Wizzup | freemangordon: sounds good | 00:06 |
---|---|---|
freemangordon | right. but in the meanwhile I modified rules instead of builddeb, going to try | 00:07 |
Wizzup | great | 00:11 |
Wizzup | I'm going to and get some sleep early tonight, ttyl | 00:11 |
freemangordon | night! | 00:12 |
freemangordon | Wizzup: I shall re-tag the kernel to do new build, right? | 09:23 |
Wizzup | freemangordon: yes | 10:04 |
freemangordon | Wizzup: BTW https://github.com/maemo-leste/droid4-linux/blob/maemo-6.1.y/scripts/package/builddeb#L205 goes into maintainer scripts substituted, is that on purpose? | 10:10 |
freemangordon | https://pastebin.com/T02RDqz6 | 10:13 |
freemangordon | this is what I mean | 10:13 |
Wizzup | freemangordon: hm, not so sure, that looks wrong | 10:28 |
freemangordon | mhm | 10:28 |
Wizzup | well, unless what you pasted is the postinstall | 10:28 |
freemangordon | umm... | 10:28 |
freemangordon | this is postinst script, yes | 10:28 |
freemangordon | so, this is on purpose? | 10:29 |
freemangordon | because postrm looks exactly the same, besides the switch argument | 10:29 |
Wizzup | freemangordon: does it say case postinst in ? | 10:30 |
freemangordon | obviously it works, but looks wrong to multiply the same script all over the place and decide on what to do based on hard-coded switch argument | 10:30 |
freemangordon | Wizzup: tis is postinst file in linux-image-omap | 10:31 |
freemangordon | *this | 10:31 |
freemangordon | it is generated by the code I linked ^^^ | 10:31 |
Wizzup | yes, but the postrm | 10:31 |
Wizzup | does it have the correct replacement? | 10:31 |
freemangordon | yes | 10:31 |
freemangordon | it is 'case "postrm" in' | 10:31 |
freemangordon | but looks weird to me, that's why I asked is that on purpose | 10:32 |
Wizzup | ok | 10:42 |
Wizzup | I agree it's weird, but I don't think it's worth fixing it up tbh | 10:43 |
Wizzup | but up to you :P | 10:43 |
freemangordon | well, if it is made like that on purpose, I will not 'fix' it | 10:47 |
freemangordon | Wizzup: heh | 11:18 |
freemangordon | https://github.com/dell/dkms/issues/37 | 11:18 |
freemangordon | this is exactly what we do it seems | 11:18 |
freemangordon | Wizzup: https://pastebin.com/izhxkhw9 | 12:53 |
freemangordon | the only side effect is that dkms modules are left on the filesystem on upgrade | 12:54 |
Wizzup | hm, do they ever go away eventually, or? | 12:55 |
freemangordon | no. But it seems debian does rm -rf on purge | 12:56 |
freemangordon | they do that in postrm | 12:56 |
Wizzup | ok | 12:57 |
freemangordon | well, not sure what we shall do | 12:57 |
freemangordon | maybe unpatch dkms and call it a day | 12:58 |
freemangordon | upstream dkms does 'dkms uninstall' | 12:58 |
Wizzup | what do you mean, unpatch dkms? | 12:59 |
freemangordon | see https://salsa.debian.org/debian/dkms/-/commit/dc9cfc36aa96351fd8329f547973f7b6aa3faa1d#32bdcca4e541258f53f056ad06abccc273e739db_20_20 | 12:59 |
freemangordon | but, don't know how would that affect VM | 12:59 |
freemangordon | no idea what to do | 13:08 |
Wizzup | freemangordon: I am not sure if I understand the root cause | 13:11 |
Wizzup | I thought you figured out the problem with dkms | 13:12 |
freemangordon | Wizzup: what happens is that dkms module gets uninstalled on kernel upgrade | 13:35 |
freemangordon | the reason for that is the change in debian dkms ^^^ | 13:35 |
freemangordon | upstream debian kernels hack around that by not calling /etc/kernel/prerm.d scripts | 13:35 |
