**** BEGIN LOGGING AT Wed Jul 04 02:59:58 2012 Jul 04 06:31:09 i'll try again, what will I have to do to get http://bit.ly/N6L7lx into preware? Jul 04 06:32:24 filmor! Jul 04 06:33:13 most certainly :) Jul 04 06:33:20 i don't have an answer for that, other than that the best thing to do to get a response is to wait for someone who knows to see the message Jul 04 06:33:38 which could be anywhere from a minute to hours or more, so don't hold your breath in case everyone's asleep or away Jul 04 06:34:00 I was going to send you a message earlier, but then I got sidetracked Jul 04 06:34:03 no worries, I'll just pop in from time to time and scavenge the logs :) Jul 04 06:34:42 it's not that it is completed, but I squashed some more bugs Jul 04 06:35:03 btw, you don't happen to know why Mojo.Controller.errorDialog doesn't work in my code? Jul 04 06:35:40 I was wondering if you knew, or knew where to point me to look at least, if the native messaging plugs like yours, somehow appropriately handle sending a message via the correct account, when there are more than one account of the same type? Jul 04 06:36:44 hmm Jul 04 06:36:44 my knowledge of Mojo unfortunately stops about where node service initialization begins, and does not extend at all into the user-facing side Jul 04 06:37:33 I don't really want to plug in two yahoo accounts, as the sync for my 1100 yahoo contacts will take half a day Jul 04 06:37:50 :) Jul 04 06:37:59 from the code comments: Jul 04 06:38:01 * luna://im.libpurple.palm/sendMessage '{"serviceName":"aol","usernameTo":"palm","messageText":"test send IM","username":"xxx@aol.com"}' Jul 04 06:38:47 so using the correct account to send is actually done on the com.palm.messaging-side of things Jul 04 06:39:12 hmm. yeah, i think i didn't explain that right .. sorry, got a headache Jul 04 06:39:23 and it's late :) and i'm dumb sometimes heh Jul 04 06:39:40 the messaging app doesn't seem to indicate what account the outgoing message should be going to, as far as i can tell Jul 04 06:39:44 er, going from Jul 04 06:39:58 or, rather, it does, but it always uses the same Jul 04 06:40:16 hmm Jul 04 06:40:19 that's weird Jul 04 06:40:45 I actually didn't change any of this code, but from what I see here "username" is actually used so it has to be correct … Jul 04 06:40:52 example, in SynerGV, say i set up two voicemail accounts. I receive a message from say 1234567890 on one, and one from 0987654321 on the other. Jul 04 06:41:10 When I type in a reply in the messaging app, it will automatically pick the first account that has a matching serviceName Jul 04 06:41:16 and fill that in as the from account Jul 04 06:41:32 I think that is a bug in the UI Jul 04 06:41:48 it definitely is, and there's a comment right above it that talks about what a bad idea it was Jul 04 06:41:53 :D Jul 04 06:42:28 so, i was wondering if somehow the native plugins have some way around it, or if they are just broken when dealing with multiple accounts with the same serviceName Jul 04 06:42:32 I have the same problem when chatting with someone on the pre3 (which indicates it's no new problem …) over facebook Jul 04 06:42:57 they're not broken I guess Jul 04 06:43:06 are you sure "username" isn't filled properly? Jul 04 06:43:29 give me a second to dig out impostah Jul 04 06:49:59 i am having a rather difficult time with testing this, since sending from one number will get combined Jul 04 06:51:51 https://github.com/filmor/webos-messaging/blob/master/src/SendOneMessageHandler.cpp Jul 04 06:52:05 that is the file which handles outgoing messages Jul 04 06:52:20 in particular the first function doSend Jul 04 06:52:58 from line 88 on it reads the username-value and fails if that one is not found Jul 04 06:53:26 ok, so, i've just verified that if i just try to send a message using that type, it will send with the first account on the list ... if i send a message to the second account, it's username and to.addr are set properly Jul 04 06:53:42 ah, but "from" is the key, not "username" Jul 04 06:54:27 right, that's the incoming side Jul 04 06:55:19 if i then reply to that, the outgoing message username and from.addr is set to the first account, not