**** BEGIN LOGGING AT Wed Jul 25 02:59:58 2012 Jul 25 04:16:35 anyone know of a way to get either gitorious or a local clone of a gitorious repository to display a nice view of the flow of commits/branches/etc? I know its possible, I did it before on a totally unrelated git project but now I cant remember how Jul 25 04:16:45 the local thing that is Jul 25 04:36:46 aha, turns out that calendar-backend/master is not the latest version, its calendar-backend/PE1 that has the N900 PR1.3 current version Jul 25 04:38:17 so the links on http://wiki.maemo.org/Fremantle_closed_packages are wrong (or were until I fixed them) Jul 25 05:15:41 hi. I am bidding on an n900 on eBay, already have an n810 that I used tethered to a PalmOS smartphone a lot at one time. Jul 25 05:16:06 how's getting maemo.org migrated to community instead of Nokia coming along? Jul 25 05:21:12 guess everyone's asleep or something. Jul 25 05:28:57 EdLin: we have to sleep some time ;) Jul 25 06:42:43 freemangordon: I've been trying to fix the fennec compile failure. It stems from the SharedTextureHandle being of type uintptr_t, which should be able to be set to nsnull only by virtue of it's pointer-like properties, but it's being resolved to unsigned int, which of course can't hold a null ptr. Definitely a bug, and perhaps something that doesn't show up in older gcc's? Not sure how to fix it, I'm attempting to compile Jul 25 06:43:19 zeq1: you're slow :P https://bugzilla.mozilla.org/show_bug.cgi?id=777241 Jul 25 06:43:20 04Bug 777241: was not found. Jul 25 06:43:46 hey man, I've gotta sleep! :P Jul 25 06:44:37 freemangordon: I thought about changing it to return 0. Jul 25 06:44:48 zeq1: how to reclaim free space on my rootfs? Jul 25 06:44:51 in SB Jul 25 06:45:03 But, shouldn't uintptr_t support being a null ptr? Jul 25 06:45:17 should Jul 25 06:45:48 freemangordon: remote dbg packages? Jul 25 06:45:57 s/remote/remove/ Jul 25 06:45:58 zeq1 meant: freemangordon: remove dbg packages? Jul 25 06:46:15 zeq1: i movet /scratchbox to a different partition Jul 25 06:46:19 clean out your /var/lib/apt/archives Jul 25 06:46:44 yet my freespace is 7.7 GB, while used space is 12GB Jul 25 06:46:53 on new partition Jul 25 06:46:58 s/lib/cache/ Jul 25 06:46:59 zeq1 meant: clean out your /var/cache/apt/archives Jul 25 06:47:16 that's a lot of used space Jul 25 06:47:26 just for SB Jul 25 06:47:45 hmm, hold on, I forgot to rm one thing Jul 25 06:47:49 :D Jul 25 06:49:40 freemangordon: I take it you haven't tested the toolchain yet? Jul 25 06:49:54 I managed to rebuild qt4 with it :) Jul 25 06:49:56 zeq1: no, I was fighting with SB Jul 25 06:50:10 but it is ok now, I am going to create a new target Jul 25 06:50:18 ok Jul 25 06:50:28 hmm, or better update my current one Jul 25 06:50:46 zeq1: ^^^ any risk you are aware of? Jul 25 06:50:53 of rebuilding qt4? Jul 25 06:51:08 update my current target with a new toolchain Jul 25 06:51:10 oh you mean updating your current profile Jul 25 06:51:38 it's okay as long as you don't extract the rootstrap on top Jul 25 06:51:40 ;) Jul 25 06:52:06 also make sure manually install the compiler/../packages/* Jul 25 06:52:13 yeah Jul 25 06:52:20 ok, going to try it Jul 25 06:53:15 good luck!:) Jul 25 06:54:07 I'm glad I managed to get the size of the toolchian down quite a bit ;) Jul 25 06:56:06 from 215Mb to 58M is pretty good going :) Jul 25 06:56:16 yeah :) Jul 25 06:56:32 lets see if it works :D Jul 25 06:57:55 what are you going to test it with? Jul 25 06:58:01 yes Jul 25 06:58:22 it works? Jul 25 06:58:30 still don't know Jul 25 06:58:38 lemme setup the target first Jul 25 06:59:55 zeq1: how to tell tar to unbx2? Jul 25 07:00:00 *unbz2 Jul 25 07:00:31 found it Jul 25 07:00:32 it does it automatically Jul 25 07:00:35 just use xf Jul 25 07:00:43 xfj ;) Jul 25 07:00:49 you can do :P Jul 25 07:01:06 automatic works too :) Jul 25 07:01:43 I would have used xz, but not everybody probably has that Jul 25 07:05:53 zeq1: gcc version 4.7.2 20120701 (prerelease) (Linaro GCC 4.7-2012.07) Jul 25 07:06:11 using -std=gnu++11 didn't work Jul 25 07:06:22 looks correct :) Jul 25 07:06:41 lemme install .debs and check if it works Jul 25 07:06:57 I've changed the return nsnull to return 0 Jul 25 07:08:34 zeq1: i can't see libgcc-dev package Jul 25 07:08:51 should there be one? Jul 25 07:09:26 no :) Jul 25 07:09:31 scratch that Jul 25 07:09:35 :D Jul 25 07:10:31 be aware that installing the libstdc++6-dbg package makes your ldconfig output ugly Jul 25 07:10:47 too late :D Jul 25 07:10:56 (the dynamic linker doesn't like (nor quietly ignore) the .py file Jul 25 07:11:11 aah, i can easily remove it by hand Jul 25 07:11:19 it's there for a reason ;) Jul 25 07:11:27 but what reason Jul 25 07:11:32 NFC Jul 25 07:11:36 :) Jul 25 07:11:42 ok, so what exactly is the reason for doing all this GCC stuff? Jul 25 07:11:43 something to do with gdb Jul 25 07:11:46 Newer version of GCC? Jul 25 07:11:55 different compile thats better in some way? Jul 25 07:12:04 jonwil: 4.7.2? Jul 25 07:12:07 jonwil: it gives a really nice performance boost if nothing else Jul 25 07:12:13 