freemangordon | on package upgrade that is | 13:36 |
freemangordon | what I don;t understand is how they remove kernel modules build by dkms on upgrdae | 13:37 |
Wizzup | maybe dkms does it itself | 13:38 |
freemangordon | no, because it is not called :) | 13:38 |
freemangordon | Wizzup: oh, they do not 'upgrade' | 13:45 |
freemangordon | they install diferent kernel packages | 13:46 |
freemangordon | linux-image-5.10.0-20-amd64 vs linux-image-5.10.0-21-amd64 | 13:46 |
freemangordon | that's why all this works | 13:47 |
arno11 | Wizzup: sicelo: i found a magic trick to reduce latency from 600ms to 30ms using pactl :) | 13:48 |
arno11 | calls are good | 13:48 |
arno11 | the only thing to improve is microphone level | 13:49 |
arno11 | a bit low even at 100 purcent | 13:49 |
Wizzup | arno11: great, I', happy that was it | 13:50 |
Wizzup | could you maybe hsare this trick too on the m-l? | 13:50 |
Wizzup | freemangordon: ah | 13:50 |
arno11 | Wizzup: sure | 13:51 |
sicelo | arno11: maybe there's another control to toggle? | 13:51 |
arno11 | sicelo: oh. maybe | 13:52 |
arno11 | i'll have a look | 13:52 |
arno11 | in fact we have already in leste specific pulseaudio modules to do lot of things automatically :) | 13:53 |
Wizzup | sicelo: 30ms sounds good to me tbh | 13:54 |
arno11 | in real call it seems a bit more than that but not bad compared to my andro phone | 13:58 |
Wizzup | sicelo: what other control are you thinking of? | 13:58 |
Wizzup | arno11: yeah, glad to know that was it | 13:58 |
Wizzup | it is what I told pavel at FOSDEM, probably just this | 13:58 |
sicelo | arno11: how does cmtspeech impact cpu ... did you notice any worse power consumption perhaps? | 14:00 |
arno11 | sicelo: cpu is running between 92 and 95 purcent with overclock | 14:01 |
freemangordon | IIRC it is NEON optimized, so someone else is loading the system | 14:01 |
freemangordon | PA most-probably | 14:02 |
arno11 | but it is still possible to navigate in the phone with no real troubles | 14:02 |
sicelo | i see. and current draw? | 14:02 |
arno11 | sicelo: no idea for current draw. i have to check | 14:03 |
arno11 | in cmtspeech source code there is a tweak to reduce cpu cons in cost of latency | 14:04 |
Wizzup | there's probably a lot more we can do once we have the initial working setup ... set up | 14:05 |
arno11 | indeed | 14:05 |
arno11 | and i'm not a pulseaudio professional lol | 14:05 |
arno11 | back in bit. testing current draw | 14:10 |
freemangordon | oh, it is actually upstream that changed it, not debian | 14:10 |
arno11 | sicelo: phone call drains arround 400mA | 14:17 |
arno11 | with overclock. maybe blocking cpu at 250 could decrease power cons a lot | 14:18 |
sicelo | yeah, i guess that's alright. i was interested in how much more if any, running cmtspeech in background affects the power consumption | 14:19 |
Wizzup | sicelo: during a call or not | 14:19 |
sicelo | i.e. when there's no active call | 14:19 |
Wizzup | right | 14:19 |
Wizzup | that should be none, but yeah | 14:19 |
arno11 | with no active call there is no cons with cmtspeech in fact | 14:20 |
sicelo | oh sweet | 14:20 |
sicelo | :-) | 14:20 |
arno11 | yes | 14:20 |
arno11 | :) | 14:20 |
arno11 | once call is starting that's another story | 14:20 |
arno11 | cmtspeech is registering other ofono event too | 14:22 |
arno11 | like receiving sms | 14:22 |
