**** BEGIN LOGGING AT Fri Nov 28 02:59:58 2014
Nov 28 05:14:35 <DougReeder> I have a minimal sample app which attempts to subscribe to a service: https://github.com/DougReeder/subscribe-enyo-webos
Nov 28 05:15:08 <DougReeder> Under webos 3.0.5, it works as expected, receiving updates at roughly 10s intervals indefinitly.
Nov 28 05:17:11 <DougReeder> Under Black Eye, it usally stops reciving updates after the first response.  Occasionally, it receives 2 or 3 before it stops receiving.
Nov 28 06:17:42 <Herrie> DougReeder: That might be a bug... I recall reading a similar bug from Garfonso in the bug tracker if I'm correct
Nov 28 06:18:02 <DougReeder> Oh, can you give me something to search for?
Nov 28 06:19:31 <DougReeder> Searching for “subscrib” doesn’t find anything like that.
Nov 28 06:21:31 <DougReeder> Is there a way to search issues by reporter?
Nov 28 06:23:46 <Herrie> LEmme check
Nov 28 06:25:38 <DougReeder> A search for open issues w/ autthor Achim Koenigs doesn’t find anythng like that
Nov 28 06:28:13 <Herrie> http://issues.webos-ports.org/issues/493
Nov 28 06:28:35 <Herrie> That seems closed
Nov 28 06:28:40 <Herrie> Not sure if it's even related
Nov 28 06:28:50 <Herrie> Anything useful in the logs you've seen?
Nov 28 06:28:58 <Herrie> Could be something is just dying :S
Nov 28 06:31:32 <DougReeder> That’s activities, rather than subscriptions.
Nov 28 06:32:49 <DougReeder> Under Balck Eye, using ls-monitor, I see a “cancel” call beign issued soon after the initial call.
Nov 28 06:33:12 <DougReeder> I’m writing it all up now that I have two sample apps.
Nov 28 06:39:53 <Herrie> It could be just a bug
Nov 28 06:40:10 <Herrie> Lot's of these "more advanced" things haven't been really tested yet to be honest
Nov 28 06:40:19 * DougReeder nods
Nov 28 06:40:22 <Herrie> We've been mainly focussing on getting the OS running and improving
Nov 28 06:40:35 <Herrie> So we're happy we have devs around who play with this and report bugs ;)
Nov 28 06:48:29 <DougReeder> Here’s the bug report: http://issues.webos-ports.org/issues/757
Nov 28 06:49:12 <DougReeder> Here’s the other app, which uses webos-lib, to do the same thing: https://github.com/DougReeder/subscribe-webos-lib
Nov 28 06:50:29 <DougReeder> Damn, that was exhausting, working up two sample apps. But that should make it much easier to evoke the bug.
Nov 28 06:54:34 <Garfonso> morning
Nov 28 06:54:55 <Garfonso> I had not tried subscriptions, yet... so no bugreport from me. :)
Nov 28 06:56:45 <DougReeder> Well, then.
Nov 28 06:57:09 <DougReeder> I have an idea how to code around that for now, in Contacts.
Nov 28 06:58:12 <Garfonso> so db.find with watch=true does not yet work?
Nov 28 06:58:41 <DougReeder> Right.
Nov 28 06:59:04 <DougReeder> But I can issue a separate “watch” command.
Nov 28 07:00:11 <DougReeder> … or, rather, apps can’t make use of db.find with watch=true, because subscriptions don’t work.
Nov 28 07:00:25 <DougReeder> I suspect db.find with watch=true would work using luna-send.
Nov 28 07:00:59 <DougReeder> … because subscriptions work using luna-send.
Nov 28 07:02:45 <Garfonso> but that hopefully is a temporary thing, so don't make the work around to big. ;)
Nov 28 07:04:41 <Herrie> DougReeder: It might actually be related to http://issues.webos-ports.org/issues/755
Nov 28 07:05:12 <Herrie> Nov 03 16:30:07 mako activitymanager847: [] [pmlog] ActivityManager SM_BOOTSTS_UPDATE_RETRY {} Subscription to System Manager failed retrying: {"errorCode":-1,"errorText":"com.palm.systemmanager is not running.","returnValue":false,"serviceName":"com.palm.systemmanager"}
Nov 28 07:06:18 <DougReeder> Dunno.  "errorText":"com.palm.systemmanager is not running."
Nov 28 07:06:52 <Herrie> Yeah could be it's not running or got killed along the way
Nov 28 08:10:20 <Tofe> morning
Nov 28 08:10:51 <DougReeder> Morning. Late night here. :-)
Nov 28 08:10:58 <Tofe> morphis: couldn't check the kbd last evening, I was a bit ill; but tonight should be ok :)
Nov 28 08:11:39 <Tofe> DougReeder: yes, seems like it :)
Nov 28 08:14:38 <morphis> DougReeder: btw. the comparision with com.palm.wifi/findnetworks is a bit not sufficient
Nov 28 08:14:48 <DougReeder> How so?
Nov 28 08:14:53 <morphis> main reason is that com.palm.wifi is completely reimplemented for openwebos
Nov 28 08:15:09 <morphis> and it's not guranteed that new results are send every 10 seconds
Nov 28 08:15:27 <morphis> those should only come when something has changed
Nov 28 08:15:31 <DougReeder> Okay, but all we need for this, is that it send results repeatedly.
