| 01:32.34 | *** join/#maemo-ssu M4rtinK (~M4rtinK@mail.melf.eu) |
| 02:06.08 | *** join/#maemo-ssu amiconn_ (amiconn@rockbox/developer/amiconn) |
| 02:12.08 | *** join/#maemo-ssu M13 (~Miranda@83.149.38.204) |
| 02:48.48 | *** join/#maemo-ssu ekze-nyan (~nyan@bakaekze.ru) |
| 02:50.20 | *** join/#maemo-ssu LaoLang_cool (~LaoLang_c@221.226.175.139) |
| 03:17.44 | *** join/#maemo-ssu taziff1 (~Taziff@cyr108.internetdsl.tpnet.pl) |
| 03:23.26 | *** join/#maemo-ssu taziff (~Taziff@cyr108.internetdsl.tpnet.pl) |
| 04:52.21 | *** join/#maemo-ssu taziff1 (~Taziff@cyr108.internetdsl.tpnet.pl) |
| 05:18.08 | *** join/#maemo-ssu fruct (~fruct@cl122-184-119-212.cl.metrocom.ru) |
| 05:56.51 | *** join/#maemo-ssu taziff (~Taziff@cyr108.internetdsl.tpnet.pl) |
| 06:07.48 | *** join/#maemo-ssu taziff1 (~Taziff@cyr108.internetdsl.tpnet.pl) |
| 06:18.16 | *** join/#maemo-ssu freemangordon (~freemango@130-204-50-168.2074221835.ddns.cablebg.net) |
| 06:46.29 | *** join/#maemo-ssu taziff (~Taziff@cyr108.internetdsl.tpnet.pl) |
| 06:53.43 | *** join/#maemo-ssu taziff1 (~Taziff@cyr108.internetdsl.tpnet.pl) |
| 07:04.49 | *** join/#maemo-ssu taziff (~Taziff@cyr108.internetdsl.tpnet.pl) |
| 07:18.03 | *** join/#maemo-ssu LaoLang_cool (~LaoLang_c@221.226.175.141) |
| 07:48.15 | *** join/#maemo-ssu jonwil (~jonwil@27-33-137-199.static.tpgi.com.au) |
| 07:48.44 | freemangordon | jonwil: https://gitorious.org/community-ssu/hildon-im-vkbrenderer3 :P |
| 07:49.05 | freemangordon | hildon-input-method-plugins almost ready |
| 07:49.18 | *** join/#maemo-ssu BCMM (~ben@unaffiliated/bcmm) |
| 07:50.27 | jonwil | nice :) |
| 07:50.46 | freemangordon | jonwil: BTW I think I found a problem with imlayouts header file |
| 07:50.52 | jonwil | yes what problem? |
| 07:51.17 | freemangordon | https://gitorious.org/community-ssu/hildon-im-vkbrenderer3/blobs/master/src/imlayout_vkb.h#line59 |
| 07:51.21 | jonwil | Also does the vkbrenderer3 look and work the same as the stock widget or are there differences? |
| 07:51.42 | freemangordon | the definitions of the shift, tab, etc should be 2 bytes |
| 07:51.53 | freemangordon | jonwil: it is exactly as stock |
| 07:51.58 | jonwil | ok |
| 07:52.03 | freemangordon | thanks to HexRays :D |
| 07:52.14 | freemangordon | it is RE, not a rewrite |
| 07:52.16 | jonwil | :) |
| 07:52.44 | freemangordon | so, what about those defines, where did you get them from? |
| 07:52.46 | jonwil | if you think there are errors in the imlayout_vkb.h file, feel free to change it |
| 07:52.52 | jonwil | I think they were RE |
| 07:53.23 | freemangordon | hmm, ok. Do you want to help and make -dev package for imlayouts? |
| 07:54.08 | jonwil | Feel free to make one, I dont have enough time to do it myself |
| 07:54.19 | freemangordon | yeah, right :D |
| 07:54.20 | freemangordon | ok |
| 07:55.40 | jonwil | that and my Gentoo box seems to be broken for reasons I dont know |
| 08:04.05 | jonwil | ok, now I know why my gentoo box is broken and can fix it |
| 08:04.12 | jonwil | Still dont have N900 time though :P |
| 08:04.27 | jonwil | Good job on the hildon-input-method stuff btw |
| 08:08.15 | freemangordon | thanks |
| 08:09.59 | jonwil | btw, hildon-im-common-virtual-settings, osso-applet-textinput and hildon-input-method-configurator would be worth reverse engineering if you are reverse engineering things |
| 08:10.04 | jonwil | All 3 have x86 packages |
| 08:14.06 | jonwil | those 3 packages along with libi18n-locale-resolver0, libi18n and locale-resolver-data are the 6 packages that are concerned with the language/input language/etc selection |
| 08:14.22 | jonwil | i.e. everything connected to languages, locales, keyboard layouts (hw and virtual) etc |
| 08:16.18 | jonwil | The libi18n stuff, everything comes from locale-resolver-data so one would need to figure out the format of those data files :) |
| 08:20.36 | jonwil | so reverse engineer libi18n-locale-resolver0 might help with that :) |
| 08:22.47 | freemangordon | yeah, but those are noot needed for portrait VKB aiui |
| 08:24.40 | jonwil | no but its still nice to have them because people want to add new languages to software :P |
| 08:24.42 | freemangordon | WTF has happened with gitorious? |
| 08:24.54 | freemangordon | jonwil: feel free to RE those :P |
| 08:25.02 | jonwil | heh :P |
| 08:25.17 | jonwil | but yeah vkb stuff is good |
| 08:26.06 | freemangordon | jonwil: BTW I will upload IDA databases on merlin1991's server once I am ready with him-plugins, you may want to look at it |
| 08:26.37 | jonwil | not really, I have no interest in vkb stuff anymore if you have made 1:1 clone of it |
| 08:27.00 | jonwil | I was sort of interested in the input method language stuff I mentioned |
| 08:27.09 | jonwil | but I am not anymore |
| 08:27.23 | freemangordon | jonwil: but it still lacks portrait support. And I am not good enough with GTK to implement it properly |
| 08:27.24 | jonwil | since I doubt anyone is still interested in changing the list of valid languages in those binaries |
| 08:27.37 | jonwil | my GTK skills are worse :P |
| 08:27.43 | freemangordon | aah, I see |
| 08:27.55 | freemangordon | well, will pester nikolai then :D |
| 08:28.53 | jonwil | but yeah now I can remove the input stuff from my lists :) |
| 08:29.02 | freemangordon | fuck, gitorious is down :( |