sicelo | as long as it doesn't impact power at 'rest', i guess consuming more during times of modem activity is acceptable | 14:22 |
arno11 | yes i think so | 14:23 |
Wizzup | better than none for now at least | 14:23 |
Wizzup | it's possible that cmt speech code is currently overly aggressive wrt latency because the pa latency msec thing wasn't set | 14:23 |
Wizzup | which it won't need to be, perhaps | 14:23 |
arno11 | hmm yes according to pactl pa latency is not set | 14:24 |
arno11 | but it's easy to configure through pactl | 14:24 |
Wizzup | I can find the libpulse call quite soon probably if you share which parts you set/chagne using pactl | 14:26 |
Wizzup | https://www.freedesktop.org/wiki/Software/PulseAudio/Documentation/Developer/Clients/LatencyControl/ | 14:27 |
arno11 | here is the trick | 14:28 |
arno11 | pactl load-module module-loopback latency_msec=1 | 14:28 |
arno11 | and then unload module immediately | 14:28 |
Wizzup | Pass a pa_buffer_attr struct in the buffer_attr parameter. In the fields of this struct make sure to initialize every single field to (uint32_t) -1, with the exception of tlength (for playback) resp. fragsize (for recording). | 14:29 |
Wizzup | I think the current API used by utils/pulse.c doesn't let you control the latency | 14:30 |
arno11 | good question | 14:31 |
arno11 | that's why latency seems dynamic on my device | 14:32 |
Wizzup | it uses pa_simple_new which doesn't let you set flags | 14:32 |
Wizzup | see my link above | 14:33 |
Wizzup | (I was looking at changing the source code of the libcmtspeechdata) | 14:33 |
arno11 | i see | 14:33 |
Wizzup | https://juho.tykkala.fi/Pulseaudio-and-latency this also let's you see the latency for streams it seems | 14:35 |
Wizzup | in any case IMO the right fix here is to have libcmtspeechdata request proper latency | 14:36 |
arno11 | it is the link where i found the tweak :) | 14:36 |
arno11 | yes agree | 14:36 |
Wizzup | The PA_STREAM_ADJUST_LATENCY flag is automatically set for all streams | 14:37 |
Wizzup | created with the simple API. | 14:37 |
Wizzup | ah, so maybe you just need to set fragsize correctly and everything else to -1 | 14:37 |
arno11 | yes not difficult | 14:37 |
arno11 | fragsize=15 you think ? | 14:40 |
Wizzup | arno11: can you try https://github.com/maemo-leste/libcmtspeechdata/commit/094db9323241cf1a9756677a339cc318c1091413 ? | 14:42 |
Wizzup | actually just a second | 14:42 |
arno11 | let me have a look | 14:43 |
Wizzup | hang on, I need to fix it up | 14:43 |
arno11 | ok | 14:44 |
Wizzup | I am trying to figure out if tlength and fragsize are in msec or some other fmt | 14:45 |
arno11 | ah ok | 14:45 |
Wizzup | the fdo link seems to suggest you need to use pa_usec_to_bytes to figure out what to set it to, but that seems to take some other struct | 14:46 |
Wizzup | oh | 14:46 |
Wizzup | this branch could work: https://github.com/maemo-leste/libcmtspeechdata/tree/latency-test ( or just https://github.com/maemo-leste/libcmtspeechdata/commit/ba3d714a953bd083ea0b61b282daaf81e954cb4a.patch ) | 14:48 |
Wizzup | arno11: ^^ :) | 14:49 |
Wizzup | if you could test it, I'd hope it would bring down latency without needing the extra hack | 14:49 |
Wizzup | if I understand correctly, if pavel set it to 1024, that's probably no good | 14:50 |
arno11 | ok i remember something about that | 14:50 |
arno11 | let me check | 14:50 |
Wizzup | it's a nice trap | 14:51 |
Wizzup | ;) | 14:51 |