should be better than 4.2.1 Jul 25 07:12:32 it's also compatible with more recent standards Jul 25 07:12:40 particularly with c++ Jul 25 07:13:01 it also have NEON autovectorization ;) Jul 25 07:13:07 neat :) Jul 25 07:13:09 s/have/has/ Jul 25 07:13:09 zeq1 meant: it also has NEON autovectorization ;) Jul 25 07:13:32 LTO should help with binary sizes and give better optimization too Jul 25 07:14:28 Is this something that's going to be used for CSSU so that when binaries are built for CSSU, they get the benefits? Jul 25 07:14:52 jonwil: maybe, if it proves itself :) Jul 25 07:15:11 src/Makefile.am:128: CXXFLAGS must be set with `=' before using `+=' Jul 25 07:15:14 :( Jul 25 07:15:22 hmm Jul 25 07:15:27 what's that from ? Jul 25 07:15:34 dpkg-buildpackage Jul 25 07:15:43 that's autoconf Jul 25 07:15:46 automake/configure step Jul 25 07:15:51 yeah Jul 25 07:15:58 what package? Jul 25 07:16:01 fennec? Jul 25 07:16:07 libgtkmozembed Jul 25 07:16:28 you're not setting CXXFLAGS before appending to them Jul 25 07:16:46 as it says ^^^ Jul 25 07:17:00 it was building just fine with my toolchain Jul 25 07:17:13 you must have had CXXFLAGS defined in your env Jul 25 07:17:33 ? Jul 25 07:18:33 fixed? Jul 25 07:19:30 no :( Jul 25 07:19:56 will I get the package with apt-get source? Jul 25 07:20:07 yes Jul 25 07:20:16 libgtkmozembed Jul 25 07:20:46 taking a look Jul 25 07:23:52 I don't think it's a problem with my toolchain Jul 25 07:24:02 could be Jul 25 07:24:57 I can't see how it could have ever worked tbh :S Jul 25 07:26:02 automake is quite fussy about having things defined properly Jul 25 07:26:11 perhaps moreso in recent versions Jul 25 07:27:17 maybe my SB is broken because of moving Jul 25 07:27:51 CXXFLAGS := Jul 25 07:28:05 at the top of the Makefile.am will fix it Jul 25 07:28:27 hmm, only THUMB target is broken, armel works just fine Jul 25 07:28:38 hmm Jul 25 07:28:47 zeq1: romaxa is asking if the patch is split, on #mobile Jul 25 07:28:47 ugh.. apnews pissed me off Jul 25 07:28:50 is thumb not working? Jul 25 07:28:55 zeq1: yes Jul 25 07:29:02 going there on my transformer keeps senging me to a page that talks about their stupid android app Jul 25 07:29:05 it that yes it isn't working? Jul 25 07:29:16 with no option to continue to their mobile page Jul 25 07:29:21 i don't want to use their app Jul 25 07:29:35 zeq1: would you join and discuss with romaxa, as I am not sure what the status is Jul 25 07:29:38 ok Jul 25 07:29:50 zeq1: yes, thumb is not working, armel is ok Jul 25 07:29:57 armel -4.2.1 Jul 25 07:30:03 thumb - 4.7.2 Jul 25 07:30:15 oh right Jul 25 07:30:42 can you just test the compiler without autotools? Jul 25 07:31:04 I don't understand what could be different Jul 25 07:31:49 if I revert the toolchain to 4.6.2 it is again ok :( Jul 25 07:33:02 :( ??? Jul 25 07:33:22 yeah, pity Jul 25 07:33:48 freemangordon: have you managed to split-up your patch and check if it works without that coment/condition confict Jul 25 07:34:03 zeq1: ^^^ Jul 25 07:34:24 I gtg, will continue later Jul 25 07:34:26 bb Jul 25 07:34:30 I hmm I'm on #mobile now Jul 25 07:34:45 see you in a bit Jul 25 07:35:26 zeq1: I saw, just pasting here the question romaxa asked :) Jul 25 07:36:49 thanks freemangordon Jul 25 07:36:50 :) Jul 25 07:41:31 zeq1: the problem with your toolchain could be that it does not have default values for those vars Jul 25 07:41:41 anyway, i am off Jul 25 07:42:42 MohammadAG: ping Jul 25 07:47:15 A pregnant Amish woman picking berries in the woods near her Pennsylvania home was killed by lightning on her due date. The fetus also died. Jul 25 07:47:26 wow... seriously makes you question the amish faith Jul 25 07:51:30 freemangordon: I'm trying to track it down Jul 25 07:53:02 freemangordon: could be a different libtool version (it's provided by the toolchain) Jul 25 08:01:23 is there an n900 app that automatically sends an email when it gets a txt or a missed call? Jul 25 08:07:02 zeq1: i am going to setup a new target, maybe some remnants from 4.6.2 are to blame Jul 25 08:07:06 freemangordon: I'm _certain_ it's because the new toolchain has a more recent libtool version. Jul 25 08:07:42 (this is inherited from the more recent sb-toolchain Jul 25 08:07:45 ) Jul 25 08:08:20 IMHO the correct thing to do is correct the packages, since it's actually just stricter with real autotools bugs Jul 25 08:09:25 freemangordon: the fennec OpenGL support has been completely messed up Jul 25 08:10:26 GLdouble is getting defined twice all over the place because the GLDefs.h header is getting included too early. Jul 25 08:10:49 ie before QGLContext Jul 25 08:11:07 there is a bug report for that too, romaxa is working on it Jul 25 08:11:13 ah ok :) Jul 25 08:11:33 I'm obviously a bit slow today ;) Jul 25 08:12:18 freemangordon: what do you think about autotools Jul 25 08:12:19 ? Jul 25 08:13:00 I could try to force sb-toolchain-extras to use an ancient libtool version, but I *know* that will break other things Jul 25 08:13:19 LTO for example Jul 25 08:13:40 I know this from Gentoo :P Jul 25 08:14:42 When I've attempted to emerge @world