| 08:29.11 | jonwil | since you did basically what I would have done had I done it (1:1 clone of the vkb renderer) |
| 08:29.36 | freemangordon | hmm, someone should edit wiki page with the closed stuff |
| 08:30.07 | jonwil | http://wiki.maemo.org/Fremantle_closed_packages seems pretty close to accurate AFAIK |
| 08:30.23 | jonwil | Whats missing from it? |
| 08:30.28 | freemangordon | 3 packages to be removed - vkbrenderer, western_fkb, western_fkb_common |
| 08:30.44 | freemangordon | jonwil: not saying something is missing |
| 08:30.56 | freemangordon | but the above 3 are no more closed source |
| 08:31.02 | jonwil | http://wiki.maemo.org/Fremantle_closed_packageswhich 3? |
| 08:31.07 | jonwil | which 3? |
| 08:31.14 | jonwil | oh those 3 :) |
| 08:31.17 | freemangordon | <PROTECTED> |
| 08:31.19 | jonwil | didn't see that |
| 08:31.46 | jonwil | ok, someone needs to put the git urls under "Source Code URL" |
| 08:32.12 | jonwil | and set FOSS Replacement as appropriate |
| 08:32.50 | jonwil | get me the right URLs and I will edit that page :) |
| 08:34.24 | freemangordon | <PROTECTED> |
| 08:34.33 | freemangordon | for vkbrenderer |
| 08:34.46 | freemangordon | https://gitorious.org/community-ssu/hildon-input-method-plugins |
| 08:34.54 | freemangordon | for the other two packages |
| 08:35.26 | freemangordon | that one is still not ready, but it will be in a couple of days, so NP to edit the page now |
| 08:37.20 | freemangordon | gitorious still down :( |
| 08:38.36 | *** join/#maemo-ssu user3003 (~andre@ip-89-176-24-140.net.upcbroadband.cz) |
| 08:39.19 | *** join/#maemo-ssu andre__ (~andre@Maemo/community/bugmaster/andre) |
| 08:40.27 | freemangordon | jonwil: are you sure package names there are correct? |
| 08:40.41 | jonwil | which package names? |
| 08:41.14 | freemangordon | hmm my bad, they are |
| 08:43.08 | freemangordon | hildon-im-fkb, libhildon-im-western-plugin-common3 and libhildon-im-vkbrenderer3 are the actual packages |
| 08:45.10 | jonwil | so you are only cloning some of the packages of hildon-input-method-plugins and not all of them? |
| 08:45.47 | freemangordon | jonwil: do you know which packages are produced by source hildon-input-method-plugins? |
| 08:46.00 | jonwil | yes you can find that out from the package files |
| 08:46.15 | freemangordon | besides hildon-im-fkb, libhildon-im-western-plugin-common |
| 08:46.32 | freemangordon | * libhildon-im-western-plugin-common3 |
| 08:46.57 | freemangordon | I can re the others too, NP with that |
| 08:47.15 | freemangordon | just gimme the names |
| 08:47.23 | jonwil | the ones I can see are hildon-im-virtual-keyboard-layouts |
| 08:47.34 | jonwil | hildon-im-plugin-base-settings |
| 08:47.42 | jonwil | hildon-im-fkb |
| 08:47.54 | jonwil | hildon-im-keyboard-assisant-scv |
| 08:48.04 | jonwil | hildon-im-keyboard-assistant-scv |
| 08:48.13 | jonwil | hildon-im-common-virtual-settings |
| 08:48.21 | jonwil | hildon-input-method-plugins-western |
| 08:48.39 | jonwil | libhildon-im-western-plugin-common3 |
| 08:48.52 | jonwil | hildon-im-keyboard-assistant |
| 08:52.30 | jonwil | the ones that exist in x86 are |
| 08:52.37 | jonwil | hildon-im-common-virtual-settings |
| 08:52.50 | jonwil | hildon-im-fkb |
| 08:53.05 | jonwil | hildon-im-plugin-base-settings |
| 08:53.23 | jonwil | hildon-im-virtual-keyboard-layouts |
| 08:53.40 | jonwil | libhildon-im-western-plugin-common3 |
| 08:54.39 | jonwil | hildon-im-virtual-keyboard-layouts is non-binary |
| 08:54.56 | freemangordon | hmm, lots of work |
| 08:55.12 | jonwil | as is hildon-im-plugin-base-settings |
| 08:55.31 | freemangordon | any clue what those -scv packages are? |
| 08:55.53 | jonwil | you are doing hildon-im-fkb and libhildon-im-western-plugin-common3 |
| 08:55.59 | freemangordon | yep |
| 08:56.19 | jonwil | so the only x86 package we have for hildon-input-method-plugins that remains is hildon-im-common-virtual-settings |
| 08:56.25 | jonwil | the rest are armel only |
| 08:56.30 | jonwil | and therefore presumably out of your scope |
| 08:57.00 | jonwil | The -scv package is for the symbols keyboard |
| 08:57.09 | freemangordon | well, it depends, if those are small libs, I am ok to RE them from ARM without decompilers |
| 08:57.22 | jonwil | and contains both the layout files for it and the the binary for it |
| 08:57.58 | freemangordon | are you sure? as vkbrenderer3 draws symbols KBD too iirc |
| 08:58.05 | jonwil | Its not the renderer |
| 08:58.14 | jonwil | it handles the layouts and talking to the renderer |
| 09:00.20 | freemangordon | yeah, I see |
| 09:00.41 | freemangordon | arm binary is 18k, could be possible to re it |
| 09:01.10 | freemangordon | bu why those are missing on x86? |
| 09:01.17 | jonwil | no idea |
| 09:01.19 | jonwil | so on x86 you have /usr/lib/hildon-input-method/western_common_vim_settings.so |
| 09:04.03 | jonwil | and on ARM you have /usr/lib/hildon-input-method/hildon_keyboard_assistant_scv.so, /usr/lib/hildon-input-method/hildon_keyboard_assistant.so, /usr/lib/hildon-input-method/hildon_im_keyboard_monitor.so and /usr/bin/hildon-im-xkbtool |
| 09:04.24 | jonwil | that covers everything thats in hildon-input-method-plugins |
| 09:05.09 | freemangordon | only those 5? |
| 09:05.15 | jonwil | yes |
| 09:05.20 | jonwil | those plus the ones you are also doing |
| 09:05.27 | freemangordon | good, will keep the list somewhere |