arno11 | arghh i can't remove my hack ATM. i need to reverse things to be able to compare | 15:07 |
arno11 | and i have to go. i'll have time this evening to test stuff. ttyl | 15:08 |
Wizzup | aight ty | 15:09 |
Wizzup | good progress in any case | 15:09 |
freemangordon | Wizzup: I'll just hack dkms support in postinst file | 16:23 |
freemangordon | Wizzup: are we ok with having bash prerm script? | 17:46 |
freemangordon | #!/bin/bash, not #!/bin/sh that is | 17:46 |
Wizzup | freemangordon: I am | 18:24 |
freemangordon | I was able to avoid that | 18:32 |
freemangordon | Wizzup: this | 18:33 |
freemangordon | LMK if you are find with it. If yes, I'll issue a new kernel build | 18:34 |
freemangordon | https://pastebin.com/CY70DnxH | 18:35 |
freemangordon | this^^^ | 18:35 |
arno11 | Wizzup: unfortunatly the new pulse.c is not working: only cracking sound both ways with arround 5 sec latency | 21:03 |
arno11 | maybe a conflict with my settings | 21:03 |
arno11 | but really not sure | 21:03 |
arno11 | back to normal reinstalling libcmtspeech stock version | 21:06 |
Wizzup | ok | 21:38 |
Wizzup | freemangordon: lgtm | 21:38 |
freemangordon | Wizzup: already built it :) | 21:40 |
freemangordon | unfortunately iphb-dkms will have to be reinstalled after new kernel is as current kernel prerm script is broken | 21:41 |
freemangordon | Wizzup: BTW, I think we are ready for release. Unless I am missing something | 21:42 |
freemangordon | we have everything login session related in place, including auth agent | 21:42 |
freemangordon | power usage is back to normal | 21:42 |
freemangordon | etc | 21:43 |
buZz | does gpu accel work again? | 21:45 |
* buZz tries chromium | 21:46 | |
buZz | nope, still black | 21:46 |
buZz | weird btw, because 3D accel seems quite fine | 21:46 |
freemangordon | buZz: it has never worked in chromium | 21:47 |
buZz | yes it did | 21:47 |
freemangordon | no | 21:47 |
buZz | ok | 21:47 |
buZz | either way, prior to chimaera , chromium worked without --disable-gpu | 21:47 |
buZz | now it doesnt | 21:47 |
freemangordon | start chromium with --disable-gpu and disable hw accel from settings | 21:47 |
freemangordon | because before it was not detecting gpu at all | 21:48 |
buZz | ahh, thats what you mean, it auto-disabled prior? | 21:48 |
buZz | right, ok | 21:48 |
buZz | yeah its not compiling the shaders | 21:48 |
freemangordon | now it detects gpu and tries to use it | 21:48 |
freemangordon | even if it does, it still fails | 21:48 |
buZz | seems the driver misses some glsl libs? | 21:48 |
freemangordon | you can start it with --no-sanbox or whatever | 21:48 |
buZz | but havent dug too deep | 21:48 |
freemangordon | I played a bit | 21:48 |
freemangordon | buZz: chromium --no-sandbox | 21:51 |
freemangordon | with gpu enabled | 21:51 |
buZz | can it compile glsl then? | 21:51 |
freemangordon | yes | 21:51 |
freemangordon | but is slow as hell | 21:51 |
buZz | nice! :) | 21:51 |
freemangordon | [6611:6611:0415/225126.332080:ERROR:gl_display.cc(504)] EGL Driver message (Error) eglGetSyncValuesCHROMIUM: glXGetSyncValuesOML failed. | 21:51 |
freemangordon | lots of those errors | 21:52 |
buZz | https://github.com/KhronosGroup/OpenGL-Registry/blob/main/extensions/OML/GLX_OML_sync_control.txt#L53 | 21:52 |
buZz | lol ok | 21:52 |
buZz | might have some flag to ignore refresh rates | 21:52 |
freemangordon | was not able to find it | 21:53 |
freemangordon | didn't try hard though | 21:53 |
buZz | --disable-gpu-vsync maybe | 21:53 |
