind him that he is missing something important Dec 15 14:33:34 without setting it, it was showing warning in every qemux86 build Dec 15 14:34:18 where does qemux86 include it? Dec 15 14:35:03 sorry cannot look right now Dec 15 14:35:13 my build dir is in middle of meta-luneos creation Dec 15 14:35:14 ok Dec 15 14:35:37 I will revert that change so we have atleast a usable system again Dec 15 14:36:49 why not just change it to weak? Dec 15 14:38:01 JaMa: as choosing just one version of the android-headers and not the right one can be breaking your whole system Dec 15 14:38:11 had this two times and finally found out the headers were wrong Dec 15 14:39:32 making it non-deterministic doesn't make it less error-prone, does it? Dec 15 14:40:24 can't we select a dummy-provider by default? Dec 15 14:40:34 which then warns on the build output? Dec 15 14:40:37 what's dummy-provider? Dec 15 14:41:04 just one to have something set Dec 15 14:41:14 which doesn't do anything in the end Dec 15 14:41:25 installs nothing to sysroot Dec 15 14:42:52 cannot we set COMPATIBLE_MACHINE in both headers recipes to prevent using them from compatible MACHINE? Dec 15 14:43:04 hm Dec 15 14:43:15 we have phablet-headers in meta-android right now Dec 15 14:43:24 if we move it to meta-webos-ports then we can do that Dec 15 14:43:34 which is somethign we should do anyway Dec 15 14:43:42 as phablet-headers is something wop specific Dec 15 15:41:15 Garfonso: ping Dec 15 19:19:22 JaMa: btw. why are we still using WEBOS_SYSTEM_BUS_SKIP_DO_TASKS? Dec 15 19:23:11 because you guys just love CAPS lock that much :P Dec 15 19:24:41 morphis: ? I've only changed the default, so that the recipes which need the .bbclass to install it, sets it to empty value Dec 15 19:25:06 but still some recipes don't install it with cmake/qmake/make, so this bbclass is still needed to do it Dec 15 19:25:18 that was the reason for my emulator build not working on Friday Dec 15 19:26:45 ah I see, they need it to be set for the file-to-package association Dec 15 19:29:57 that's one part, but this variable is needed for do_install_append to install the files Dec 15 19:31:06 also see https://github.com/webOS-ports/meta-webos-ports/commit/894b20bc5f1fd6474676b7410182ca02a7e54277 from today for further improvement on this Dec 15 19:31:21 webos-filesystem-paths.inc will be pain to eliminate correctly Dec 15 19:31:52 because all variables are exported and in theory can be used from any generated Makefile Dec 15 19:32:31 my current plan is to generate buildhistory for all MACHINEs and then start eliminating exports while checking files-in-image.txt reports if something wents missing Dec 15 19:32:56 but all system bus files were missing when building luna-* in distro-less config (without webos-filesystem-paths.inc) Dec 15 19:33:19 other option is to introduce webos_paths.bbclass and inherit it from every webos related recipe to be sure Dec 15 19:34:21 JaMa: basically we need some of those paths defined Dec 15 19:34:32 why do you want to remove them? Dec 15 19:37:42 I want them included only in recipes which are using them Dec 15 19:38:08 and all exported variables invalidate all the sstate every time they are changed or new one added/removed Dec 15 19:38:51 ah ok Dec 15 19:38:53 that is good Dec 15 19:39:12 right now zlib is rebuild when someone decides to export webos_pkgconfig variable Dec 15 19:39:40 (zlib and everything else) Dec 15 19:39:56 just because maybe few webos components are using this variable somewhere Dec 15 19:40:15 and if it's used only in the recipes (e.g. in FILES_*) then it shouldn't be exported Dec 15 19:50:25 DougReeder: regarding the subscribe bug Dec 15 