**** BEGIN LOGGING AT Thu Nov 26 03:00:43 2015 Nov 26 03:12:19 ~seen pali Nov 26 03:12:23 pali <~pali@Maemo/community/contributor/Pali> was last seen on IRC in channel #neo900, 22d 3h 39m 48s ago, saying: 'campaign for what?'. Nov 26 03:36:17 hmm. i have portrait vkb. I thought all devices have it. It's just about enabling vkb from menu. Nov 26 06:23:45 jonwil: hi! Nov 26 06:23:50 hi Nov 26 06:23:55 I saw your changes, good work so far :) Nov 26 06:24:15 where did you get cmtspeech code from? Nov 26 06:24:28 I took the code that was in https://github.com/community-ssu/cmtspeech/ Nov 26 06:24:36 and reverse engineered/changed it Nov 26 06:24:42 I mean the fremantle version Nov 26 06:24:48 only RE? Nov 26 06:24:58 yes only RE Nov 26 06:25:07 I reverse engineered it all out of module-nokia-voice Nov 26 06:25:12 ok Nov 26 06:25:21 great job Nov 26 06:25:25 on top of the existing FOSS Meego cmtspeech cod Nov 26 06:25:27 code Nov 26 06:26:49 jonwil: we'll have big troubles with that thread_func, for some reason reset_call_stream_states and the functions it calls were declared inline in the header file Nov 26 06:27:23 so the code is replicated all over the place, thus the size of that function Nov 26 06:28:01 thats going to be annoying Nov 26 06:28:25 yeah Nov 26 06:28:46 at least we have the code from the Meego source to help us out Nov 26 06:29:00 I copied the code from meego, but until we start testing it on the device, we wouldn't know how close it is Nov 26 06:29:55 It wont be correct, I can say that now Nov 26 06:30:12 I know :) Nov 26 06:30:29 but at least we have some code to start with Nov 26 06:30:51 you seem to have gotten voice_cmt_dl_deactivate somewhat correct at least (for example) Nov 26 06:31:15 and also voice_cmt_ul_deactivate from the look of it Nov 26 06:31:32 I tried hard to have at least that part correct Nov 26 06:32:16 actually, besides thread_func and the function it calls, the other parts should be more or less correct Nov 26 06:32:25 *functions it calls Nov 26 06:33:28 I am going to take a look at thread_func and see if I can see anything you got wrong :) Nov 26 06:33:51 that would be great Nov 26 06:34:13 I can already see one mistake, you missed a pa_rtpoll_install call at the start of the function Nov 26 06:35:38 yeah Nov 26 06:35:53 will fix it Nov 26 06:36:13 I will fix it when I commit any other fixes I find :) Nov 26 06:36:27 ok Nov 26 06:37:53 I also see a missing check against the return value of cmtspeech_open Nov 26 06:38:07 where it does (if !cmtspeech) do stuff Nov 26 06:38:26 oh wait that's part of check_cmtspeech_connection i think Nov 26 06:38:33 yes Nov 26 06:38:44 it is like spaghetti :( Nov 26 06:39:42 looks like the whole if (!cmtspeech) bit is a copy of check_cmtspeech_connection Nov 26 06:40:12 oh, now I see why they put the stuff in headers, see voice_aep_sink_process Nov 26 06:40:12 so once we verify that its properly cloned, we can tick it off and move on to the next step Nov 26 06:40:55 chech_cmt... should be correct I think, but nevertheless won't hurt if you check it as well Nov 26 06:41:00 *check... Nov 26 06:44:53 I think the hardest part of all this is going to be the mountain of math that sits behind functions like wrap_a_audio_enh_downlink and wrap_a_audio_enh_uplink Nov 26 06:45:35 you mean the NEON code? Nov 26 06:46:16 oh, I see what you mean Nov 26 06:47:12 but the good news is that those functions are relatively small, albeit lots of Nov 26 06:48:54 the tricky part there will be figuring out all the data structures Nov 26 06:49:06 yeah Nov 26 06:49:20 init_main is where a lot of the stuff there will need to start Nov 26 06:50:36 init_main looks like its the main function from something standalone taken and thrown into a blob Nov 26 06:51:48 yes Nov 26 06:52:10 I am not sure we should RE it at all Nov 26 06:53:12 so you are suggesting we go with my idea of simply re-using all of this math stuff (i.e. anything that doesn't call PulseAudio or talk to the