freemangordon | no idea | 21:53 |
freemangordon | chrome just crashed gpu here :) | 21:53 |
buZz | it seems pretty fast, beside the hanging on that error | 21:54 |
freemangordon | "--disable-frame-rate-limit --disable-gpu-vsync" | 21:54 |
freemangordon | according to google | 21:54 |
buZz | oh, and one other thing thats still broken since chimeara ; desktop widgets | 21:54 |
buZz | they keep getting removed when i 'sudo poweroff' and boot back up | 21:55 |
freemangordon | you have a widget that crashes | 21:55 |
buZz | and adding 'calendar widget' adds no widget plus removes all others | 21:55 |
freemangordon | that one then | 21:55 |
buZz | might be the calendar one yeah | 21:55 |
freemangordon | hildon-home keeps itself from a boot loop by disabling all widgets on one crashing | 21:55 |
buZz | alas; adding --disable-frame-rate-limit --disable-gpu-vsync gives the same errors | 21:55 |
freemangordon | well... | 21:56 |
buZz | freemangordon: do they ever get re-tested for crashing? or is it forever labelled 'bad' | 21:56 |
buZz | can i unlabel it? :) | 21:56 |
freemangordon | yes, but not sure how | 21:57 |
freemangordon | next restart maybe | 21:57 |
buZz | well, then 'calendar widget' is just completely broken ? :( | 21:59 |
buZz | sad, was so useful | 21:59 |
freemangordon | I will have a look at it tomorrow, just remind me | 21:59 |
buZz | i'd like to remove the broken one then, so i can -keep- widgets over reboots :P | 22:00 |
buZz | i'll try to remove it with apt / appmanager i guess | 22:00 |
buZz | whats sorta funny, readding the widgets one by one puts em back in exact location with exact config | 22:01 |
buZz | ok, removed through appmanager, gonna try a restart to see if widgets stay ^_^ | 22:02 |
buZz | omg yes! | 22:06 |
buZz | w00t | 22:06 |
buZz | :D :D | 22:06 |
buZz | so just -having it installed- and not visible causes this , impressive | 22:06 |
freemangordon | I have some ideas what might be | 22:07 |
freemangordon | hmm, no, not that | 22:08 |
freemangordon | buZz: /usr/bin/hildon-home: symbol lookup error: /usr/lib/x86_64-linux-gnu/hildon-desktop/cal-home-widget.so: undefined symbol: time_get_local | 22:13 |
freemangordon | :) | 22:14 |
buZz | looks like a symbol that should have a definition? | 22:14 |
freemangordon | missing -l flag | 22:14 |
freemangordon | sec | 22:14 |
buZz | ooooo | 22:14 |
buZz | freemangordon: you're too early, you said tomorrow! | 22:15 |
buZz | :) | 22:15 |
freemangordon | it turned out that tomorrow I will be MIA | 22:15 |
buZz | oo | 22:15 |
buZz | freemangordon in concert! | 22:15 |
freemangordon | no | 22:15 |
buZz | MIA is such a cool artist :D | 22:15 |
freemangordon | heh | 22:16 |
buZz | hehe i know, just kidding ;) | 22:16 |
Wizzup | arno11: maybe try it only for source then | 22:22 |
Wizzup | arno11: ah you siad both ways | 22:22 |
Wizzup | freemangordon: regarding release, let's see, just a sec | 22:25 |
freemangordon | buZz: https://pastebin.com/x9GCvQdG | 22:25 |
freemangordon | feel free to make a PR with that patch, I am going to watch Ronnie :) | 22:26 |
freemangordon | ttyl | 22:26 |
buZz | :) enjoy! | 22:26 |
buZz | thanks! | 22:26 |
Wizzup | buZz: do you want me to commit and build? | 22:31 |
buZz | if you want to? i was still searching for how ;) | 22:31 |
Wizzup | building now | 22:35 |
buZz | woot :) it tend to use the maemo calendar a lot, even with the broken layout and colors :P | 22:36 |
buZz | i tend* | 22:36 |