with LTO enabled, packages which ship their own old libtool or don't get autoreconf'd break. Jul 25 08:15:46 It's usually not too difficult to fix up the bugs Jul 25 08:20:59 zeq1: no joy with new target too :( Jul 25 08:21:27 freemangordon: like I said ^ it's the newer libtool version Jul 25 08:21:54 zeq1: why default is --with-fpu=neon? Jul 25 08:22:02 why not? Jul 25 08:22:24 because the original is with --with-fpu=vfp, we don;t know what that will break Jul 25 08:23:01 it shouldn't break anything that doesn't break with the new gcc IMO Jul 25 08:23:40 it just allows NEON instruction selection/intrinsics Jul 25 08:23:54 well, ok Jul 25 08:24:50 -mfpu=vfp still works Jul 25 08:25:08 if it's really needed Jul 25 08:25:32 (if for example a package uses neon intrinsics, but the code is broken) Jul 25 08:25:48 ok Jul 25 08:26:15 although --disable-neon or something for that package would probably be more approriate probably Jul 25 08:26:20 anyway.... Jul 25 08:27:09 shall I fixup the automake stuff liblibgtkmozembed Jul 25 08:27:12 ? Jul 25 08:28:00 how? Jul 25 08:28:08 new toolchain? Jul 25 08:28:36 zeq1: ^^^ Jul 25 08:28:40 as I said it's real bugs exposed by the newer libtool Jul 25 08:29:02 we can't really use an old libtool with new gcc Jul 25 08:29:11 but, but, it works with 4.6.2 Jul 25 08:29:13 since it breaks gcc support Jul 25 08:29:26 and we cannot fix all the packages Jul 25 08:29:27 depends which features you use Jul 25 08:29:39 most probably aren't broken Jul 25 08:29:41 the same libgtkmozembed compilation is ok Jul 25 08:30:02 well, configure Jul 25 08:30:15 yes it's just the libtoolize/automake Jul 25 08:30:34 hmm, lemme try to upgrade automake Jul 25 08:30:48 ok Jul 25 08:31:11 good morning gents Jul 25 08:31:28 I suspect (hope) you were just unlucky with libgtkmozembed Jul 25 08:31:28 Do you know if it is possible to just dump the raw nmea strings from the N900 gps? Jul 25 08:31:31 hi vi__ Jul 25 08:31:56 But what if he comes bac?? zeqs dead baby, zeqs dead! Jul 25 08:32:48 freemangordon: fixing libgtkmozembed was easy Jul 25 08:38:30 freemangordon: I've put a patch up on my site Jul 25 08:38:46 libgtkmozembed-libtool.patch Jul 25 08:38:51 ok Jul 25 08:42:40 need to add it to quilt Jul 25 08:42:46 zeq1: well, hildon-desktop builds ok, so no new toolchain :D Jul 25 08:43:04 libgtkmozembed was the first package i tried Jul 25 08:43:09 forget about it Jul 25 08:43:56 I've built it with the patch Jul 25 08:44:22 it's probably worth you taking a look at the patch just to see what had to be changed Jul 25 08:44:34 ok Jul 25 08:45:23 zeq1: nice, i will keep that in mind for a referenc Jul 25 08:45:26 e Jul 25 08:47:02 freemangordon: if you want to try autovectorization (will probably break some things) you should use -ftree-vectorize in conjunction with -ffast-math or -Ofast Jul 25 08:47:13 -fgraphite is optional Jul 25 08:47:20 but works fine too Jul 25 08:48:03 fast-math is needed because neon instructions don't get used where they aren't ieee-math compliant without it Jul 25 08:48:09 zeq1: for some reason your messages crash my IRC client :( Jul 25 08:48:13 (it falls back to vfp) Jul 25 08:48:26 impressive :S Jul 25 08:48:41 not all of them, but still :S Jul 25 08:48:49 I'm using irssi Jul 25 08:49:06 leafchat here, used to work ok till half an hour Jul 25 08:49:09 that certainly shouldn't cause a problem !?! Jul 25 08:49:25 maybe it's the things I say it doesn't like (!) Jul 25 08:49:32 :D:D:D Jul 25 08:50:16 well, toolchain looks ok. did youtry to install gcc-base,libstdc++ and libgcc1 on the device? Jul 25 08:50:34 I did with my old b0rken versions... Jul 25 08:50:42 it's going to be fun ... Jul 25 08:51:03 (I also installed qt4 with dep on epoch 4 gcc debs Jul 25 08:51:07 ) Jul 25 08:51:08 could you try the latest (stripped). I don't want to risk to remain without a phone :D Jul 25 08:51:24 yes, ofc Jul 25 08:51:25 well, you could tell dpkg to ignore dependencis Jul 25 08:51:36 I'll probably have to! Jul 25 08:52:29 * zeq1 goes ahead and tries not to break his N900 anymore than he already has... Jul 25 08:54:42 pali: ping Jul 25 08:54:49 jonwil, pong Jul 25 08:55:16 I updated a couple of links in http://wiki.maemo.org/Fremantle_closed_packages Jul 25 08:55:29 yes I got emails Jul 25 08:55:47 specifically calendar-backend-*, camera-firmware and tone-generator were updated to point to the correct PR1.3 revisions Jul 25 08:55:57 or in the case of tone-generator, as close as its possible to get Jul 25 08:56:08 freemangordon: I'm replacing libqt4* at the same time, but it would have been a good idea not to have installed the dbg and dev packages... crap Jul 25 08:57:08 jonwil, decompiled prolog code show that there is no regular way to enable jack bias Jul 25 08:57:26 there must be incoming call or active call Jul 25 08:58:04 so manually sending something to alsaped wont help? Jul 25 08:58:08 and you can create "virtual" call via libplayback library (register your app as voip and set state active) Jul 25 08:58:13 because prolog will disable it off Jul 25 08:58:34 ohm more times check prolog predicates and