Nov 28 08:15:42 <morphis> so either a network added/removed, signal strength changed
Nov 28 08:15:45 <morphis> right
Nov 28 08:15:55 <morphis> better would be some predictable
Nov 28 08:16:10 <morphis> like a watch for a db8 object we then change on our own and verify we get the subscription response
Nov 28 08:16:18 <DougReeder> They key point, to my thinking, is ls-monitor shows the app issuing a cancel.
Nov 28 08:16:24 <morphis> hm
Nov 28 08:16:33 <Herrie|Veer> morphis: ofono seems to behave a lot nicer now :)
Nov 28 08:16:37 <morphis> Herrie|Veer: :)
Nov 28 08:16:43 <morphis> Herrie|Veer: will fix firstuse today
Nov 28 08:16:56 <morphis> DougReeder: you want to look at the native code for that?
Nov 28 08:17:09 <DougReeder> I could take a look
Nov 28 08:17:30 <morphis> https://github.com/webOS-ports/qtwebkit/tree/webOS-ports/master/Source/WebCore/platform/webos
Nov 28 08:17:45 <Herrie|Veer> morphis: thnx, where are these country/language combinations stored?
Nov 28 08:18:12 <DougReeder> Calling luna-send -n 5 luna://com.palm.wifi/findnetworks '{"subscribe": true}' gets repeated results, under eithere webOS or LuneOS.
Nov 28 08:18:13 <Herrie|Veer> I always have all my OS-es in English, but when I select English my country is not in the list :P
Nov 28 08:18:36 <morphis> Herrie|Veer: the main difference here is that locale and region are different
Nov 28 08:18:43 <morphis> so you can have english - uk as locale
Nov 28 08:18:49 <morphis> and netherlands as region
Nov 28 08:19:00 <morphis> but also english -us as locale
Nov 28 08:19:04 <morphis> and that is what I did wrong
Nov 28 08:19:16 <Herrie|Veer> Ah :)
Nov 28 08:20:10 <Herrie|Veer> But I guess we have 2 bugs for this right? One the systemmanager not running and 2nd something with the language/locale/region?
Nov 28 08:20:21 <DougReeder> Does PalmServiceBridge handle JS -> PalmBus?
Nov 28 08:21:51 <morphis> DougReeder: yes
Nov 28 08:22:03 <morphis> PalmServiceBridge is the object you instantiate from Js
Nov 28 08:22:12 <morphis> it's behind every enyo ls2 kind
Nov 28 08:22:34 <morphis> that in turn uses the LunaServiceBridge to have just one connection to ls2 per app
Nov 28 08:25:29 <Herrie|Veer> morphis: ofonod now gets started on the TP4G :) Need to confirm on N4
Nov 28 08:25:42 <morphis> ok
Nov 28 08:26:44 <Garfonso> a bit strange is that m_subscribed is never read...
Nov 28 08:27:18 <morphis> the code mostly came as is from legacy
Nov 28 08:28:10 <DougReeder> Yes, I get the impression it’s simply not doing what it should when the subscribe flag is on.
Nov 28 08:28:25 <morphis> possible
Nov 28 08:28:49 * DougReeder yawning “Anyway, I need to get to bed. I won’t make sense of code tonight.”
Nov 28 08:29:11 <morphis> the key thing is when cancel is called
Nov 28 08:29:35 * DougReeder nods
Nov 28 08:29:59 <Garfonso> or it's the otherway round, it missbehaves if subscribe is off and fixed somewhere else? Like I can't see why a non subscribing call should not receive multiple responses from the PalmServiceBridge code
Nov 28 08:30:12 <Garfonso> night DougReeder :)
Nov 28 08:30:12 <morphis> Garfonso: right
Nov 28 08:30:24 <morphis> generally there is no difference between a LSCall and subscriping
Nov 28 08:30:52 <morphis> see https://github.com/openwebos/luna-service2/blob/master/include/public/luna-service2/lunaservice.h#L462
Nov 28 08:31:07 <morphis> there are OneReply variants of LSCall which you should call when you expect just one reply
Nov 28 08:31:29 <morphis> otherwise LSCall will return response as long as you listen for them and don't call cancel on the token you got
Nov 28 08:35:39 <Garfonso> LunaServiceManager only calls LSCallFromApplication, never LSCallFromApplicationOneReply
Nov 28 08:44:26 <morphis> right
Nov 28 08:44:36 <morphis> but afaik the enyo kind handels the subscription case
Nov 28 08:44:50 <morphis> where it calls PalmServiceBridge.cancel when it didn't subscribed
Nov 28 19:03:17 <DougReeder> In PalmServiceBridge.cpp, I see that call() contains the code if (LSMESSAGE_TOKEN_INVALID == listenerToken) {ExceptionCode ec; cancel(ec);}   but I can’t see where the value of listenerToken comes from.
Nov 28 19:07:02 <DougReeder> PalmServiceBridge.idl defines       readonly attribute unsigned long token;
Nov 28 19:08:06 <DougReeder> and PalmServiceBridge.h defines    int token();
Nov 28 19:08:58 <DougReeder> token() just returns listenerToken, but I don’t see where listenerToken comes from.
Nov 28 19:14:20 <nizovn> DougReeder: it comes from https://github.com/webOS-ports/qtwebkit/blob/webOS-ports/master/Source/WebCore/platform/webos/LunaServiceMgr.cpp#L243
Nov 28 19:19:01 <DougReeder> So it’s coming from further in.  Hmm, well I don’t see anything else in the source code.
**** ENDING LOGGING AT Sat Nov 29 02:59:59 2014