sicelo | buZz: the widget positions on desktop are stored in gconf. that's why they go back to original location on being added again :-) | 22:39 |
buZz | fancy :) | 22:39 |
Wizzup | buZz: should be in repos | 22:39 |
* buZz tries | 22:40 | |
buZz | w00t, instantly even adds it back to desktop :D | 22:42 |
buZz | \o/ | 22:42 |
buZz | it works, Wizzup :) | 22:43 |
buZz | oo also new kernel and new hildon-base | 22:45 |
buZz | arf | 22:49 |
buZz | rebooted for the new kernel -> all widgets gone again | 22:49 |
Wizzup | maybe you have another widget with this problem | 22:50 |
buZz | i just have 3 widgets installed ; command exec, personal IP , and the calendar one | 22:51 |
buZz | they all are on 1 desktop page and work now | 22:51 |
buZz | the command exec is 8x on desktop even , 8 different commands | 22:51 |
buZz | maybe something else causes the same behaviour? like a broken statusmenu plugin? (afaik orientation lock is still installed here) | 22:52 |
sicelo | i hope it's not my widgets, although I'll admit i haven't tested them on chimaera | 22:54 |
buZz | i'm not sure, those arent in appmanager? | 22:56 |
sicelo | iow, personal ip and cmd exec ... i ported them and should maintain them | 22:56 |
buZz | iow? | 22:56 |
buZz | the personal ip and cmd exec ones i use, and they seem to work totally fine | 22:57 |
sicelo | "in other words" .. | 22:57 |
buZz | oh, right, ok | 22:57 |
sicelo | nice to hear they work good :-) | 22:57 |
buZz | :) | 22:58 |
sicelo | i haven't run Leste in a while now ... haven't had much time to spare, although arno11's work is encouraging me to 'return' asap | 22:58 |
buZz | i removed orientationlock and trying a powercycle now | 22:59 |
Wizzup | could also be the ip address one | 22:59 |
buZz | but its working? | 22:59 |
buZz | oh wait | 23:00 |
buZz | maybe hildon -remembered- that the calendar one is broken | 23:00 |
buZz | and doesnt retest ever | 23:00 |
buZz | powercycle now removes em all again | 23:00 |
sicelo | if ip address or cmd exec cause issues, please feel free to ping. I'm still willing to maintain them | 23:01 |
sicelo | buZz: look at or enable logs | 23:01 |
buZz | just gonna try a hunch | 23:01 |
buZz | (removing cal-home-widget, to see if hildon has some memory about this) | 23:02 |
buZz | hahahaha it does | 23:05 |
buZz | so uninstalling the functional cal-home-widget made hildon assume they're all good , so displays them as before poweroff | 23:05 |
buZz | but, adding it back makes hildon remember the old crashing state of that 1 plugin, removing -all- from desktop on reboot | 23:06 |
buZz | amazing UI nokia :D | 23:06 |
sicelo | sounds weird | 23:06 |
buZz | i guess 'restore factory defaults' or something would fix it | 23:07 |
sicelo | maybe need gconf cleanup | 23:07 |
buZz | yeah likely | 23:07 |
buZz | or wherever it stored it | 23:07 |
Wizzup | I think what would make more sense is that you run hildon-home in debug mode | 23:07 |
Wizzup | and see if any other errors occur | 23:07 |
Wizzup | this might be based on too many assumptions on your side | 23:07 |
buZz | right, ok, where is it launched? | 23:07 |
Wizzup | you first have to stop it using dsmetool and then start it from gdb or so | 23:08 |
Wizzup | https://leste.maemo.org/Debugging | 23:08 |
Wizzup | https://leste.maemo.org/Debugging#Dealing_with_Maemo_Launcher_.2F_Maemo_Invoker | 23:08 |
freemangordon | buZz: /usr/sbin/dsmetool -k /usr/bin/hildon-home | 23:09 |