emits dbus signals Jul 25 08:59:02 so something which may work is to listen for jackbias off signal and then send jackbias on Jul 25 08:59:23 but you need to sent that array of dicts via dbus - and I do not know how Jul 25 08:59:48 I changed OMP code to act as ipcall - voip Jul 25 08:59:56 and set state active Jul 25 09:00:03 it really enabled jackbias Jul 25 09:00:15 ok, so if we can pretend to be voip, does that solve the problem? Jul 25 09:00:17 or not? Jul 25 09:00:22 but in voip active call is MAFW audio output disabled Jul 25 09:00:28 ok Jul 25 09:00:40 ohm set volume level to 0 Jul 25 09:00:47 and it cannot be changed Jul 25 09:00:57 seems like the only hope would be to convince Nokia (or Kristztian Litkey as the maintainer of policy-settings-rx51) to publish the prolog code so we can modify it... Jul 25 09:01:12 That or apply big hacks to ohm and various other parts of the audio system Jul 25 09:01:34 jonwil, see: http://talk.maemo.org/showpost.php?p=1242365&postcount=39 Jul 25 09:01:50 call_active or call_alerting must be true Jul 25 09:01:52 yes I saw that Jul 25 09:02:01 otherwise jackbias us off Jul 25 09:02:46 but if call_active or call_alerting is true audio output is enabled only from audio call Jul 25 09:03:16 so this cannot be enabled without modifing context_variable(jackbias, B) :- predicate Jul 25 09:03:49 freemangordon: installing the qt4 dbg and dev debs on the N900 is a *very bad* idea Jul 25 09:03:57 why is it bad? Jul 25 09:04:12 because thet don't fit! Jul 25 09:04:15 you cannot enable jack bias for OMP Jul 25 09:04:18 s/thet/they/ Jul 25 09:04:20 zeq1 meant: because they don't fit! Jul 25 09:04:20 ok Jul 25 09:04:28 so headset button will not work in OMP Jul 25 09:04:51 package gets left in bad state after no space left Jul 25 09:04:52 jonwil, if we can load that binary prolog code to interpretor there should be way to add new predicate and then compile that prolog code again to file Jul 25 09:04:54 zeq1: yeah :D Jul 25 09:05:31 jonwil, current loaded code can be compiled by: qsave_program('/path/to/output/file', []). Jul 25 09:05:37 freemangordon: otherwise, it seems to be working ok Jul 25 09:06:01 WTF is with my irc client :( Jul 25 09:06:10 so the way I see it, we have several options: 1 is to somehow get the prolog source and modify it to do what we want (way to toggle jackbias without being in audio call), 2 is to write something to send jackbias stuff to alsaped (and detect when ohm tries to override it), 3 is to hack ohmd and whatever else to override the thing where it turns off the other audio when a call is in progress Jul 25 09:06:45 zeq1: could you ping me? Jul 25 09:07:01 jonwil, can you ask that nokia guy about it? Jul 25 09:07:21 freemangordon: ping Jul 25 09:07:22 if you can help me come up with some details of what to say, yeah, I can send the email I guess Jul 25 09:07:49 hmm, yeah, it was that i turned sounds, and now I am connected through remote desktop Jul 25 09:08:03 now with sounds disabled it is ok Jul 25 09:08:31 ask if he can send us source code of that packge, write that all ohm/prolog/dres code is published but policy-settings-rx51 is missing Jul 25 09:08:46 or if there is other way how to enable jack bias Jul 25 09:09:17 freemangordon: you're back with us then? :) Jul 25 09:09:29 zeq1: yeah :D Jul 25 09:10:08 zeq1: could you repeat what is the status of packages when istalled on the device? Jul 25 09:10:23 i missed those messages Jul 25 09:11:11 the libstdc++6 and libgcc1 seem to work fine Jul 25 09:11:36 I've just got a few incomplete installs of dbg and dev debs I need to clean up Jul 25 09:11:49 what about gcc-base? Jul 25 09:11:55 yep Jul 25 09:11:57 that too Jul 25 09:12:03 btw pali, it looks like tonegend should be possible to get matching N900 if you were to merge the fix from "stream don't kill if disconnection isn't possible" with the tree just before "removed debian packaging" Jul 25 09:12:18 zeq1: ok, thats good news Jul 25 09:12:37 zeq1: should we mad an oficial announcement? Jul 25 09:12:46 s/mad/make/ Jul 25 09:12:47 freemangordon meant: zeq1: should we make an oficial announcement? Jul 25 09:13:39 freemangordon: just for thumb? I don't see why not :) Jul 25 09:13:52 why only for thumb? Jul 25 09:14:05 that's fully functional SB toolchain Jul 25 09:14:44 I was asking :) Jul 25 09:14:54 aah :) Jul 25 09:15:19 btw, is there a list somewhere of the packages that have so far been recompiled for thumb? Jul 25 09:15:31 jonwil: in repo? Jul 25 09:15:40 yeah I guess so Jul 25 09:15:42 cssu-thumb repo Jul 25 09:15:56 http://maemo.merlin1991.at/cssu/community-thumb/ Jul 25 09:16:08 get packages.gz from there Jul 25 09:16:10 do you know something when will be cobs? Jul 25 09:16:50 Pali: NFC. but I wouldn't hold my breath Jul 25 09:17:29 do we have a thumb compiled libc yet? Jul 25 09:17:35 freemangordon, there is problem with kp packages again Jul 25 09:17:40 missing in repo :-( Jul 25 09:17:44 fuck :( Jul 25 09:17:49 which version? Jul 25 09:17:53 52? Jul 25 09:17:58 51 Jul 25 09:18:03 r2? Jul 25 09:18:21 I already contated council, problem is seems same: