01:25.58 | *** join/#maemo-ssu LauRoman (~LauRoman@5-14-22-54.residential.rdsnet.ro) |
02:03.41 | jonwil | xes: ping |
03:39.02 | *** join/#maemo-ssu povbot (~supybot@178-236-196-236.csc.lt) |
03:47.23 | *** join/#maemo-ssu esoare|2 (~kvirc@71-222-117-68.ptld.qwest.net) |
04:55.40 | jonwil | bencoh: ping |
05:41.44 | freemangordon | bencoh: yeah, happened here as well |
06:03.48 | jonwil | Still haven't found something to reverse engineer, every binary I fiddle with ends up having some reason why its too hard |
06:03.53 | jonwil | All the easy stuff has been cloned already :( |
06:17.09 | *** join/#maemo-ssu futpib (~futpib@176.104.194.78) |
06:26.10 | bencoh | jonwil: pong? |
06:26.33 | jonwil | Looked at the device lock stuff, the actual UI is drawn by libcodelockui |
06:26.43 | bencoh | yup |
06:26.50 | jonwil | and I hit snags reverse engineering both it and osso-systemui-devlock |
06:27.04 | jonwil | libdevlock is just for managing the actual settings in CAL |
06:27.08 | jonwil | and has nothing to do with UI |
06:27.39 | bencoh | I think we have a .h somewhere for libcodelockui |
06:27.45 | jonwil | we do |
06:27.52 | jonwil | the dev package was released by nokia |
06:28.05 | jonwil | but that doesn't help with reverse engineering either it or osso-systemui-devlock |
06:28.24 | jonwil | we have a .h file for libdevlock too FYI |
06:28.36 | jonwil | created by me as part of my MCE work as it happens |
06:28.50 | bencoh | :) |
06:29.59 | bencoh | so ... we have a "documented" header, we know which symbols it depends on ... what's missing for libcodelockui? |
06:30.36 | bencoh | (apart from "I dont wanna write gtk stuff" :D) |
06:31.26 | jonwil | it just needs someone to write a clone that implements the documented API |
06:31.55 | bencoh | weren't you looking for a package to clone? |
06:31.56 | jonwil | well I would clone it myself but its got some wierdness going on that makes reverse engineering it hard |
06:32.08 | jonwil | hence why I am not cloning it :) |
06:32.25 | bencoh | which is? |
06:35.11 | jonwil | well for starters its got some private data structures (and my skills in regards to figuring those out are non-existent) |
06:36.18 | jonwil | and the dialog has the usual stuff that makes reverse engineering GTK bits annoying |
06:36.37 | jonwil | hence why I have decided to pick up work on libconnui and friends |
06:36.50 | jonwil | specifically figuring out the currently-unknown functions from those libs |
06:38.25 | bencoh | okay :) |
06:38.55 | jonwil | anyone know who runs the CSSU repos? |
06:39.14 | jonwil | I think it might be merlin1991 but I am not 100% sure |
06:41.39 | jonwil | or is it freemangordon? |
06:42.13 | bencoh | it's merlin1991 |
06:42.44 | jonwil | ok |
06:42.58 | bencoh | thumbs are hosted by merlin1991 as well, but -testing/-stable are hosted on maemo infra |
06:43.45 | jonwil | ok |
06:56.46 | freemangordon | bencoh: could you try if adding worker->xid = 0; on line 2256 in mafw-gst-renderer-worker.c (between eos and seek_position) fixes the issue for you |
07:06.44 | jonwil | merlin1991: ping |
07:10.36 | bencoh | freemangordon: line 2256? |
07:10.51 | freemangordon | hmm, well, my bad |
07:11.06 | freemangordon | it is other on your side (I added some g_warnings) |
07:11.27 | freemangordon | anyway the function is p, li { white-space: pre-wrap; } mafw_gst_renderer_worker_stop |
07:11.39 | freemangordon | stupid QtCreator !!! |
07:11.46 | freemangordon | mafw_gst_renderer_worker_stop |
07:12.23 | freemangordon | bencoh: after the comment /* Reset worker */ |
07:12.52 | freemangordon | init worker->xid to 0 somewhere there |
07:15.08 | freemangordon | bencoh: got it? |
07:15.44 | bencoh | yeah |
07:16.11 | freemangordon | ok, as I am going afk |
07:16.15 | freemangordon | bbl |
08:23.46 | *** join/#maemo-ssu freemangordon_ (~ivo@212.5.158.15) |