| 09:07.15 | jonwil | bah, gitorius is still down :( |
| 09:07.21 | freemangordon | yes :( |
| 09:11.55 | *** join/#maemo-ssu taziff (~Taziff@cyr108.internetdsl.tpnet.pl) |
| 09:12.03 | jonwil | Going to see what other hildon-input-method related binaries exist that you could consider cloning whilst you are doing it :P |
| 09:12.16 | jonwil | in case any are worth it for portrait vkb or other CSSU things |
| 09:21.46 | jonwil | usr/bin/hildon-input-method-configurator.bin seems quite important to h-i-m stuffs :) |
| 09:21.53 | jonwil | and there is x86 version :) |
| 09:22.07 | freemangordon | jonwil: hold on, I want to finish with what is really needed and return to CSSU kernel :P |
| 09:22.14 | jonwil | :P |
| 09:23.11 | jonwil | yeah its not worth cloning libimlayouts0 for example |
| 09:23.18 | jonwil | now that we have -dev package for it |
| 09:23.51 | freemangordon | do we? |
| 09:23.59 | freemangordon | where is it? |
| 09:24.01 | jonwil | well we have a .h for it |
| 09:24.05 | freemangordon | aah, ok |
| 09:24.07 | jonwil | and can make a dev package |
| 09:24.10 | jonwil | :P |
| 09:24.29 | freemangordon | I know, that is what i asked an hour ago :P |
| 09:24.40 | jonwil | so the only things worth doing for hildon-input-method are |
| 09:25.20 | jonwil | x86 /usr/lib/hildon-input-method/western_common_vim_settings.so, /usr/bin/hildon-input-method-configurator.bin and arm /usr/lib/hildon-input-method/hildon_keyboard_assistant_scv.so, /usr/lib/hildon-input-method/hildon_keyboard_assistant.so, /usr/lib/hildon-input-method/hildon_im_keyboard_monitor.so and /usr/bin/hildon-im-xkbtool |
| 09:25.37 | freemangordon | yes |
| 09:25.41 | jonwil | How many of those you want to clone is up to you of course :) |
| 09:25.45 | freemangordon | i'll update my list |
| 09:25.58 | jonwil | and no I have no idea exactly what those do |
| 09:26.14 | freemangordon | well, I won't touch ARMs until I have plenty of free time |
| 09:26.22 | jonwil | :) |
| 09:26.30 | jonwil | do the 2 x86 ones then :P |
| 09:26.57 | freemangordon | will see, lemme finish with the important stuff first, I have about 20-30 functions remaining |
| 09:27.01 | jonwil | :) |
| 09:27.02 | merlin1991 | damn gitorious, always down when one wants todo stuff |
| 09:27.05 | freemangordon | yes |
| 09:27.36 | merlin1991 | in other words, I just wanted todo several git pulls and prepare a new -testing release |
| 09:27.54 | jonwil | so what do hildon-im-fkb and libhildon-im-western-plugin-common3 do? |
| 09:28.45 | freemangordon | <PROTECTED> |
| 09:29.07 | freemangordon | <PROTECTED> |
| 09:29.13 | jonwil | ok :) |
| 09:29.31 | freemangordon | hildon-im-fkb also handles word prediction |
| 09:29.42 | freemangordon | it is huge, 60k or so |
| 09:30.25 | freemangordon | thanks god we have x86 and HexRays, that would be nearly impossible to RE from ARM |
| 09:30.49 | jonwil | yeah :) |
| 09:31.49 | jonwil | If I had x86 version of icd stuff, systemui stuff, csd stuff or connui stuff, I would be seriously far along with reverse engineering those |
| 09:31.51 | jonwil | :) |
| 09:35.47 | jonwil | so yeah hildon_keyboard_assistant_scv.so is what draws the rest of the symbols keyboard (and calls the vkbrenderer to render the buttons) |
| 09:38.10 | *** join/#maemo-ssu ivgalvez (~ivgalvez@89.140.113.138.static.user.ono.com) |
| 09:39.30 | ivgalvez | hi freemangordon and jonwil |
| 09:40.48 | jonwil | hi |
| 09:41.07 | ivgalvez | looking at your comments about closed packages in Fremantle, is there any plans to introduce in CSSU any of the already existing replacements for widgets? |
| 09:41.18 | jonwil | which widgets are those? |
| 09:41.27 | freemangordon | ivgalvez: hi |
| 09:41.36 | ivgalvez | for example calendar, media player, ... |
| 09:41.44 | ivgalvez | I think most of them are from Nicolia |
| 09:41.47 | ivgalvez | Nicolai |
| 09:42.03 | jonwil | I guess it depends on if those are replacements or just new apps for the same thing |
| 09:42.14 | jonwil | Look at what happened with the new camera-ui that wasnt a proper replacement |
| 09:42.23 | jonwil | people complained about it AFAIK |
| 09:42.44 | ivgalvez | people complaining for bugs in testing releases? |
| 09:42.46 | jonwil | but yeah if there are genuine replacements for closed packages that are not listed here http://wiki.maemo.org/Fremantle_closed_packages then they should be listed here |
| 09:42.47 | ivgalvez | funny |
| 09:43.10 | jonwil | no, not complaints about bugs, complaints that the new camera-ui did not do what the old camera-ui did |
| 09:43.20 | jonwil | i.e. it wasnt a replacement, it was a new app that happened to do a few of the same things |
| 09:43.44 | ivgalvez | people complains, it's the nature |
| 09:43.49 | jonwil | whereas the calculator replacement is just that, its a replacement with possible new features :) |
| 09:43.51 | ivgalvez | Gnome 3, KDE 4 |
| 09:44.01 | merlin1991 | well currently there is a bug in the calculator :/ |
| 09:44.19 | ivgalvez | fixable now, as it's OSS |
| 09:45.10 | ivgalvez | also jonwil, don't forget that complainers make more noise than satisfied users |
| 09:45.16 | jonwil | true :) |
| 09:45.32 | ivgalvez | me for example I'm satisfied with camera replacement since day 1 |
| 09:45.44 | ivgalvez | I haven't shout it though |
| 09:45.50 | chem|st | ivgalvez: so you don't use it? |