admin must manually set new version Jul 25 09:18:24 r1 Jul 25 09:18:44 I will not push new version because it cause other problems again... Jul 25 09:18:52 jonwil: no, and i would prefer libc to stay ARM, for performance reasonc Jul 25 09:19:00 only kernel-power and kernel-power-headers are OK Jul 25 09:19:11 this is really not problem with version numbers Jul 25 09:19:30 Pali: lets obsolete KP Jul 25 09:19:35 freemangordon: it's probably worth recomiling libc though Jul 25 09:19:35 I guess thumb<->arm transitions cost performance Jul 25 09:19:36 but something else, because always random kernel-power-* are OK and some not Jul 25 09:19:59 yes definatly worth recompiling libc in arm with the latest and best GCC Jul 25 09:20:07 it just not make sense to play the same circus every time Jul 25 09:20:07 I think that maemo.org has problem with parsing debian/control Jul 25 09:20:09 same with any other key system libraries like glib Jul 25 09:20:10 freemangordon: it could give us a faster libm Jul 25 09:20:12 or something else Jul 25 09:20:28 we have really big debian/control due to 2 or 3 icons and a lot of provides Jul 25 09:20:59 and something like random packages are ok seems like some segfault or buffer problem... Jul 25 09:21:26 Pali: lets split KPS from KP, put KPS in extras-devel, and switch to kernel-cssu only Jul 25 09:21:49 KPS is in separate source package Jul 25 09:22:00 also this is missing in maemo repo too Jul 25 09:22:18 freemangordon: we could also provide a hardfp libc in case somebody wants to use their N900 for numbercrunching ;) Jul 25 09:22:21 * zeq1 runs Jul 25 09:22:43 zeq1: kernel does not support hrdfp afaik Jul 25 09:22:53 zeq1: that type of talk will get you a free N9 Jul 25 09:22:58 * Macer hides under a rock Jul 25 09:23:02 finally, I can use my n900 as a rainbow table generator and a pocket heater. Jul 25 09:23:03 freemangordon, council may get admin rights for repos and then I belive it will be easier Jul 25 09:23:22 Pali: I know, but noone knows how long that will take Jul 25 09:23:34 ffs, just give the CSSU developers admin rights... Jul 25 09:23:36 Pali: might be a good time to move maemo.org somewhere before nokia deletes it all Jul 25 09:23:47 and having in mind extrash will be gone in couple of months,... Jul 25 09:24:03 vi__: ;) Jul 25 09:24:12 merlin1991: ping Jul 25 09:24:21 vi__: give everybody admin rights! Jul 25 09:24:33 REWT FOR EVERYONE! Jul 25 09:24:42 what ever became of "free and open" ?? Jul 25 09:24:44 :) Jul 25 09:25:00 Macer: DO you know how to improve the n900 gps accuracy? Jul 25 09:25:15 vi__: yes.. use an external GPS :-P Jul 25 09:25:22 -_- Jul 25 09:25:34 i honestly never had a problem with the gps not being accurate Jul 25 09:25:39 my problem was more in the lock on time Jul 25 09:25:54 I don't have problems on both issues Jul 25 09:25:59 I use it to map my runs and the tracks put through buildings and randomly crossing the roads. Jul 25 09:26:04 it is acurate and locks in seconds Jul 25 09:26:05 has anybody mirrored maemo.org? Jul 25 09:26:06 freemangordon: try not using supl :-P Jul 25 09:26:20 zeq1: merlin1991 afaik Jul 25 09:26:32 Macer: and why is that? Jul 25 09:26:42 it is a-gps after all Jul 25 09:26:48 I've seen GPS probs in narrow streets lined with ~7 stories buildings Jul 25 09:26:57 freemangordon: my sim reader/modem is flakey Jul 25 09:27:05 since i swapped to att i don't use the sim in the n900 anymore Jul 25 09:27:11 aah Jul 25 09:27:13 Yesterday It told me I hit 30km/h Jul 25 09:27:13 tried to use it with ecoach to go running... gave up on it Jul 25 09:27:22 total horse shit. Jul 25 09:27:26 vi__: running? Jul 25 09:27:31 yes. Jul 25 09:27:35 Macer: is you clock set accurately? Jul 25 09:27:35 :) Jul 25 09:27:35 zeq1: trying to build fennec with 4.7 Jul 25 09:27:48 yes. Jul 25 09:27:50 zeq1: used ntpdate to update it Jul 25 09:27:51 so yes Jul 25 09:27:54 freemangordon: it's broken at the moment Jul 25 09:27:54 wifi works fine ;) Jul 25 09:28:01 freemangordon: the GL stuff Jul 25 09:28:08 damn, horseshit makes me run at 30km/h?! Jul 25 09:28:09 there's a fix for it Jul 25 09:28:15 ah ok :) Jul 25 09:28:18 So I ran the track through gpsbabel telling it to discard wonky waypoints and it reduced that to a healthy 21km/h. Jul 25 09:28:37 vi__: maybe you did hit 30km/h Jul 25 09:28:40 DocScrutinizer05: yah, gives you gas propulsion Jul 25 09:28:41 and just didn't think you did Jul 25 09:28:46 DocScrutinizer05: Convieniently I was adjacent to a zoo. With zebras. WHich are a kind of a horse. Jul 25 09:28:46 freemangordon: it works with crazy CFLAGS Jul 25 09:28:47 did you hear rocky music playing? Jul 25 09:29:06 export LDFLAGS="-Wl,-rpath-link,$PWD/dist/bin/:/usr/lib:/lib -Wl,-rpath,/opt/qtm12/lib -Os -fno-strict-volatile-bitfields -flto=4 -fgraphite -ftree-vectorize -ffast-math" Jul 25 09:29:06 zeq1: what? Jul 25 09:29:09 export CFLAGS="${CFLAGS} -fno-strict-volatile-bitfields -flto=4 -fgraphite -ftree-vectorize -ffast-math" Jul 25 09:29:09 zebra is actually family of horse Jul 25 09:29:10 vi__: were they running with yu and you were keeping up? Jul 