freemangordon | sudo apt install hildon-home-dbgsym | 23:09 |
buZz | i'll install gdb first :P | 23:09 |
freemangordon | gdb maemo-summoner | 23:09 |
freemangordon | r /usr/bin/hildon.home.launch | 23:09 |
buZz | go to your movie ;) | 23:10 |
freemangordon | ummm.... | 23:10 |
freemangordon | r /usr/bin/hildon-home.launch | 23:10 |
buZz | :) | 23:10 |
freemangordon | movie? Ronnie is not a movie ;) | 23:10 |
freemangordon | https://en.wikipedia.org/wiki/Ronnie_O%27Sullivan :) | 23:11 |
buZz | oh, i thought Reagan, the actor | 23:11 |
freemangordon | no | 23:11 |
buZz | hildon-home-dbg , it seems? | 23:14 |
freemangordon | could be, yeah | 23:14 |
Wizzup | dbgsym probably, use apt-cache search | 23:14 |
buZz | hildon-home-dbg/stable 0.3.79+m7 armhf | 23:15 |
buZz | yeah | 23:15 |
Wizzup | huh, that one is still on dbg eh? | 23:15 |
freemangordon | yeah | 23:15 |
freemangordon | compat is 5 | 23:15 |
buZz | but, how can i run hildon-home through gdb to read debug msgs, when this only happens on boot? | 23:21 |
freemangordon | buZz: see what I wrote ^^^ | 23:22 |
freemangordon | start with "/usr/sbin/dsmetool -k /usr/bin/hildon-home" | 23:23 |
freemangordon | then "G_MESSAGES_DEBUG=all gdb maemo-summoner" | 23:23 |
freemangordon | then in gdb: "r /usr/bin/hildon.home.launch" | 23:23 |
buZz | yeah but, then the issue is already gone? | 23:24 |
freemangordon | sorry, r /usr/bin/hildon-home.launch | 23:24 |
freemangordon | what do you mean? | 23:24 |
buZz | because that would be the second launch of hildon-home since boot | 23:24 |
freemangordon | and? | 23:24 |
freemangordon | so what? | 23:24 |
buZz | i thought i was trying to find why it removes the widgets :P | 23:24 |
freemangordon | does it display them on second launch? | 23:24 |
Wizzup | probably because is crashes and then restores safe state | 23:25 |
freemangordon | mhm | 23:25 |
Wizzup | why is why you're starting it like this | 23:25 |
Wizzup | which is why* | 23:25 |
buZz | ok, well, lets see | 23:25 |
buZz | :D | 23:25 |
freemangordon | buZz: lets try as I say, if we cannot repro, will try other things | 23:26 |
buZz | i'm installing it with hildon-home in gdb , to see if any errors directly appear, then restarting and seeing if its giving more new errors | 23:26 |
freemangordon | hmm? | 23:26 |
freemangordon | please, do what I wrote ^^^ | 23:26 |
freemangordon | do not restart anything | 23:27 |
buZz | but, eh | 23:27 |
buZz | ok | 23:27 |
buZz | only lacked G_MESSAGES_DEBUG=all :) | 23:28 |
freemangordon | does not matter | 23:28 |
freemangordon | and? | 23:28 |
buZz | no errors on adding the calendar widget back | 23:29 |
freemangordon | sure | 23:29 |
buZz | and it displays | 23:29 |
freemangordon | now, break into debugger (ctrl-c) and quit | 23:29 |
buZz | but how can i now see the behaviour its showing on boot, of removing all the widgets? | 23:29 |
freemangordon | buZz: patience :) | 23:30 |
freemangordon | do as I say | 23:30 |
freemangordon | "now, break into debugger (ctrl-c) and quit" | 23:30 |
buZz | 'quit anyway (y/n) | 23:30 |
freemangordon | yes | 23:30 |
buZz | done | 23:30 |
freemangordon | now start it again in gdb | 23:30 |
freemangordon | G_MESSAGES_DEBUG=all gdb maemo-summoner | 23:31 |
freemangordon | etc | 23:31 |
buZz | first line 'hildon-home: program did not exit properly on the previous session. All plugins will be disabled' | 23:32 |
freemangordon | mhm, that's the issue | 23:32 |
freemangordon | sec | 23:32 |