08:29.39 | bencoh | freemangordon: works fine thus far |
08:33.51 | *** join/#maemo-ssu freemangordon_ (~ivo@212.5.158.15) |
08:46.30 | bencoh | freemangordon: 11:29 < bencoh> freemangordon: works fine thus far |
10:21.15 | *** join/#maemo-ssu freemangordon_ (~ivo@212.5.158.15) |
10:22.18 | freemangordon_ | bencoh: seems there is more to fix |
10:37.14 | xes | freemangordon_: i think to recall that most of time the problem happens while trying to resume a previously played video |
10:37.28 | freemangordon_ | :nod: |
10:40.44 | freemangordon_ | seems the change I proposed fixes the issue at least partially |
10:40.56 | freemangordon_ | xes: you're on cssu? |
10:42.40 | merlin1991 | jonwil: pong |
10:45.42 | jonwil | xes: about your message earlier (which I saw in chanlogs) regarding per-contact ringtones, I took a look into it and in theory its possible. To make it work you would need to change hildon-sv-notification-daemon (part of hildon-home) so that it listens for the relavent "incoming call" signal from the cellular services daemon and grabs the incoming phone number so it can find the right... |
10:45.43 | jonwil | ...ringtone. Then when it gets the normal "play ringtone" signal from librtcom-call-ui/rtcom-call-ui it plays that sound instead of the default one. |
10:45.48 | jonwil | Wont work for VoIP though |
10:46.06 | jonwil | and you might need to clone the hildon-plugins-notify-sv library to pull it off |
10:46.13 | jonwil | but its definatly possible |
10:46.37 | jonwil | merlin1991: where does cssu get hald-addon-bme-dev from? |
10:47.03 | jonwil | its not in the nokia SDK repos or in any of the cssu git stuff |
10:47.12 | merlin1991 | one sec, let me check |
10:47.35 | merlin1991 | iirc it's in one of our gits that has a different name |
10:48.53 | jonwil | There should be something somewhere listing any git repos other than https://github.com/community-ssu that are needed by CSSU (like this one and anything else) |
10:49.44 | merlin1991 | actually there should be nothing else needed |
10:50.06 | merlin1991 | if something is missing on https://github.com/community-ssu then that's a mistake |
10:50.20 | jonwil | well hald-addon-bme-dev isn't there |
10:50.35 | jonwil | My guess is that hald-addon-bme-dev is comming from the bme replacement stuff |
10:51.13 | freemangordon_ | :nod: |
10:51.28 | freemangordon_ | we should pester Pali |
10:53.10 | jonwil | also merlin1991, would it be possible to get commit access on github to https://github.com/community-ssu/libconnui-dev and https://github.com/community-ssu/osso-systemui-dev ? I have plans to do some more work on reverse engineering the missing pieces to go into those dev packages so they can be made complete. |
10:53.25 | jonwil | Or if you would rather have me make patches and submit them for review, thats ok too, I can do that |
10:53.51 | freemangordon_ | jonwil: iirc you had commit access on gitorious, ain't? |
10:53.59 | freemangordon_ | s/commit/push/ |
10:54.07 | merlin1991 | jonwil: you should have mail |
10:54.21 | jonwil | yes I had push access on gitorious but not sure to which trpos |
10:54.30 | jonwil | repos |
10:54.41 | freemangordon_ | I guess to those you're working on :) |
10:55.44 | jonwil | Just reinstalled vmware so I can load up my Maemo DEV VM (I hadn't reinstalled it since I bought an SSD and re-installed Windows) so now I can properly work on those repos :) |
10:56.50 | merlin1991 | jonwil: where do you need hald-addon-bme-dev? because I don't have that in my scratchbox where I build all the cssu releases |
10:57.04 | jonwil | the package mce references it |
10:57.13 | jonwil | so if mce isn't built as part of cssu then that would explain it |
10:58.03 | merlin1991 | hm mce was in cssuT11 |
10:58.09 | merlin1991 | gotta check how I built that |