| 09:45.56 | ivgalvez | I use it a lot |
| 09:46.15 | ivgalvez | it's by all means superior to Nokia |
| 09:46.25 | chem|st | ivgalvez: then you did not find bugs others did... |
| 09:46.28 | jonwil | btw, I wonder if anyone will make any use of my bluetooth-sysinfo and wl1251-cal replacements for anything useful... |
| 09:46.35 | chem|st | ivgalvez: true at least |
| 09:46.58 | ivgalvez | chem|st: I remember a few quirks with the first iterations |
| 09:47.10 | *** join/#maemo-ssu M4rtinK (~M4rtinK@mail.melf.eu) |
| 09:47.15 | ivgalvez | but that's normal, being testing stuff |
| 09:47.18 | chem|st | ivgalvez: speaking of day 1 |
| 09:47.42 | freemangordon | chem|st: what exactly bugs exists in latest nikocam? |
| 09:48.51 | chem|st | freemangordon: well it segfaults quiet often but that might be another issue |
| 09:49.10 | ivgalvez | works for me (tm) |
| 09:49.13 | chem|st | often for $user that is |
| 09:49.29 | freemangordon | segfaults? |
| 09:49.48 | freemangordon | any chance to get a coredump? |
| 09:50.10 | ivgalvez | provide a link to bugzilla, otherwise it didn't happen ;) |
| 09:50.18 | chem|st | if waking up from standby it does 1/10 something like 1/30 would be for users I guess |
| 09:50.36 | freemangordon | chem|st: which version is that? |
| 09:50.47 | chem|st | cssu testing one |
| 09:50.53 | freemangordon | T4? |
| 09:51.08 | chem|st | freemangordon: as I said it might be another issue as my device plays up from time to time |
| 09:51.24 | freemangordon | chem|st: do me a favor and capture a coredump please |
| 09:51.38 | chem|st | freemangordon: I do not use it as productive phone currently so I do not have it with me sry |
| 09:51.45 | freemangordon | chem|st: well, you said it is not stable and segfaults |
| 09:52.02 | freemangordon | now you're saying that might be it is stable |
| 09:52.11 | chem|st | freemangordon: and I do not know if it is my device or the camapp |
| 09:52.24 | freemangordon | i see |
| 09:52.38 | chem|st | I can tell you how to get there if that happens on no other 3 devices I guess it is my hardware |
| 09:52.58 | freemangordon | ok, tell me |
| 09:53.13 | ivgalvez | CSSU Testing is much more stable than what it should be |
| 09:53.22 | ivgalvez | there's almost not real testing there |
| 09:53.25 | freemangordon | chem|st: that is even better than coredump |
| 09:55.12 | chem|st | settings all auto and 2 seconds reviewer, take picture in fullauto, tab gallery in reviewer; wait some 10 seconds to let the camapp go to standby for real; go back - BOOM |
| 09:55.38 | chem|st | maybe it is not 1/10... maybe it is 1/15-20 but still |
| 09:55.57 | freemangordon | chem|st: ok, will try it |
| 09:55.59 | chem|st | most of the time the app does not get idle long enough |
| 09:56.24 | chem|st | but if I go back and forth and review alot it crashes on me |
| 09:56.58 | merlin1991 | hm gitorious gives at least a 503 now |
| 09:57.13 | freemangordon | yeaah, much better :D:D:D |
| 09:57.24 | chem|st | with mbarcode that happens instantly when cammodule gets a wake up so it might be a broken lib instead of the cam app |
| 09:58.12 | chem|st | but that started after cssu and was all fine before |
| 09:58.40 | chem|st | iirc |
| 10:11.59 | *** join/#maemo-ssu bsdmaniak (~bsdmaniak@std93-20-88-120-139-80.fbx.proxad.net) |
| 10:20.17 | freemangordon | hehe, a memleak in hildon_western_fkb.so |
| 10:33.38 | merlin1991 | time to fix it ;) |
| 10:34.52 | merlin1991 | also gitorious is back (finally) |
| 10:39.01 | DocScrutinizer05 | nicocam "segfault" been traced down to DSP-segfault. Now you could argue that's not anything nicocam touched/tweaked, but fact is I've not seen similar failures with stock camera-ui. And a missing link is just that: something you haven't found out the details of it yet |
| 10:51.25 | merlin1991 | damn I'd need pali and luf |
| 11:01.20 | merlin1991 | freemangordon: ping |
| 11:01.34 | freemangordon | merlin1991: pong |
| 11:01.54 | merlin1991 | anything on gitorious from that isn't thumb related and should go into the testing release that I'm preparing currently? |
| 11:01.55 | freemangordon | DocScrutinizer05: who traced it? |
| 11:02.26 | freemangordon | merlin1991: Qt, but you should pull the commit from thumb to master |
| 11:02.33 | freemangordon | tinymail too |
| 11:02.59 | DocScrutinizer05 | btw for nicocam - like for so many other packages that could get 'replaced' - applies the principle of not making something mandatory just because of it been mandatory in original maemo MP. E.G. for nicocam (or any other cam-ui replacement) it's ridiculously simple to make stuff optional, so user can decide in a simple picker GUI which of the available programs shall be his default |
| 11:03.03 | freemangordon | merlin1991: better wait for me to do it, i'll try to do it when I am bick home |
| 11:03.25 | freemangordon | *back |
| 11:03.25 | merlin1991 | freemangordon: is tinymail in -thumb too? |
| 11:03.30 | freemangordon | yep |
| 11:03.41 | merlin1991 | err i mean the commit in the thumb branch? :D |
| 11:03.52 | freemangordon | not sure, need some time to check |
| 11:04.19 | freemangordon | but anyway, I need a lunch otherwise i will fall on the ground, bbl |
| 11:04.22 | DocScrutinizer05 | freemangordon: I traced it a few weeks ago, and reported "live" here in this chan |
| 11:04.32 | merlin1991 | I hate to build qt, it takes forever |