25 09:29:12 export CXXFLAGS="${CXXFLAGS} -fno-strict-volatile-bitfields -flto=4 -fgraphite -ftree-vectorize -ffast-math" Jul 25 09:29:17 WOW Jul 25 09:29:31 freemangordon: the above didn't break it (somewhat surprisingly) Jul 25 09:29:35 :) Jul 25 09:29:39 yeah :D Jul 25 09:29:43 The point is the gps accuracy sucks ass. Jul 25 09:29:50 is this the latest? Jul 25 09:29:58 vi__: i just use my lumia 900 to run ;) and endomondo Jul 25 09:30:11 actually that is not strictly true. Jul 25 09:30:28 but when i did use my n900's gps it seemed ok to me Jul 25 09:30:31 zeq1: is this the latest m-c? Jul 25 09:30:32 I just want a program that can intelligently discard bogus gps waypoints. Jul 25 09:30:38 minus the ovi maps being total crap Jul 25 09:30:41 that runs with those flags? Jul 25 09:31:04 it was the latest when you asked me to rebase yesterday Jul 25 09:31:08 ovi maps has to be the absolutely worst type of nav software i have ever used my entire life Jul 25 09:31:19 ok Jul 25 09:31:34 i think the 1980s PLGRs we had in the military were better than ovi maps Jul 25 09:31:43 using that and an actual paper map would get me somewhere faster Jul 25 09:31:45 yes, and the ssl problem had gone away (not sure if that was related to using crazy flags though) Jul 25 09:32:16 it defintely felt faster (although we know the js jit wasn't as quick as it has been) Jul 25 09:32:39 when is someone going to find out how i can get maemo on my transformer? :-P Jul 25 09:32:55 jonwil, I know how to add new predicate to binary prolog file Jul 25 09:33:13 jonwil, but I need your env where you can load that binary file into swipl Jul 25 09:33:30 freemangordon: m-c has been updated, shall I rebase? Jul 25 09:34:40 zeq1: yes Jul 25 09:34:54 zeq1: BTW libfakeroot is sb1 version Jul 25 09:35:02 hmmm, one problem is knowing if that policy-settings-rx51 guy is even still at Nokia Jul 25 09:35:12 freemangordon: also remember fennec won't build without the qt header being fixed Jul 25 09:35:22 ok Jul 25 09:35:40 jonwil, send email Jul 25 09:36:05 if he is not in nokia you immedialely got email responce about address is not correct Jul 25 09:36:16 freemangordon: I think that's okay though (correct) since it's for SB, not the device, right? Jul 25 09:36:16 googling now to see if I can find mention of him somewhere Jul 25 09:37:36 jonwil, you can add new predicate to binary file: load binary file into swipl, write new predicate to text file 'new.pl', run consult('new.pl'). and then run qsave_program('new_binary', []). Jul 25 09:37:59 will that override any existing predicate of the same name? Jul 25 09:38:09 no Jul 25 09:38:16 it will extend predicate Jul 25 09:38:58 but in prolog is unification, so prolog will try next term if he got false Jul 25 09:40:04 freemangordon: I've pushd the rebase Jul 25 09:40:35 jonwil, this http://pastebin.com/86t4iQ6u should enable jackbias when headset is connected Jul 25 09:40:44 (deleted that active call check) Jul 25 09:41:04 ok, gcc4.7 libs are in thumb repo Jul 25 09:42:00 cool :) Jul 25 09:42:10 /cs voice #maemo povbot_ Jul 25 09:42:30 meh Jul 25 09:42:47 morning DocScrutinizer05 Jul 25 09:42:54 ok, google shows that Krisztian was clearly involved in the opening of ohm bits for MeeGo (including the creation of policy-settings-basic) Jul 25 09:43:16 at the very least he may be able to share why policy-settings-rx51 cant be shared Jul 25 09:44:06 jonwil: hassle stkeeps. Jul 25 09:44:14 He seems to have access to EVERYTHING. Jul 25 09:44:19 yeah, on #meego Jul 25 09:44:36 Although he cannot tell you stuff covered by NDA, he may be able to hint... Jul 25 09:44:39 forgot about stkeeps Jul 25 09:44:42 ok Jul 25 09:44:48 if you say so pali Jul 25 09:45:05 I asked for scripts which genereate list of closed and open packages Jul 25 09:45:06 jonwil: stkeeps has access to ALL the closed source shizzle. Jul 25 09:47:32 ok, will email Krisztian then Jul 25 09:47:49 I doubt hassling stskeeps will help in this case Jul 25 09:48:43 given that Krisztian was involved in the opening of the ohm bits, he at the very least might be able to explain why policy-settings-rx51 was not opened and policy-settings-basic was created instead Jul 25 09:51:30 I think its worth linking to http://talk.maemo.org/showthread.php?t=83447&page=4 in the email to explain why we care about this stuff and whats important about it Jul 25 09:58:29 I think a question of "why did you do XY, in context of FOSSing stuff" is usually futile. Too much involvement of legal department, and management assigning work packages to particular groups Jul 25 09:59:07 find better questions, you only got a limited number of them you'll get answered Jul 25 10:00:27 e.g. "what are the differences between policy-settings-rx51 and policy-settings-basic, either on a diff or on an abstract level?" might already a better wuation than anything starting with "why" Jul 25 10:01:05 s/wua/ques/ Jul 25 10:01:06 DocScrutinizer05 meant: e.g. "what are the differences between policy-settings-rx51 and policy-settings-basic, either on a diff or on an abstract level?" might already a better question