the account that should be replying Jul 04 06:56:34 but I'm quite sure that the libpurple plugins don't handle that in any way Jul 04 07:00:22 yeah it looks like most likely it would fail here too Jul 04 07:00:53 although it doesn't look like it's an error, in that file, if there's no username Jul 04 07:02:12 hmm? Jul 04 07:02:22 i rather couldn't believe that i was getting the wrong username in the outgoing message, and then i found "//NOTE: this is a dumb attempt at getting an account. It should really figure out which account // is associated with the buddy (recipient.value)" Jul 04 07:04:02 it looks like it doesn't bail out if there's no user object found in the doSend() function Jul 04 07:06:45 hmm Jul 04 07:06:47 indeed Jul 04 07:07:06 but the query should give no results … Jul 04 07:07:12 i'd guess that the transport function that it gets passed along to eventually then guesses at it Jul 04 07:08:13 won't this fail? Jul 04 07:08:14 MojDbQuery query; query.where("serviceName", MojDbQuery::OpEq, m_serviceName); query.where("username", MojDbQuery::OpEq, m_username); query.from(IM_LOGINSTATE_KIND); Jul 04 07:08:52 m_username should be empty in the case where no from-block is given … Jul 04 07:10:22 hmm. so why does the fromAddrObject setup work differently than the serviceName and currentMsgdbId ... Jul 04 07:11:09 maybe a username can be nothing Jul 04 07:12:56 maybe we should just fix the messaging-programs to be rigorous :) Jul 04 07:14:21 if the messaging program just took the "to" in the conversation i'm replying to, and set the "from" to it, i'd be happy Jul 04 07:15:08 yep, that would be the way to go Jul 04 07:18:29 there seems to be an assumption that it doesn't matter what actual account is used to send a message, only that the same type of account is used Jul 04 07:19:42 the whole concept is a bit weird Jul 04 07:20:17 like this servicename-stuff … Jul 04 07:20:43 yeah most all of the checks are for serviceName, and have little if any regard to what actual account to use Jul 04 07:21:51 yay, 1.1M logfile … Jul 04 07:22:51 and that's only from 2 days of use Jul 04 07:23:10 i should really reduce logging a lot before calling this 1.0 ;) Jul 04 07:24:18 LunaSysMgr output is pretty insane too in the woce Jul 04 07:25:10 i've got a big long running post that i've been adding to every time i run into weird stuff in Messaging that I think should be changed .. input is welcome :) https://developer.palm.com/distribution/viewtopic.php?f=105&t=19951&p=100803#p100803 Jul 04 07:26:11 EricBlade, why not post them up on issues.webos-ports.org Jul 04 07:26:32 then everything would be in one place, and perhaps a different developer could pick up an issue and work it ;) Jul 04 07:26:51 that doesn't seem appropriate for things we don't have open code for yet Jul 04 07:26:52 issues.webos-ports.org is designed to be a centralized location for feature requests and bug reports for WOCE Jul 04 07:27:01 regarding the first one, that is fixed in my pluggins ;) Jul 04 07:27:25 which "first one" Jul 04 07:27:33 see, exactly, if it was a bug report on issues, then filmor could attach his patch to the issue, then it could get pulled into WOCE, and perhaps eventually upstreamed to HP Jul 04 07:27:58 there was a simple bug (wrong parentheses) in the code that made it commit every buddy individually to the db :) Jul 04 07:28:10 maybe that one is in skype too Jul 04 07:29:14 I haven't examined the process fully, but I think it's more to do with the contact watcher / person creator Jul 04 07:29:30 okay Jul 04 07:29:48 one thing that really bugs me are the messages in /var/log/messages when it is initially pulling everything in and it is saving the contact images Jul 04 07:29:55 it is something along the lines of... Jul 04 07:30:19 "couldn't create file /some/path/file-cache/somewhere/file.ext This error can be ignored if it is within the file-cache" Jul 04 07:30:33 that was just the impression that i was getting from the logs, which seemed to indicate that the person db handler as well as the buddy tempdb were picking up each change individually and processing them Jul 04 07:30:36 if it can be ignored if it is within the file-cache, why