hardware)? Nov 26 06:54:16 no, I'd prefer to not carry blobs around Nov 26 06:54:30 ok, so what do we do? Nov 26 06:54:49 the same we did with xprot, record and music Nov 26 06:55:05 once we have the PA part, I'll RE the algos Nov 26 06:55:25 for testing we can use the blobs though, to verify PA part is ok Nov 26 06:55:40 sounds ok to you? Nov 26 06:55:43 ok Nov 26 06:55:53 what about init_main? We need that if we want all the algos to work :) Nov 26 06:56:13 do we? well, we'll have to RE it then :D Nov 26 06:56:28 I though it isn't called Nov 26 06:56:31 it is Nov 26 06:56:34 ok Nov 26 06:56:44 voice_update_parameters calls it Nov 26 06:57:17 basically it looks like they originally wrote something to test out aep Nov 26 06:57:42 which took a bunch of command line parameters (there is even "usage" info in that function which might help) Nov 26 06:57:54 then they wrapped that up into the blob Nov 26 06:58:19 yes, I saw that when I looked at the code back then Nov 26 06:58:43 iirc there are code parts that try to open files Nov 26 07:00:40 anyhow, I am sure we will get there in the end Nov 26 07:01:00 and have a usable replacement for yet another nokia blob Nov 26 07:02:41 "yet another"? Nov 26 07:03:02 hmm, I'd say this is one of the major stoppers for porting fremantle around Nov 26 07:03:14 for Neo900 it is Nov 26 07:03:28 since we are copying the audio system Nov 26 07:04:02 not only, you need that for n900 as well, you you want to use newer kernels Nov 26 07:04:07 yes we do Nov 26 07:04:09 *if you Nov 26 07:04:39 but if you just wanted Fremantle apps and core on an existing device that had its own audio setup, you wouldn't need pulseaudio-nokia to make it happen Nov 26 07:04:40 also, I guess it will be needed ,at least partially, if one wants to port to some android HW Nov 26 07:05:03 you need that for voip Nov 26 07:05:09 afaik Nov 26 07:05:53 Any android device will have its own stuff for making voip audio sound good (that will be tuned to the audio bits of that hw) Nov 26 07:06:17 but we won;t be able to reuse those Nov 26 07:08:42 you wouldn't be able to run Fremantle on most Android devices in any case, the GPU alone would be a problem (I dont think you can get x11 GPU drivers for most Snapdragon series SOCs for example) Nov 26 07:09:12 that said, our mission here is focused on the N900 and Neo900 where we most definatly DO need a clone of pulseaudio-nokia to move forward Nov 26 07:09:38 ok, looks like check_cmtspeech_item and pollfd_update are both correct Nov 26 07:09:39 you don;t really need GPU to run fremantle, hildon will be happy without it afaik Nov 26 07:10:14 but yeah, this is not in the first stage anyway Nov 26 07:12:17 however, /me gtg Nov 26 07:12:31 bye, see you later Nov 26 07:13:01 cya Nov 26 07:25:30 Nota Bene we are _not_ copying any cmtspeech related hardware! Neo900 cmt audio is a simle PCM "card" Nov 26 07:25:36 simple even Nov 26 07:31:50 no but we copied the rest of the n900 audio system Nov 26 07:31:55 most of it anyway Nov 26 07:35:05 basically all of it Nov 26 07:40:34 huh Nov 26 07:41:04 something split Nov 26 15:04:49 freemangordon: I've found a valid fix for the cssu thumb installation issues re repositories Nov 26 15:05:08 new enabler is in testing, you probably want to merge master into your thumb branch and upload a new version to -thumb Nov 26 16:50:02 merlin1991: thanks, will do Nov 26 18:19:15 does communication error mean that my SIM card is dieing? Nov 26 18:19:38 could be Nov 26 18:19:50 but it can also be dying modem Nov 26 18:19:59 or dirty sim contacts Nov 26 18:20:00 well, I don't think its the cellphone tower Nov 26 18:20:07 heh Nov 26 18:24:27 i get those errors on occasion Nov 26 18:32:36 Vajb: I don't believe you will have proper portrait keyboard. Nov 26 18:34:22 enabling forced rotation will make the vkb squash into portrait format, but I wouldn't call this portrait vkb. It's not fully usable IMO. Nov 26 18:44:48 throw the ball up in the air and it takes a panorama