19:50:40 Hmm? Dec 15 19:54:52 DougReeder: http://issues.webos-ports.org/issues/757 Dec 15 19:55:13 I tried that with a simple example Dec 15 19:55:27 palm://com.palm.systemservice/setPreferences / getPreferences Dec 15 19:55:31 the settings app does that Dec 15 19:55:54 And you got different results? Dec 15 19:56:01 you only saw this problem with the com.palm.wifi service? Dec 15 19:56:27 Also with luna://com.palm.wifi/findnetworks Dec 15 19:57:13 hm Dec 15 19:57:49 DougReeder: btw. can you add your test case to https://github.com/webOS-ports/org.webosports.app.testr? Dec 15 19:57:50 You subscribed to getPreferences, and used setPreferences in another thred? Dec 15 19:57:59 Sure. Dec 15 19:58:26 right Dec 15 19:58:55 com.palm.wifi/findnetworks works differently in legacy and LuneOS Dec 15 19:59:00 When you use my test apps, do you get the same results? Dec 15 19:59:00 so not a good comparisio Dec 15 19:59:07 didn't tried it yet Dec 15 19:59:25 https://github.com/DougReeder/subscribe-webos-lib , right? Dec 15 19:59:40 The key point is the difference between subscribing to a service using luna-send, and from an app. Dec 15 20:00:00 Yes, and https://github.com/DougReeder/subscribe-enyo-webos Dec 15 20:00:12 yes Dec 15 20:00:23 the subscription in the settings app uses webos-lib Dec 15 20:00:31 and I triggered the change with luna-send Dec 15 20:00:46 DougReeder: there is an ipk in the repo, is that the last version? Dec 15 20:01:04 Maybe, maybe not. I don’t remember. Dec 15 20:01:16 ok Dec 15 20:01:21 Use tools/deploy.sh -i to install the latest version Dec 15 20:02:25 We should test the subscribe apps with getPreferences Dec 15 20:03:41 did already Dec 15 20:04:05 And subscription worked properly? Interesting. Dec 15 20:04:44 I see the app subscribing but returning only one result Dec 15 20:04:47 but that is fine Dec 15 20:04:54 it didn't detected any more networks yet Dec 15 20:05:26 but now a: Dec 15 20:05:27 ĉ1884.201 TX [PRV] call 220 com.palm.luna-1563 (/var/run/ls2/TiTtqT) com.palm.wifi (/var/run/ls2/H7HfyQ) /com/palm/luna/private/cancel «{"token":217}» Dec 15 20:05:33 interesting Dec 15 20:05:33 * DougReeder smiles “You should try in in my neighborhood” Dec 15 20:08:59 DougReeder: you tried with a db8 watch also? Dec 15 20:09:11 Yeah, that’s where I started. Dec 15 20:11:27 and that didn't worked, right? Dec 15 20:11:40 I am also not sure how the db8 watch is used to work Dec 15 20:11:47 if it just fires once or multiple times Dec 15 20:12:18 A watch fires once, but you need a subscription to make it work Dec 15 20:12:46 A find with a watch has 2 responses: the initial result & later the watch. Dec 15 20:13:19 A watch-by-itself has 2 responses: an immediate “yeah, i can do that” and the later watch. Dec 15 20:13:54 ah ok Dec 15 20:14:03 thought first it fires everytime something changes Dec 15 20:14:58 but then it is fine Dec 15 20:15:07 I belive not. The watch desn’t give you the changed records, so you need to do another find. Dec 15 20:15:15 ok Dec 15 20:15:25 DougReeder: so can you do three test cases in testr: Dec 15 20:15:33 1. with com.palm.wifi as you already did Dec 15 20:15:44 2. with com.palm.systemservice and changing some property Dec 15 20:15:54 3. with a db8 object which changes Dec 15 20:16:20 hope we can nail this down then Dec 15 20:17:04 * DougReeder nods Dec 15 20:17:43 it doesn't look like it is broken in general just in some cases Dec 15 20:20:59 That would be an interesting clue. Dec 15 20:21:23 For wifi, occasionally the app will get two or three responses. Dec 15 20:21:39 the wifi service is broken anyway .. Dec 15 22:53:01 morphis: pong? **** ENDING LOGGING AT Tue Dec 16 02:59:58 2014