than anything starting with "why" Jul 25 10:02:13 jonwil: ^^^ Jul 25 10:02:18 yeah true Jul 25 10:04:19 ok, so pali, can you tell me exactly what jackbias is and why we want to be able to turn it on? Jul 25 10:06:35 jackbias aka micbias is a ->phantom powersupply for the headset microphone. The relevant paper says: 2V via 2kR. It's needed to power the mosfet amp in electret microphones, and also it detects pushbutton press on headset, which simply shorts mic line to GND Jul 25 10:08:34 basically it's just a pullup via 2k2 on ring3 of hs-jack, from a stabilized noise-free power surce Jul 25 10:08:39 source* Jul 25 10:09:33 we want to turn it on to enable recording from headset microphone and to enable detection of button press on headset Jul 25 10:10:20 freemangordon: when you get the chance, can you try rebuilding qmltube? It's the package that has DH_COMPAT=7... Jul 25 10:11:30 zeq1: there is a new autotools in extras-devel Jul 25 10:11:57 or new debhelper, or whatever Jul 25 10:12:43 http://talk.maemo.org/showthread.php?t=21182 Jul 25 10:12:44 except the debian-etch devkit gets used! I must be missing something. Jul 25 10:13:12 Pali: jonwil: ^^^ Jul 25 10:13:19 zeq1: where were debian Qt patches you pointed to some weeks ago Jul 25 10:14:19 freemangordon: hold on Jul 25 10:14:22 I want to merge once performance patch from there Jul 25 10:14:32 s/once/one/ Jul 25 10:14:34 freemangordon meant: I want to merge one performance patch from there Jul 25 10:15:40 MohammadAG: ping Jul 25 10:16:00 freemangordon: I looked the other day and they've moved onto 4.8 :/ Jul 25 10:16:18 I couldn't see any 4.7 in the debian patch system anymore Jul 25 10:16:39 fuuuck, isn't there som archive? Jul 25 10:16:57 there *must* be Jul 25 10:17:10 is there a reason we can't move onto 4.8? Jul 25 10:17:16 yea Jul 25 10:17:30 cssu is 4.7.2 Jul 25 10:17:31 isn't it compatible? Jul 25 10:17:36 NFC Jul 25 10:17:40 I mean cssu too Jul 25 10:17:48 but that is a major change Jul 25 10:18:13 it's not like jumping straight to qt5 though Jul 25 10:18:15 and having in mind tha buch of problems we had with 4.7.2 I will not do it Jul 25 10:18:30 zeq1: it is, maemo5 support was dropped long ago Jul 25 10:18:36 ah Jul 25 10:18:53 so it would need forward porting Jul 25 10:19:08 :( Jul 25 10:19:17 so, at least there is a very important reason to upgrade to 4.8 and/or above, I wouldn't touch it Jul 25 10:19:33 s/so, /,so is / Jul 25 10:19:33 freemangordon meant: ,so is at least there is a very important reason to upgrade to 4.8 and/or above, I wouldn't touch it Jul 25 10:19:46 hehe, typos day again Jul 25 10:19:52 :) Jul 25 10:20:28 maybe I'll take a look if I find myself with too much time on my hands Jul 25 10:21:45 is HARM also staying at 4.7.2? Jul 25 10:21:52 yes Jul 25 10:22:03 I take it Mer is using qt5? Jul 25 10:22:10 zeq1: no matter it is 4.8 now, where is the address? Jul 25 10:22:22 zeq1: Qt5 is not released yet Jul 25 10:23:27 looking for it Jul 25 10:26:03 Pali: http://talk.maemo.org/showpost.php?p=593405&postcount=64 http://talk.maemo.org/showpost.php?p=593468&postcount=67 http://talk.maemo.org/showpost.php?p=593592&postcount=74 !!! http://talk.maemo.org/showpost.php?p=659580&postcount=80 !!! Jul 25 10:27:31 Doc, postcount=74 is incorrect for our last pr1.3 prolog code Jul 25 10:27:36 freemangordon: http://patch-tracker.debian.org/package/qt4-x11/4:4.8.2-1 Jul 25 10:27:41 Doc, see decompiled prolog code Jul 25 10:27:57 I already written where is problem Jul 25 10:28:34 jackbias ON is defined if there is fact: active call or incoming call Jul 25 10:29:12 postcount 80 is hacking with one config file Jul 25 10:29:24 but this does not work correcly for pr1.3 Jul 25 10:29:42 ohm send some signals to alsaped to disable it sometimes Jul 25 10:29:57 Doc, I already read all this info month ago Jul 25 10:30:52 jonwil, jackbias in alsa/ohm/prolog/policy/other prog is switch which enable/disable button on wired headset, nothing more in our application context Jul 25 10:31:42 Doc, look at pr1.3 prolog code here: http://talk.maemo.org/showpost.php?p=1242365&postcount=39 Jul 25 10:34:07 DocScrutinizer05, I already tried to use libplayback to report active call via: PB_CLASS_VOIP and PB_STATE_PLAY Jul 25 10:34:18 This really enabled jackbias but MAFW in this state is muted Jul 25 10:34:25 so not useble for OMP Jul 25 10:35:12 freemangordon: any luck finding your patch? Jul 25 10:35:24 freemangordon, I assume the gcc-4.7 thumb compiled binaries will be up in a few days? Jul 25 10:37:26 Hurrian: it will be as and when they are built. The idea with having libstdc++6 and libgcc1 already available is to support them as they appear. Jul 25 10:37:33 Pali: whatever, I started recorder and it recorded from headset mic, so evidently micbias been enabled Jul 25 10:37:56 Pali: I did this when I tried to find out about multibutton-headset protocol used Jul 25 10:38:05 Doc, really button worked? Jul 25 10:38:14 zeq1, cool. My N900's a nice testing tablet nowadays, my daily phone stuff moved to N9. Jul 25 10:38:35 speaking of which, the battery life is fantastic now that