image https://www.panono.com/home Nov 26 19:27:48 what's happening to freenode? Nov 26 19:36:27 sixwheeledbeast: ok. Then that's what it is (portrait vkb). Usable easily with pointer IMO. Harder with fingers though. Nov 26 19:36:59 i mean i have forced rotation enabled... Nov 26 19:37:19 not all button are accessible Nov 26 19:39:00 seems to me that they are Nov 26 19:39:33 can't see symbols on two of them, but they r still accessible with stick atleast Nov 26 19:40:27 ok four of them r missing proper signs Nov 26 19:40:56 layout switcher, shift, backspace and special character button Nov 26 20:24:37 Is this just a channel about the Maemo software or can I ask a question about the N900 hardware? Nov 26 20:24:56 yes you can Nov 26 20:25:44 n900 is the best linux phone available (in sane price tag range) Nov 26 20:26:49 Timo: it's a chan for both :) Nov 26 20:27:52 um kay, oscp revision -54 upped Nov 26 20:28:14 i'm planning to add earwax in -55 Nov 26 20:28:23 that being? Nov 26 20:28:47 Timo: we're curios now .. what did you want to ask :) Nov 26 20:28:55 http://pastebin.com/raw.php?i=6eHJ6JjM Nov 26 20:31:08 Sicelo: Well, the usual USB port thing. However, it doesn't seem to be the soldering on the board. It's more like the problem is *inside* the port. Charging only works when the USB cable is in a specific position and the problem worsens over the time... Nov 26 20:31:23 ~usbfix Nov 26 20:31:24 i guess usbfix is http://talk.maemo.org/showthread.php?t=75920 - and **NEVER** use epoxy (unless you want to seal your device for underwater), or https://www.youtube.com/watch?v=fYz5nIHH0iY#t=1866, you will basically need two irons: a small good one (or better hot-air reflow) and a 60+ Watt Nov 26 20:32:54 Timo: it happened to me .. ultimately you'll be resoldering the port Nov 26 20:33:39 i'd say the earlier the better Nov 26 20:33:59 Ah, so it isn't inside the port. It really is the soldering. Nov 26 20:34:09 or if one has weak soldering skills, just start hunting for cheap ~20-40usd second hand one Nov 26 20:34:30 it also can just be inside port .. but there's no way to fix that. you change port as a whole Nov 26 20:34:54 KotCzarny: Skills aren't the problem (one has an experienced father), but we don't have the right soldering iron. Nov 26 20:35:21 I guess buying a cheap second hand one is my best option, right? Nov 26 20:35:39 you can invest in proper iron and start fixing them Nov 26 20:35:42 Timo: i used normal soldering iron on 2 N900. first smd project at that Nov 26 20:35:54 Hmm, and it worked, Sicelo? Nov 26 20:36:18 it didn't just work ;) it work*s* perfectly even now Nov 26 20:36:43 you need to do your research properly .. so you understand the whole process of smd soldering Nov 26 20:37:21 with soldering iron, copious amounts of flux are a definite necessity. you won't get far without it Nov 26 20:37:21 Hmhm. Nov 26 20:38:34 I inspected the connections to the motherboard, but they seem to be good, not weak or something. Nov 26 20:38:48 here is the technique i used: https://www.youtube.com/watch?v=5uiroWBkdFY Nov 26 20:39:07 ideally though, use hot air Nov 26 20:39:50 and get a printed-circuits soldering flux Nov 26 20:40:10 not some corrosive flux for other soldering uses Nov 26 20:40:16 KotCzarny: i want to try that earwax thing on my pc .. tbh i like the sound 'in' my head, but would like to try this too. maybe i will like it Nov 26 20:40:27 sicelo, its in the sox Nov 26 20:40:39 it works as advertised Nov 26 20:40:48 i was using it in the xmms times Nov 26 20:40:52 (~10 years ago) Nov 26 20:41:09 so i need to install sox? Nov 26 20:41:43 yeah, and it can read mp3/play to alsa i think Nov 26 20:42:00 or just use pc and convert to wav Nov 26 20:42:30 Sicelo: Does the N900 have unleaded soldering tin? Because my iron doesn't get hot enough to melt that. :-/ Nov 26 20:43:25 i used a 30-40W iron .. for desoldering, use braid .. i will say this is a necessity too Nov 26 20:44:21 Timo: don't you have good, equiped