| 11:10.13 | DocScrutinizer05 | I still think that getting rid of MP at all - or at very least moving out of MP as many pkgs as possible - is one of the major long term goals of CSSU. We should carefully evaluate dependencies whenever opportunity arises, and replace silly MP concept by proper comprehensive dependencies for the particular pkg. Freedom of choice is an extremely high value in CSSU, and it also relieves many of the QA requirements from replacements, |
| 11:10.15 | DocScrutinizer05 | when they can get opted out by user any time |
| 11:12.44 | merlin1991 | DocScrutinizer05: well you go patch apt and watch the world burn |
| 11:13.01 | DocScrutinizer05 | (silly) well for Nokia it's probably been a sensible choice to not waste manpower on detailed dependency checking&setup, when it came to their zoo of closed blobs that didn't offer any alternative. For a system/"distro" like CSSU this rationale doesn't apply at all |
| 11:13.07 | merlin1991 | there are a few (C) nokia patches in the maemo apt that make this basically impossible |
| 11:14.07 | DocScrutinizer05 | well, we got sucha gifted bunch of developers here, don't tell me we can't tackle that |
| 11:14.35 | DocScrutinizer05 | after all apt is FOSS afaik :-) |
| 11:15.22 | DocScrutinizer05 | so whatever Nokia "invented", we probably can come up with a superset that mitigates or even cures any flaws in Nokia's concept |
| 11:20.14 | DocScrutinizer05 | first we need to internalize and grok that we *want* to get rid of MP, and ponder options on every pkg we touch |
| 11:21.53 | DocScrutinizer05 | as long as we see MP as a God-given fact and just wonder how we can live _with_ it, we neglect to think how we might live better _without_ it and miss opportunities to move into right direction |
| 11:22.27 | merlin1991 | freemangordon: I guess in qt you mean Improve performance of partial updates in raster window surface on X11. ? |
| 11:22.53 | merlin1991 | DocScrutinizer05: main problem is, that we have no update utility besides ham |
| 11:23.02 | DocScrutinizer05 | :nod: |
| 11:23.03 | merlin1991 | and ham is bs to udpate a list of packages |
| 11:23.11 | DocScrutinizer05 | but aiui we can patch HAM |
| 11:24.50 | DocScrutinizer05 | merlin1991: well, obviously HAM is missing some functionality that's equivalent to apt-get dist-upgrade ? |
| 11:25.18 | DocScrutinizer05 | again due to MP concept made such function obsolete |
| 11:25.38 | DocScrutinizer05 | AIUI |
| 11:26.00 | DocScrutinizer05 | (I'm not the brightest packager, so just my humble 2 ct) |
| 11:29.00 | merlin1991 | the fact is that nokia removed the core / not core conecept form packaging, apt would never remove a core package from the system and for every other package it has to be either be user installed or a dependency of something else |
| 11:30.32 | DocScrutinizer05 | mhm, lemme think about that. And maybe there are alternative methods to "standard apt way" so we could have both - MP compatibility plus a migration path to something better |
| 11:33.00 | DocScrutinizer05 | thinks of abstract concepts like standard apt on top of nokia apt |
| 11:33.51 | DocScrutinizer05 | or a mode switch inside HAM, depending on repo |
| 11:34.13 | DocScrutinizer05 | some pretty green spontanous ideas |
| 11:35.22 | DocScrutinizer05 | wait, you say MP is nokia's replacement for a "core" tag on some packages? |
| 11:38.44 | *** join/#maemo-ssu lizardo (lizardo@nat/indt/x-uogmxorwoyvykjvz) |
| 11:39.03 | DocScrutinizer05 | I think we *might* safely assume that there won't be any new MP version from Nokia anytime this millenium, so maybe we can base on that, and redefine the way HAM/apt works for MP |
| 11:41.43 | DocScrutinizer05 | merlin1991: when Nokia removed the core/non-core concept, why can't we re-introduce it? |
| 11:42.06 | merlin1991 | because some closed packages miss the core flag in their packagaing |
| 11:42.20 | DocScrutinizer05 | well, it's HAM/apt looking for that tag |
| 11:42.22 | merlin1991 | so if we introduce it again we can have fun like you and sip rtcom |
| 11:43.20 | DocScrutinizer05 | we can patch HAM to change the way it "looks for core tag" - e.g parse if pkg is in MP |
| 11:44.21 | DocScrutinizer05 | A pkg from repo X? Repo X also has MP pkg? A pkg in that MP pkg? -> emulate core tag |
| 11:44.34 | DocScrutinizer05 | my naive approach |
| 11:45.04 | DocScrutinizer05 | (not that I've wrapped my head around all the complexity of the problem any precentage yet) |
| 11:46.45 | DocScrutinizer05 | what we need further in HAM/apt is "don't process MP like a normal package, rather ignore", and "override emulated core tag by true core tag in cssu repo, if any newer version there" (latter probably usual standard apt feature) |
| 11:49.16 | DocScrutinizer05 | btw SIP: 2 days ago I noticed inbound SIP calls via 3G are not signalled (no ringing), but later on show up as "missed call" :-/ |
| 11:49.44 | DocScrutinizer05 | ...while outbound SIP call seems to work just fine |
| 11:50.36 | DocScrutinizer05 | first of all I ought test via WLAN instead of 3G |
| 11:51.27 | DocScrutinizer05 | meh, WLAN works. Curse O2 |
| 11:52.03 | DocScrutinizer05 | seems I have to complain about their friggin NAT |
| 11:53.30 | DocScrutinizer05 | (<merlin1991> because some closed packages miss the core flag in their packagaing) we could patch HAM to read a textfile list of such pkgs, and emulate the core tag for those. |