10:58.25 | jonwil | now I just have to figure out how to pull a github repo inside scratchbox inside my VM |
10:58.40 | jonwil | I guess I probably want to pull via ssh |
11:10.55 | jonwil | ok, now I can pull stuff from github :) |
11:20.07 | jonwil | ok, its telling me I dont have commit access to osso-systemui-dev |
11:20.31 | jonwil | My ssh keys are set properly and its letting me in, it just tells me I dont have push access to it |
11:21.18 | merlin1991 | jonwil: one sec |
11:23.10 | merlin1991 | github changed the permission system |
11:23.21 | merlin1991 | now there are "teams" inside an organization and other stuff |
11:23.40 | merlin1991 | jonwil: you should be able to push now (I hope) |
11:24.06 | jonwil | ok, I can push to that repo now |
11:24.25 | jonwil | so which repos do I now have push access to exactly? |
11:25.01 | merlin1991 | all of the community-ssu repos |
11:25.14 | jonwil | ok, thanks |
11:25.36 | merlin1991 | I thought I gave you enough permissons when I added you, but due to the fact the we had no proper team configured yet you were a member without any rights :D |
11:32.50 | jonwil | did you figure out how you compiled mce and where hald-addon-bme-dev came from? |
11:38.18 | bencoh | freemangordon: yeah, I've just seen a strange bug |
11:40.22 | bencoh | click on a video, window changed to player waiting (dots), failed and moved back to video list |
11:40.43 | bencoh | re-clicking on video openeded a new window with player inside that closed almost instantly |
11:41.24 | *** join/#maemo-ssu freemangordon_ (~ivo@212.5.158.15) |
11:42.01 | freemangordon_ | bencoh: hmm? |
11:42.03 | bencoh | not sure what happened, maybe the player process of mediaplayer that kept crashing/relaunching |
11:42.40 | freemangordon_ | never happened here |
11:42.53 | bencoh | first time I see this behavior as well |
11:43.08 | bencoh | I havent been able to reproduce it from the begining |
11:44.17 | bencoh | but when it happened, reclicking on the mediaplayer window from task manager led to window closing/reopening |
11:45.19 | freemangordon_ | I guess it is a bug in OMP |
11:45.30 | bencoh | stock mediaplayer |
11:45.31 | bencoh | but maybe |
11:45.35 | freemangordon_ | ooh |
11:45.38 | freemangordon_ | OMP here |
11:45.55 | freemangordon_ | maybe that's why i've never seen it |
11:46.11 | freemangordon_ | however, besides that xid, there is something else |
11:46.17 | freemangordon_ | could be in gst-dsp this time |
11:46.20 | bencoh | player hanging seems to be a common bug |
11:46.23 | bencoh | so gst or mafw |
11:46.28 | freemangordon_ | yeah |
11:46.34 | freemangordon_ | mafw for sure |
11:46.54 | freemangordon_ | otherwise it won't recover on mafw-gst-renderer restart |
11:47.09 | bencoh | oh, and ... how the hell did they manage to overlay controls over video and switch it on/off without any playback artifact? |
11:47.12 | freemangordon_ | it is supposed to recreate the gst pipeline on every new video |
11:47.41 | freemangordon_ | bencoh: omap3 overlay supports it |
11:47.47 | freemangordon_ | it is a HW feature |
11:48.09 | bencoh | hmmm... |
11:48.23 | bencoh | mafw-gst-renderer uses xvimagesink with xoverlay |
11:48.30 | freemangordon_ | :nod: |
11:48.34 | bencoh | so ... where does the omap3 magic happen? |
11:48.48 | freemangordon_ | toldya, in HW |
11:48.50 | bencoh | OMP is able to overlay controls as well, but playback stutters |
11:49.02 | freemangordon_ | do gst-inspect on xvimagesinl |
11:49.04 | bencoh | stock mediaplayer has no playback issue, it's smooth |
11:49.10 | freemangordon_ | *xvimagesink |
11:49.22 | freemangordon_ | well, OMP has it glitches still |
11:50.17 | freemangordon_ | but this "hang" is driving me nuts, that is the last bug in Maemo still standing (for me) |
11:50.52 | bencoh | I see glitches as well when playing with a plain, simple gst-launch playbin2/xvimagesink pipeline and notifications/volume pop |