electronics shops around? they should be able to do it for you :) Nov 26 20:44:38 just make it clear that any mistake my lead to their instant death, haha Nov 26 20:44:45 :> Nov 26 20:45:16 I know one electronics shop in my town but I think they only sell parts, and don't do any reparations themselves. Nov 26 20:45:30 I can ask it though. Guess it's safer than doing it myself. ;) Nov 26 20:45:36 look around .. you may find experienced people Nov 26 20:45:43 "safer", hmm Nov 26 20:45:46 you never know Nov 26 20:46:04 give them your samsung first ;) Nov 26 20:46:30 Hehehe Nov 26 20:47:14 People don't seem to sell their N900s anymore. Nov 26 20:47:27 depends on the region Nov 26 20:47:33 Netherlands. Nov 26 20:47:41 lots of them in .pl Nov 26 20:47:54 most in 30-60usd range Nov 26 20:47:58 I think there were a few ones on maemo recently Nov 26 20:48:18 Timo: i've been searching markplaats for maemo a lot lately, but no results Nov 26 20:48:27 bencoh: Those are rather expensive, given the (usually quite high) shipping costs. Nov 26 20:48:33 The only reason I will get rid of my N900 is if it breaks in a way that is unfixable or if by some miracle I am able to afford a new better phone (which means Neo900) Nov 26 20:48:36 managed to pickup a n900 last year for 25,- Nov 26 20:48:47 henkjan_: I've bought one for €10,- lately, but it did have quite some problems. Nov 26 20:48:59 so you have two Timo? Nov 26 20:49:03 Yep. Nov 26 20:49:07 :) Nov 26 20:49:12 it begins Nov 26 20:49:16 then take care of your babies :) Nov 26 20:49:40 my n900 has a defective camera shutter button :( Nov 26 20:49:49 no taking pictures for me Nov 26 20:50:09 um .. nicocam allows you to capture with C key on hwkbd ;) Nov 26 20:50:17 Sicelo: Of course. Unfortunately one has a non-functioning power button (that's the €10,- one), the other one has the USB problem. Hopefully I can buy one soon. :) Nov 26 20:50:49 if you can't find working N900, buy broken one .. move parts around Nov 26 20:50:49 henkjan_: I'm rather sure the white square on the left in cameraui2 does the job. Nov 26 20:51:16 Sicelo: The point is both the power button and the usb port are too difficult to move. Nov 26 20:51:48 Sicelo: thnx for pointing me to that software. i'll give it a try Nov 26 20:52:00 Timo: I found a shop(internet) that still sells unboxed n900, though not exactly cheap Nov 26 20:52:10 Timo: i'll try that one also Nov 26 20:52:12 henkjan_: it is that cameraui2 that Timo mentioned Nov 26 20:52:26 freemangordon: how much? Nov 26 20:52:42 about 300 euros Nov 26 20:52:52 eek Nov 26 20:53:02 Err, that's a bit of a lot too much. :-P Nov 26 20:53:29 well, I bout it for about 450 back then, so the price is reasonable Nov 26 20:53:35 *bought Nov 26 20:53:39 yes, was writing same :) Nov 26 20:53:53 I was thinking of ~50 euros. I have a brand new digitizer and the screen from my previous ones so a cheap one with a working motherboard is fine. Nov 26 20:54:06 bought mine for EUR220 in 2011 Nov 26 20:54:20 Heh, €150 in 2011. Nov 26 20:56:25 Timo: get the L3/L4 Service Manuals, read up on tmo about the experiences of others .. you will find the usb and power button won't be as difficult as they seem right now Nov 26 20:57:47 wtf! Nov 26 20:58:06 Sicelo: If I can't find a decent replacement device I'll definitely try! Nov 26 20:59:06 * Timo parties. Nov 26 20:59:41 I asked on a forum whether someone had an N900 available to sell, got a reply. Hopefully it's inside my price range. Nov 26 21:01:46 freemangordon: unboxed n900? do they still have some? Nov 26 21:01:50 sounds crazy in 2015 Nov 26 21:02:07 unsold stock Nov 26 21:02:31 sure, but still ... 