| 11:54.09 | DocScrutinizer05 | hell, we probably can even patch apt's database to set the core flag on those pkgs |
| 11:54.30 | DocScrutinizer05 | somebody claimed he's doing such stuff all the time |
| 11:54.36 | merlin1991 | in that case it would be dpkgs database |
| 11:54.42 | DocScrutinizer05 | yep |
| 11:54.44 | merlin1991 | and apt's db aswell |
| 11:54.50 | merlin1991 | but apt rebuilds the db on each apt-get update |
| 11:55.07 | DocScrutinizer05 | hmm, we got control over that as well |
| 11:55.51 | merlin1991 | in the end we have a 3000 lines commit on gitorious that starts with // bloody ugly hack to work around nokia mp starts here |
| 11:55.59 | DocScrutinizer05 | probably 5 source lines to find pkg name in exception list and set core flag if found |
| 11:56.41 | DocScrutinizer05 | well, 3000 lines are not bad actually |
| 11:56.57 | DocScrutinizer05 | though I guess it's mor elike 600 max |
| 11:57.36 | DocScrutinizer05 | resets the "missed call" notification on 4 devices |
| 12:00.21 | DocScrutinizer05 | btw //pretty cute patch to work around ugly Nokia MP hack |
| 12:02.00 | DocScrutinizer05 | we could even write a wrapper around "apt-get update" to postprocess databases each time an update been done. Could probebly even been done as shellscript, with sed etc |
| 12:05.56 | DocScrutinizer05 | and to be absolutely sure, we also could implement a simple "daemon" that just does "while true; do inotifywait -e modify <database> && do-our-fix-databases; done" |
| 12:13.51 | DocScrutinizer05 | even: while true; do inotifywait -e modify <database> && (while kill -0 $(fuser <database>); do sleep 0.2; done) && kill -SIGSTOP $(pidof apt apt-worker hildon-app-manager dpkg) && do-our-fix-databases; && kill -SIGCONT $(pidof apt apt-worker hildon-app-manager dpkg) done |
| 12:15.24 | DocScrutinizer05 | get a script for sed to edit the databases, done? :-) |
| 12:16.01 | DocScrutinizer05 | (if there are no better dpkg/apt related tools for editing database) |
| 12:17.33 | DocScrutinizer05 | doesn't look all that daunting to me |
| 12:18.04 | DocScrutinizer05 | and for sure no 3000 lines of code ;-) |
| 12:19.14 | DocScrutinizer05 | let's just discuss feasibility instead of denying to think about possible pathes to get rid of MP |
| 12:19.34 | DocScrutinizer05 | after all it's a constant PITA |
| 12:20.33 | DocScrutinizer05 | and I'd happily volunteer to write such scripts/daemons to cope with fixing the missing core tags in Nokia's repos |
| 12:21.04 | DocScrutinizer05 | after we discussed the concept and found it should be ok |
| 12:22.35 | DocScrutinizer05 | as mentioned above, I'm not the brightest packager, so I miss guidance on requirements for the design/concept. I however can implement it once we got together all the requirements and operation principle |
| 12:34.56 | freemangordon | merlin1991: yes, partial updates performance commit |
| 12:35.51 | freemangordon | merlin1991: tinymai commits are in master |
| 12:36.13 | freemangordon | but both Qt and tinymail are missing proper debian/changelog |
| 12:36.25 | freemangordon | merlin1991: aah, microb-engine :D:D:D |
| 12:39.04 | freemangordon | merlin1991: https://bugs.maemo.org/12564 and https://bugs.maemo.org/12555 are fixed in microb-engine |
| 12:39.05 | povbot_ | Bug 12564: microb-engine: Daemon does not shut down properly when no connection and browser closed when connection dialog opens |
| 12:39.06 | povbot_ | Bug 12555: microb-engine: CVE-2010-1205: Buffer overflow in embedded libpng |
| 12:39.58 | freemangordon | proper debian/changelog missing too, sorry for that :( |
| 12:47.24 | freemangordon | merlin1991: keep in mind that maybe some dependencies in -mp are not in place (libxml and libalarm iirc) |
| 13:10.05 | *** join/#maemo-ssu taziff (~Taziff@cyr108.internetdsl.tpnet.pl) |
| 13:12.04 | DocScrutinizer05 | wow, povbot_ does multi |
| 13:13.20 | DocScrutinizer05 | I'd love to see CVE against fsckng flash getting catched (note I didn't say 'fixed') |
| 13:13.38 | DocScrutinizer05 | known open vulnerability |
| 13:14.38 | DocScrutinizer05 | I'd really prefer the process that runs flash plugin wouldn't have full access to everything on my device |
| 13:17.00 | DocScrutinizer05 | we probably should've written a PoC exploit for the very first security update published by adobe after N900 rollout, now not even real exploits out in the wild tailored particularly for N900/ARM/fremantle would make Nokia consider publishing a security patch |
| 13:21.50 | *** join/#maemo-ssu Estel_ (~Estel@Maemo/Community/council/Estel-) |
| 13:37.09 | jonwil | Its a pitty we dont have a newer flash build that we can use |
| 13:44.56 | DocScrutinizer05 | we got some, alas without hw accel afaik, and basically not any less obsolete than stock flash-plugin |
| 13:46.24 | DocScrutinizer05 | stock is flash9 with some backports from flash10 iirc, the 'leaked' one is a generic flash10 implementation without hw accel. |
| 13:48.10 | Lava_Croft | Flash on N900 is almost a reason to not get the N900 |
| 13:53.40 | jonwil | Best answer would be for someone to make flashblocker for N900 that can disable Flash completly |
| 13:54.01 | Lava_Croft | My first experience with Flash on N900 was also the last one |
| 13:54.07 | Lava_Croft | What a deception |