log the error when it is within the file-cache? Jul 04 07:31:17 yeah lots of annoyances like that Jul 04 07:34:40 it will be great to actually work on the messaging-app code :) Jul 04 07:34:59 oooh, you're a masochist aren't you Jul 04 07:35:05 :D Jul 04 07:35:49 could be, the libpurple-stuff was also kinda weird ;) Jul 04 07:36:08 and I still don't get the idea behind much of mojo … Jul 04 07:36:18 (the c++-lib, not js) Jul 04 07:36:45 it looks like pre-c++98-code in parts . Jul 04 07:36:48 .. Jul 04 07:36:48 Palm's style guidelines for Enyo make everything really difficult to read for me, and the Messaging app has so much cross communication between kinds that it's almost mind warping Jul 04 07:37:08 hehe Jul 04 07:37:47 just try to do something dynamically, that's fun stuff :) Jul 04 07:38:01 although it's a lot easier in enyo than in mojo … Jul 04 07:38:18 enyo totally makes sense to me. mojo makes me go "wth" Jul 04 07:39:22 yep, pretty much Jul 04 07:39:50 i actually had to go through the framework code to figure out how to dynamically build a preferences list … Jul 04 07:40:01 in enyo it just works as expected Jul 04 07:40:42 yeah i don't know. i dig into anything in /usr/palm/framework that is user-side oriented other than enyo, and i go insane Jul 04 07:40:46 and still I don't get any error dialogs and I have absolutely no idea why :) Jul 04 07:41:05 all the Scene and Controller and that stuff ugh Jul 04 07:41:48 oh yeah, that's fun stuff Jul 04 07:41:51 if I'd picked up webOS before Enyo, there's no way i'd be involved here. Enyo is simplicity when armed with a tiny bit of javascript knowledge Jul 04 07:42:14 true Jul 04 07:42:24 it unbreaks a lot of js :) Jul 04 07:47:26 add to that that i spent the last 6 months shaking of 20 years of rust in C (there's still about 18 years of rust left) Jul 04 07:55:32 now what i could possibly do, as a workaround to something that i really don't think i should have to work around :( is run a query where conversation == the conversationID of the outgoing message, and to.addr == [ array of account names that could be possible ], order by most recent, limit 1, and if a result shows up, use that instead of the one supplied by messaging Jul 04 07:57:33 hmm. Impostah has some display issues in landscape mode on touchpad. wonder if htat's a WOCE flaw or an Impostah flaw Jul 04 08:01:15 hmm. i wonder if it's even possible to query on an array like you can a subobject .. where conversations.0 == ? i wonder Jul 04 08:06:09 doesn't look like that works Jul 04 08:06:44 and you can't compare equality in arrays in javascript Jul 04 08:06:58 s/in arrays/between arrays Jul 04 08:30:43 what exactly where you trying? Jul 04 14:36:09 list Jul 04 21:16:06 if filmor ever looks at this log.. the stock messaging connectors appear to break completely when used with multiple accounts on TouchPad -- I just tried enabling only one of my two yahoo accounts for messaging, and the messaging system completely fails to send with it .. it's the second account in the list, the first one in the list does not have messaging enabled.. so it bails Jul 04 22:46:45 jellow jellow Jul 04 22:47:42 just checkng, 2.6.36 is the lastest known kernel with webos patches on, right ? Jul 05 00:45:24 xaiki, I thought 2.6.35 was...but I'm a bit behind Jul 05 01:03:06 ka6sox: yeah, ... compared to 'the world©' it is, I wanted to have a look to a diff between palm and 3.3 (android), but all I can find is 2.6.36... and that is too much to be manageable =/ Jul 05 01:36:40 xaiki, openwebos will be a 3.3 kernel(3 months from now) Jul 05 01:39:11 ka6sox: yeah, I was trying to work ahead of it Jul 05 01:39:13 =D Jul 05 01:39:30 but well, I guess we'll just have to wait… hope android doesn't jump to 3.5 in the mean-time. Jul 05 01:39:47 or that HP dumps a kernel like they did for woce Jul 05 01:58:14 hello Jul 05 02:07:33 xaiki, well...we have some things...just not a kernel for WOCE...however we do have kernels we built for almost all things. Jul 05 02:34:42 ka6sox: all things, beeing, 'all things that were made to run webos at some point' or does it have things like i9100 ? **** ENDING LOGGING AT Thu Jul 05 02:59:58 2012