11:51.11 | bencoh | (the same glitches) |
11:51.31 | bencoh | but I dont see any on stock mediaplayer, so I guess they found a way |
11:51.46 | freemangordon_ | that's normal as well, volume indicator is not overlaid via overlay |
11:52.09 | freemangordon_ | gimme a second to find how was it called |
11:57.49 | freemangordon_ | bencoh: http://processors.wiki.ti.com/index.php/UserGuideDisplayDrivers_PSP_03.00.00.05#Color_Keying |
11:58.13 | bencoh | yeah I came across color keying as well |
11:58.29 | freemangordon_ | that is how it is supposed to be done |
11:58.33 | freemangordon_ | no ide what OMP does |
11:58.36 | freemangordon_ | *idea |
11:59.27 | bencoh | but I dunno in which window stock MP makes mafw render the video |
11:59.35 | bencoh | (the xid) |
11:59.42 | bencoh | I'm starting to think it's the rootwindow |
12:00.09 | bencoh | oh, and ... I still dont understand how it can playback when gdb-attached (mafw-gst-renderer) |
12:08.56 | *** join/#maemo-ssu freemangordon_ (~ivo@212.5.158.19) |
12:09.26 | freemangordon_ | bencoh: it is possible that it does DMA from DSP to framebuffer |
12:09.32 | jonwil | would it be useful to poke in stock MP to see what sort of calls its making to this code (and if so, which functions should I be looking for?) |
12:10.09 | freemangordon_ | jonwil: what for? |
12:10.23 | jonwil | this mafw discussion |
12:10.33 | bencoh | freemangordon_: who is "it"? |
12:10.35 | freemangordon_ | implementing color keying in OMP is easy |
12:10.48 | jonwil | it being the stock media player |
12:10.48 | freemangordon_ | bencoh: HW :) |
12:10.54 | bencoh | gstreamer pipeline lives in mafw-gst-renderer |
12:11.02 | jonwil | i.e. figuring out what calls its making that are different to omp |
12:11.03 | freemangordon_ | not sure |
12:11.15 | jonwil | the goal here seems to be to make omp do what stock mp does |
12:11.18 | bencoh | and it's a "simple" playbin2/xvimagesink pipeline |
12:11.49 | bencoh | unless I missed someting |
12:11.51 | freemangordon_ | bencoh: I don;t think it is that simple, I guess output buffers are allocated by the player, not mafw |
12:12.06 | bencoh | output surface maybe |
12:12.14 | bencoh | (hence the xid) |
12:12.16 | freemangordon_ | also, you don;t need a running process for DMA to happen |
12:12.33 | bencoh | do you think it's end-to-end DMA? |
12:12.49 | bencoh | with no software interaction |
12:12.51 | freemangordon_ | it is not, xvimagesink does a copy |
12:13.06 | bencoh | then I dont understand how it can run when attached |
12:13.42 | bencoh | which means there has to be something else, but I dont see what :) |
12:13.59 | freemangordon_ | or you're making some mistake :) |
12:14.04 | bencoh | most probably |
12:14.06 | bencoh | :) |
12:14.43 | freemangordon_ | bencoh: anyway, even with the xid fix, OMP still hangs on me. way harder to trigger it, but still |
12:14.54 | freemangordon_ | so, there is another bug living somewhere in mafw |
12:17.10 | freemangordon_ | bencoh: what I suspect is that pipeline is not really reset on a new playback, or mafw's internal state is not reset |
12:18.29 | bencoh | yeah |
12:18.32 | freemangordon_ | jonwil: no, the goal is to find the bug in mafw which causes both players to "hang" when video playback is attempted |
12:18.35 | bencoh | funny thing is: |
12:18.37 | bencoh | <PROTECTED> |
12:18.37 | bencoh | <PROTECTED> |
12:18.44 | freemangordon_ | :D |
12:18.49 | bencoh | before _construct_pipeline |
12:18.52 | bencoh | so I guess it should be :D |
12:19.18 | freemangordon_ | camera-ui reuses the pipeline iirc |
12:19.25 | bencoh | or .... something's racy |
12:19.34 | freemangordon_ | don;t see why mafw cannot |
12:19.47 | freemangordon_ | bencoh: yeah, or that one |
12:20.00 | bencoh | camera-ui has no media/format detection issues |
12:20.07 | bencoh | pipeline can be static |