freemangordon | buZz: what is the result of "find / -name hildon-home.stamp" | 23:34 |
freemangordon | or rather ls -al /tmp/hildon-desktop/ | 23:35 |
buZz | eh, guess i should have hildon-home running then , 1 moment ;) | 23:36 |
freemangordon | wait | 23:37 |
Wizzup | no | 23:37 |
freemangordon | how do you restart the device? | 23:37 |
buZz | eh, often just sudo reboot/sudo poweroff | 23:37 |
freemangordon | so, what is the result of ls -al /tmp/hildon-desktop? | 23:37 |
buZz | sometimes i take the time to press powerbutton and tap the button | 23:37 |
freemangordon | so? | 23:39 |
buZz | there's a 1 byte hildon-home.stamp | 23:39 |
freemangordon | buZz: please... | 23:39 |
buZz | timestamped for just-now | 23:39 |
freemangordon | ls -al /tmp/hildon-desktop | 23:40 |
freemangordon | just give me the output | 23:40 |
buZz | total 16 | 23:40 |
buZz | drwxr-xr-x 2 user user 4096 Apr 15 23:39 . | 23:40 |
buZz | drwxrwxrwt 7 root root 4096 Apr 15 23:40 .. | 23:40 |
buZz | -rw-r--r-- 1 user user 0 Apr 15 23:39 desktop-started.stamp | 23:40 |
buZz | -rw-r--r-- 1 user user 1 Apr 15 23:39 hildon-home.stamp | 23:40 |
buZz | -rw-r--r-- 1 user user 1 Apr 15 23:39 status-menu.stamp | 23:40 |
buZz | here you go :D | 23:40 |
freemangordon | now rm hildon-home.stamp | 23:41 |
buZz | done | 23:41 |
freemangordon | and then /usr/sbin/dsmetool -t /usr/bin/hildon-home | 23:41 |
buZz | oh, -t | 23:42 |
buZz | dangit, did -k | 23:42 |
freemangordon | and then add calendar widget again (if it is missing) | 23:42 |
freemangordon | no problem | 23:42 |
freemangordon | it kills nothing if there is nothing to kill :) | 23:42 |
buZz | ehhhh, 'desktop menu' doesnt work after -k (and then -t) | 23:43 |
freemangordon | what does it mean "does not work"? | 23:44 |
buZz | tapping it doesnt display a menu | 23:44 |
buZz | tapping 'done' does close it | 23:44 |
freemangordon | hmm | 23:44 |
freemangordon | ok, do /usr/sbin/dsmetool -b | 23:45 |
freemangordon | this will reboot the device | 23:45 |
buZz | ok | 23:45 |
buZz | thats the same as powermenu 'switch off' ? | 23:45 |
freemangordon | almost | 23:46 |
freemangordon | will reboot though | 23:46 |
buZz | seems similar , insta-black and whiteled directly | 23:46 |
freemangordon | yes, similar but not same | 23:47 |
buZz | ok, it booted back up now | 23:48 |
freemangordon | is widget there? | 23:49 |
buZz | non of the widgets are | 23:49 |
freemangordon | which widgets do you have? | 23:49 |
freemangordon | calendar, ip and? | 23:49 |
buZz | 22:51:09 < buZz> i just have 3 widgets installed ; command exec, personal IP , and the calendar one | 23:50 |
freemangordon | ok | 23:50 |
freemangordon | i am missing exec one | 23:50 |
buZz | removing cal-home-widget fixed it prior btw | 23:50 |
freemangordon | I have calendar and ip widgets on my d4 and VM, no issues there | 23:51 |
buZz | as in , apt remove | 23:51 |
buZz | curious :) | 23:51 |
freemangordon | buZz: could you pastebin /var/log/maemo/dsme.log? | 23:52 |
freemangordon | buZz: scratch that, we have to do one more thing before that | 23:56 |
freemangordon | please re-add cal widget and then again reboot with "/usr/sbin/dsmetool -b" | 23:56 |
buZz | ok | 23:57 |
buZz | rebooting | 23:57 |
buZz | and then just last entries since boot from dsme.log? | 23:57 |
freemangordon | mhm | 23:58 |
freemangordon | well, lets see if widget will be there | 23:58 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!