| 13:54.33 | Lava_Croft | Well, it did fix one thing for me personally: I never ever long for flash on mobiles anymore, since the N900 |
| 13:54.48 | Lava_Croft | Adobe does neither, these days:) |
| 14:02.08 | *** join/#maemo-ssu javispedro (~javier@Maemo/community/contributor/javispedro) |
| 14:13.13 | *** join/#maemo-ssu toxaris (~toxaris@90-230-114-186-no34.tbcn.telia.com) |
| 14:14.44 | Raimu | Isn't deleting the Flash plugin a very apt Flash blocker on the n900? |
| 14:16.17 | Raimu | Anyway, what was that thing about n9's Flash being adaptable onto n900? I vaguely recall someone making such a statement, somewhere. |
| 14:25.12 | *** join/#maemo-ssu toxaris (~toxaris@90-230-114-186-no34.tbcn.telia.com) |
| 14:35.13 | *** join/#maemo-ssu DocScrutinizer05 (~HaleBopp@openmoko/engineers/joerg) |
| 15:06.38 | amiconn | Wouldn't it still be possible to repackage a .deb even if the contained app is closed source? |
| 15:27.06 | Estel_ | Hey, N900's flassh 9 isn't accelerated too |
| 15:27.10 | Estel_ | why everyone claim it is? |
| 15:27.18 | Estel_ | it's why only 240p videos in Youtube. |
| 15:27.35 | Estel_ | the "leaked" generic flash 10 is just horribly unoptimized amd bugged |
| 15:27.38 | Estel_ | and* |
| 15:28.18 | Estel_ | but it's not last word of adboe's empire of evil, se we will probably see N900's "native" flash 10 version, soon. But it's not accelerated just like our current flash 9 |
| 15:29.02 | *** join/#maemo-ssu Pali (~pali@unaffiliated/pali) |
| 15:33.29 | Estel_ | flash is useless, anyway ;) |
| 15:41.05 | *** join/#maemo-ssu NIN101 (~NIN@p5DD28A98.dip0.t-ipconnect.de) |
| 15:49.38 | *** join/#maemo-ssu BCMM (~ben@unaffiliated/bcmm) |
| 16:08.40 | DocScrutinizer05 | WTF would produce a genuine N900 flash10? |
| 16:30.03 | amiconn | Estel_: The leaked flash 10 (there seem to be several point releases) usually stops playing youtube videos after the pre-buffered part is used up |
| 16:35.45 | ShadowJK | It'd be easy to tell, you van see with powertop which units are active |
| 16:37.20 | ShadowJK | iva2 = video accel |
| 16:41.43 | ShadowJK | flash wants to use the video, decoded, in rgb colourspace, whereas iva2 is suited for hw decode and displa through dss which then takes care of converion to rgb, but then you can't actually "see" the video from software :) |
| 16:42.17 | freemangordon | ShadowJK: but DSP has colorspace conversion codecs |
| 16:42.28 | freemangordon | vpp that is |
| 16:43.07 | freemangordon | and actually DSP driver on n900 does not support decoding to FB |
| 16:43.33 | ShadowJK | vpp? |
| 16:43.41 | freemangordon | video post processing |
| 16:43.59 | ShadowJK | i would've thought it'd be saner to use dss |
| 16:44.16 | ShadowJK | to minimize amount of rgb you have to pass around |
| 16:44.19 | *** join/#maemo-ssu M13 (~M13@170.133-224-87.telenet.ru) |
| 16:44.42 | ShadowJK | (something that isn't compatible with flash's way of doing things) |
| 16:44.57 | freemangordon | ShadowJK: won't work, you have only 2 overlays |
| 16:45.15 | freemangordon | and there is a limit of how much up/down sampling DSS can do |
| 16:45.26 | freemangordon | that is why 720p does not play on tv-out |
| 16:46.02 | ShadowJK | I was under the impression that currently dss is indeed used by media player stuff |
| 16:46.10 | freemangordon | it is |
| 16:46.24 | ShadowJK | right |
| 16:46.28 | freemangordon | but you cannot use it for ambeded plugin in browser AIUI |
| 16:46.33 | freemangordon | *embeded |
| 16:46.58 | ShadowJK | yesx |
| 16:47.08 | ShadowJK | that's what I'm saying |
| 16:47.12 | freemangordon | imagine a web page with several flash videos |
| 16:47.27 | freemangordon | but you stil can use dsp |
| 16:47.45 | ShadowJK | that the best hw accel path is incompatible with flash |
| 16:47.52 | freemangordon | the major problem aiui is that flash does not use NEON |
| 16:48.04 | freemangordon | NEON is second to the best |
| 16:48.18 | ShadowJK | even with neon, 480p is a struggle |
| 16:48.28 | freemangordon | what bitrate? |
| 16:48.43 | freemangordon | and why do you think that? |
| 16:48.49 | ShadowJK | 1M main/high profile |
| 16:49.05 | ShadowJK | baseline probably works |
| 16:49.22 | freemangordon | camera subsystem can do up/downsampling in HW memory to memory, so you can chain it |
| 16:49.42 | freemangordon | so you youse NEON for decoding and capera for resizing |
| 16:49.54 | freemangordon | damn typos :D |
| 16:50.59 | merlin1991 | there has to be a way to fix hw ids to alasa device numbers |
| 16:51.26 | merlin1991 | my usb headset and the nvidia hdmi audio output keep swapping around on reboot which messes with my .asoundrc :/ |
| 16:51.34 | freemangordon | ShadowJK: camera subsystem is called ISP |
| 16:51.37 | javispedro | merlin1991: use device names ;P |
| 16:52.04 | ShadowJK | Is there enough bandwidth for it? |
| 16:52.08 | merlin1991 | my asoundrc looks like pcm.!default { type plug; slave.pcm dmix:2; } |
| 16:52.09 | freemangordon | merlin1991: did you see may comments for microb-engine, tinymail and Qt? |
| 16:52.13 | merlin1991 | yes |
| 16:52.28 | freemangordon | ShadowJK: well, 720p plays with no problem, so there should be |
| 16:52.58 | merlin1991 | javispedro: how would I use a devicename indstead of the dmix:2 part? and how do I find out the devicename I need? |