12:20.19 | bencoh | playbin2 has a lot of things going under |
12:20.24 | freemangordon_ | sure |
12:21.05 | bencoh | it's quite amazing that it works as it is considering the amount of "automagic" stuff :] |
12:22.25 | freemangordon_ | anyway, I put g_warning(__FUNC__) in all functions in ...-worker.c file, will try to find the difference between "ok" and "hang" once I am back home |
12:22.51 | freemangordon_ | bencoh: you may try to do the same as well |
12:23.45 | bencoh | how are you supposed to see warn messages wit mafw? |
12:24.02 | bencoh | run it without wrapper? |
12:28.52 | xes | freemangordon_: yes, cssu devel thumb |
12:30.17 | xes | jonwil: per-contact ring tone would be great if really working without slow down everything like with the other "fantasy" solutions of the past |
12:33.49 | freemangordon_ | bencoh: no idea |
12:34.00 | freemangordon_ | but I see them in syslog :) |
12:35.01 | freemangordon_ | xes: ringtones are supposed to be "prepared" as pcm, so they should not slow down the system |
12:40.08 | *** join/#maemo-ssu Sicelo009N (~sicelo@unaffiliated/sicelo) |
12:42.13 | bencoh | freemangordon_: how right, syslog |
12:42.29 | bencoh | (I dont have it since it's not a dev device) |
12:45.28 | Sicelo009N | easy to install.. have it on all my N900 |
12:46.17 | xes | freemangordon_: yes, but old tricks with a service swapping symbolic links with ringtone associations in a txt file... were a mess |
13:20.39 | *** join/#maemo-ssu sparetire_ (~sparetire@unaffiliated/sparetire) |
13:25.02 | *** join/#maemo-ssu freemangordon_ (~ivo@212.5.158.41) |
13:25.27 | freemangordon_ | xes: right, but that's unrelated |
13:25.58 | freemangordon_ | the correct way is what jonwil said (I guess) |
15:37.47 | *** join/#maemo-ssu Pali (~pali@Maemo/community/contributor/Pali) |
15:38.23 | *** join/#maemo-ssu Sicelo009N (~sicelo@unaffiliated/sicelo) |
16:33.40 | *** join/#maemo-ssu povbot (~supybot@178-236-196-236.csc.lt) |
17:02.27 | *** join/#maemo-ssu useretai- (useretail_@gateway/shell/elitebnc/x-kzzbcpkjqhdlfcmx) |
17:50.15 | *** join/#maemo-ssu futpib (~futpib@176.104.194.78) |
18:03.39 | *** join/#maemo-ssu _LauRoman (~yaaic@5-14-22-54.residential.rdsnet.ro) |
18:04.39 | *** join/#maemo-ssu LauRoman|Phone (~yaaic@5-14-22-54.residential.rdsnet.ro) |
18:34.25 | *** join/#maemo-ssu jon_y (~enforcer@2001:e68:4074:444c:e01f:8e6d:cdca:d0e5) |
19:09.26 | freemangordon | xes: btw, did you try the last modest? it should have "show more" working as it is supposed |
19:09.35 | freemangordon | bencoh: any progress? |
19:09.59 | bencoh | regarding? mafw? |
19:10.33 | freemangordon | yep |
19:10.39 | bencoh | not really, the strange mediaplayer bug occured again (player window closes and opens back like it crashed) |
19:10.52 | bencoh | (and no playback, of course) |
19:10.52 | freemangordon | hmm |
19:11.34 | freemangordon | I am going to dig it further in a while |
19:12.12 | bencoh | I'd probably need to install syslog and add g_warns everywhere as you suggested |
19:12.36 | freemangordon | oh, I thought you already did it |
19:12.46 | freemangordon | I see no other sane way to debug that |
19:12.47 | bencoh | not yet no |
19:15.41 | bencoh | I played with gstxoverlay/colorkey and managed to overlay gtk widgets on the video (just like stock mediaplayer), but that's another story :) |
19:16.51 | bencoh | oh, and ... it feels like it's easier to trigger the mafw bug when using a gst video pipeline in a different process beforehand |
19:40.51 | *** join/#maemo-ssu Sicelo009N (~sicelo@unaffiliated/sicelo) |
22:19.29 | xes | freemangordon: modest "show more" now is working correctly. But i'm still testing general behavior |
22:27.18 | *** join/#maemo-ssu jonwil (~jonwil@27-33-80-219.tpgi.com.au) |
22:31.41 | *** join/#maemo-ssu peetah (~peetah@cha92-9-82-236-202-86.fbx.proxad.net) |