6 years later Nov 26 21:02:46 what else can they do with it :p Nov 26 21:03:58 :) Nov 26 21:04:00 donate Nov 26 21:04:11 bencoh: at least they list it as avalable http://gsm-bg.net/%D1%82%D0%B0%D0%B3%D0%BE%D0%B2%D0%B5/nokia/nokia-n900 Nov 26 21:04:23 sorry, but the site is in bulgarian only Nov 26 21:04:36 however, the price is: Nov 26 21:04:45 ~669/2 Nov 26 21:04:45 334.5 Nov 26 21:05:03 and it is supposed to have 2 years of warranty Nov 26 21:05:10 no idea if this is for real Nov 26 21:05:41 oh well ... :) Nov 26 21:05:44 "nice" Nov 26 21:06:16 at that point its better to save a bit more for neo900 Nov 26 21:06:47 ...if it ever comes to production state. Nov 26 21:06:49 why not just get a second hand one? Nov 26 21:08:03 I guess it is getting harder to find one in a good condition Nov 26 21:09:21 I have several in decent condition Nov 26 21:28:05 ~repositories Nov 26 21:28:05 repositories are frequently old. Nov 26 21:28:15 ~maemo-repos Nov 26 21:28:15 well, maemo-repos is http://wiki.maemo.org/Repository#List_of_Maemo_repositories Nov 26 21:30:11 ty Nov 26 21:54:56 freemangordon: ping Nov 26 21:55:06 hi jonwil Nov 26 21:56:35 I happened to have the same changes you just made (plus some others) in my local commit so it was easier to pull your changes, overwrite the voice-cmtspeech.c file with the copy containing all my changes and push that than to undo my local commit, merge yours, make all my changes again then commit it all Nov 26 21:57:24 I guess I should pull then? Nov 26 21:57:27 yes Nov 26 21:57:29 you should pull Nov 26 21:57:34 ok Nov 26 21:58:04 read my commit to see what changes I made :) Nov 26 21:59:13 I am still reverse engineering thread_func and friends so dont change that Nov 26 22:00:12 ok, anyway I was about to go to bed :) Nov 26 22:00:21 ok :) Nov 26 22:01:57 so yeah leave voice-cmtspeech.c alone until I say otherwise (proper cloning of thread_func & friends is the last thing to do in that file) and check status.txt for other files containing needs-cloning functions to work on instead :) Nov 26 22:02:06 when you do more work on things that is :) Nov 26 22:02:10 yeah Nov 26 22:02:54 BTW, are you sure that reset_call_stream_states does not set u->cmt_connection.record_running and u->cmt_connection.playback_running? Nov 26 22:03:16 reset_call_stream_states does set both Nov 26 22:03:33 oh, sorry, a bit sleepy already: Nov 26 22:03:38 I meant: Nov 26 22:04:09 are you sure those 2 are not reset in ul/dl_deactivate? Nov 26 22:04:14 hmmm yeah maybe you are right and voice_cmt_dl_deactivate/voice_cmt_ul_deactivate does set them Nov 26 22:04:16 good call Nov 26 22:04:29 I might have miss-read what was going on in the meego source Nov 26 22:04:35 its logical that those 2 functions set them Nov 26 22:04:39 yes Nov 26 22:04:51 will fix Nov 26 22:05:30 also, close...on_error calls reset_call_stream_states, not deactivate functions IMO Nov 26 22:05:46 it cant call reset_call_stream_states Nov 26 22:05:56 why? Nov 26 22:06:07 since we know from thread_func that reset_call_stream_states has a pa_log_xxx call with its name in it Nov 26 22:06:19 and close_cmtspeech_on_error has no such log Nov 26 22:06:27 hmm, right Nov 26 22:06:52 but those 2 calls do the same, unless I am missing something Nov 26 22:07:02 2 calls == deactivate functions Nov 26 22:07:16 there are other things to add to reset_call_stream_states when I get to that bit Nov 26 22:07:38 not just the debugs but something else that close_cmtspeech_on_error doesn't do but reset_call_stream_states clearly does Nov 26 22:07:48 ok Nov 26 22:08:22 but please, fix that deactivate functions first, to not forget it later Nov 26 22:08:33 yes I fixing deactivate now Nov 26 22:08:56 just pushed :) Nov 26 22:09:16 also, please add that missing pa_rtpoll_install(u->cmt_connection.rtpoll); in thread_func :) Nov 26 22:09:43 yeah that's going to happen in a sec when I start from the top of thread_func and reverse engineer it Nov 26 22:10:10 ok Nov 26 22:15:29 missing pa_rtpoll_install added :) Nov 26 22:15:47 I now have to verify mainloop_cmtspeech and one bit in thread_func Nov 26 22:15:56 and the things mainloop_cmtspeech calls :) Nov 26 22:18:36 :) Nov 27 01:52:06 FWIW, I got my USB cable snagged and the cable broke instead of the N900 :D **** ENDING LOGGING AT Fri Nov 27 03:00:52 2015