| 16:53.10 | ShadowJK | recalls P2-P3 era, where memcpy on a decoded frame meant 30% hit on decoding speed :) |
| 16:53.44 | freemangordon | ShadowJK: according to some benchmarks memcpy using ARM/NEON/PLG is about 600 MB/s |
| 16:53.47 | javispedro | merlin1991: /proc/asound/cards lists card names |
| 16:53.51 | freemangordon | *PLD (preload) |
| 16:54.12 | freemangordon | or even 700, so no bandwith problem |
| 16:54.19 | javispedro | to MEMORY? |
| 16:54.27 | javispedro | O.o |
| 16:54.35 | javispedro | more like half of that |
| 16:54.38 | freemangordon | javispedro: memory to memory |
| 16:54.42 | merlin1991 | javispedro: which part is the card name? ie my internal card is 0 [PCH ]: HDA-Intel - HDA Intel PCH in /proc/asound/cards |
| 16:54.47 | javispedro | merlin1991: PCH |
| 16:54.52 | freemangordon | javispedro: lemme find that |
| 16:54.56 | javispedro | freemangordon: no way, half of that _at most_ |
| 16:55.09 | merlin1991 | so I just use dmix:PCH in my .asoundrc and it should work ? |
| 16:55.35 | javispedro | merlin1991: I do not know the syntax of asoundrc by heart, I do know that in the card 0 statements you can use card Intel e.g. and it will work |
| 16:56.23 | javispedro | freemangordon: I'd believe 700 for L2 to L2 |
| 16:56.29 | merlin1991 | javispedro: it worked :) |
| 16:56.51 | freemangordon | javispedro: could be, I am trying to find the article |
| 16:58.24 | freemangordon | javispedro: http://sourceware.org/ml/libc-ports/2009-07/msg00000.html |
| 16:58.34 | freemangordon | that is without L1 on 500 MHz |
| 16:58.36 | merlin1991 | btw freemangordon where doe sthe Qt speed improvement help (which use applications) |
| 16:58.45 | merlin1991 | or is it a general rendering improvement |
| 16:58.48 | freemangordon | merlin1991: read the description |
| 16:58.52 | freemangordon | general |
| 16:59.14 | merlin1991 | did you happen to think of it, or is it backported from upstream? :D |
| 16:59.30 | freemangordon | merlin1991: backported |
| 16:59.57 | freemangordon | merlin1991: I think there is a link to original patch, lemme check |
| 17:00.08 | javispedro | freemangordon: well, it says ~370, nowhere near 700 =) |
| 17:00.17 | merlin1991 | freemangordon: true, there is one |
| 17:00.19 | freemangordon | javispedro: yes, my bad |
| 17:00.21 | javispedro | still considers that quite high, goes lookup LPDDR specs for beaglebard |
| 17:00.24 | javispedro | *board |
| 17:00.49 | merlin1991 | freemangordon: sometimes I should display full commit messages, normally I just have the shortened 1 line version :D |
| 17:00.50 | freemangordon | but still is pretty enough for the case we were discussing |
| 17:01.30 | freemangordon | javispedro: I didn't find what I was looking for, I still think the speeds were about 600-700 |
| 17:02.49 | freemangordon | javispedro: and have in mind DMA can be used |
| 17:03.05 | freemangordon | both ISP and DSP support DMA |
| 17:03.11 | merlin1991 | does anybody know how busybox-power deploys the binaries exactly? |
| 17:03.38 | merlin1991 | because pali did a fix for something busybox related, and I'm not sure if we mess up busybox-power installations if we force a busybox upgrade |
| 17:05.31 | merlin1991 | FFS gitorious is 503 again |
| 17:06.07 | *** join/#maemo-ssu zeq (~s_j_newbu@2a01:348:1e3:1:e6ec:10ff:fe9a:d418) |
| 17:08.44 | DocScrutinizer05 | (<merlin1991> my usb headset and the nvidia hdmi audio output keep swapping around ) there's a howto on alsa.org or soemwhere how to load drivers in a defined way |
| 17:10.54 | DocScrutinizer05 | dafaq: http://alsa.opensrc.org/Index.php/MultipleCards#Recommended_solution |
| 17:12.03 | javispedro | freemangordon: I repeat the test on my N950. for L1 speeds are exactly doubled (<1% error), as expected (500 vs 1Ghz), L2 speeds are tripled (unexpected..), external memory access is the same (also <1% error) |
| 17:12.08 | merlin1991 | dafaq "There is currently no text in this page. You can search for this page title in other pages, or search the related logs." |
| 17:12.24 | DocScrutinizer05 | http://alsa.opensrc.org/Hotplugging_USB_audio_devices_(Howto) |
| 17:12.50 | javispedro | freemangordon: except that memcpy_neon no longer has any speed advantage (expected, as LPDDR is the bottleneck) |
| 17:12.55 | DocScrutinizer05 | look there, you'll eventually find how to pass device to modprobe alsa |
| 17:14.19 | javispedro | you ain't getting 800 out of LPDDR, considering one bank can do around ~50MBps only |
| 17:14.45 | javispedro | s/800/700 |
| 17:15.10 | javispedro | plus the usual ARM bandwidth limitations |
| 17:15.53 | freemangordon | javispedro: ok, I won't argu as I could easily be wrong, the point was that eve 300 MB/s is ok |
| 17:15.57 | freemangordon | *even |
| 17:16.04 | javispedro | (the ~50Mbps is assuming 200Mhz, no idea at what freq DDR runs on the N9/00) |
| 17:16.20 | freemangordon | isn't it ddr2? |
| 17:16.53 | javispedro | double the figure then :P |
| 17:17.12 | javispedro | (I don't remember, I though beagleboard was lpddr1) |
| 17:18.39 | javispedro | nah, bb is lpddr1, and @166Mhz |
| 17:18.47 | javispedro | therefore N9's also lpddr1 at same freq. |
| 17:19.07 | freemangordon | n900 is @ 133, ain't? |
| 17:20.32 | javispedro | I suspect it will be 166 too, but can't really check |
| 17:22.15 | javispedro | now, why is the L2 cache so fast in the N9's A8? |
| 17:27.18 | DocScrutinizer05 | syslog |