the cell radio isn't being used Jul 25 10:38:37 also, recorder can set playback mode to active call VOIP Jul 25 10:38:39 damn, whenever micbias is enabled, yes the button shorts it to ground and you can detect this via hardware Jul 25 10:38:44 Hurrian: if you don't mind testing please go ahead and install them if you haven't already :) Jul 25 10:38:54 nfc if the sw bits for dbus signalling or whatever been in place Jul 25 10:39:03 zeq1: still no :( Jul 25 10:39:05 but micbias been active Jul 25 10:39:16 there must be an archive somewhere Jul 25 10:39:22 freemangordon: ^ Jul 25 10:39:26 Doc, enabling mic != enabling jackbias Jul 25 10:39:30 see alsamixer -c 0 Jul 25 10:39:35 YES! it is! Jul 25 10:39:57 when mic *works* it *needs* micbias to do so Jul 25 10:40:30 and micbias in turn will allow detection of button shorting it to GND Jul 25 10:40:51 ok, I will try it Jul 25 10:41:05 which app did you used for recording? Jul 25 10:41:11 freemangordon: maybe we can find an outdated debian mirror with the source pkg? Jul 25 10:41:15 want to see source code Jul 25 10:41:19 freemangordon: http://lists.linuxtogo.org/pipermail/openembedded-core/2012-April/021310.html Jul 25 10:41:32 I don't suggest fuzzing with alsamixer will automatically enable micbias whenever you unmute mic Jul 25 10:41:59 zeq1, I assume that was the patch for Qt that was needed Jul 25 10:42:00 but evidently recorder did something that did both enable mic and power up micbias Jul 25 10:42:31 Hurrian: no problem with that, in debian patch system there was a patch to speedup pixmap rendering Jul 25 10:42:47 or something Jul 25 10:43:01 Doc, now I tested Recorder app Jul 25 10:43:16 It recodered from headset mic Jul 25 10:43:19 zeq: wrong channel :P Jul 25 10:43:20 teotwaki: Pong Jul 25 10:43:21 but jackbias was OFF Jul 25 10:43:27 oops! Jul 25 10:43:30 so Doc, what you wrote is not true Jul 25 10:43:49 hehe Jul 25 10:43:58 freemangordon, it wasn't simply QT_GRAPHICSSYSTEM=raster? Jul 25 10:43:58 then I diunno what type of headset you use, since normally those microphones do not work without micbias Jul 25 10:44:09 Doc, it is not micbias Jul 25 10:44:12 but jackbias alsa switch Jul 25 10:44:19 I'm using standard n900 headset Jul 25 10:44:24 dafaq, there's only one hw line Jul 25 10:44:51 I do not know how this is in HW, but I see what SW is doing... Jul 25 10:45:07 from top to bottom: left, right, mic+button+CVBS-out, GND Jul 25 10:45:25 seems that nokia created other usefull SW alsa switch Jul 25 10:45:55 as soon as you want to record from mic+button+CVBS-out, you need to apply phantom supply aka micbias aka jackbias Jul 25 10:45:56 freemangordon: anyway, as I accidently was saying on the wrong channel, it has a link to vcs at the bottm Jul 25 10:46:23 and 4.7.x is tagged Jul 25 10:46:45 and button shorts mic+button+CVBS-out to GND, thus creating a pushbutton event that gets detected by dedicated hw in N900 Jul 25 10:46:48 yeah, saw that Jul 25 10:47:02 now i only need to remember which version to look in Jul 25 10:47:48 Doc, see this: http://svn.jacekowski.org/host_mode/trunk/sound/soc/omap/rx51.c Jul 25 10:48:02 there are more jack swicthes exported to alsa Jul 25 10:48:48 there is "Mic Jack" Jul 25 10:49:12 freemangordon: the qt4.7.3 brach should have it Jul 25 10:49:19 branch8 Jul 25 10:49:23 branch* Jul 25 10:49:25 your domain to analyze the fsckdup alse sw Jul 25 10:49:34 freemangordon: you're not the only one who can't type today! Jul 25 10:49:35 hw anyway works like I described Jul 25 10:49:54 yes baby: http://anonscm.debian.org/gitweb/?p=pkg-kde/qt/qt4-x11.git;a=blob;f=debian/patches/Improve-performance-of-partial-updates-in-raster-win.patch;h=82625af1be51deb06b3bd06761acda33c0e2738a;hb=qt4.7.3 Jul 25 10:50:28 DocScrutinizer05 ok, but for HW access on n900 is possible only that nokia alsa code Jul 25 10:51:00 I'd hope it's FOSS Jul 25 10:51:04 freemangordon: any other useful patches in there? Jul 25 10:51:21 and asla "Jack Bias Switch" must be enabled for wired headset button Jul 25 10:52:25 http://pastebin.com/1DTyBtag Jul 25 10:52:36 Thats what I have for the email so far Jul 25 10:52:57 I know alsaped is a friggin undocumented closed blob, but I never expected great magic in alsa itself Jul 25 10:53:13 any feedback on the email so far? Jul 25 10:53:21 freemangordon: what about: http://anonscm.debian.org/gitweb/?p=pkg-kde/qt/qt4-x11.git;a=commit;h=290508beb1b09df83a1f9bcae2a9d9b71a21b332 Jul 25 10:53:32 alsa mixer is just confusing maybe Jul 25 10:53:40 but not obscured Jul 25 10:53:59 freemangordon: never mind, it's just a packaging patch Jul 25 10:54:08 after all alsamixer is FOSS Jul 25 10:54:19 I was hoping it might fix the QGLWidget errors Jul 25 10:54:38 zeq1: well, for now I think performance patch, and your patches for 4.7 should be enough Jul 25 10:54:43 sorry, have to run Jul 25 10:54:45 sure Jul 25 10:54:48 doc, yes alsamixer only show kernel alsa switches Jul 25 10:54:57 brb Jul 25 10:55:05 and this switch is in that rx51.c sound code Jul 25 10:55:16 jonwil, for me looks OK Jul 25 10:55:44 jonwil, maybe you should write that jackbias is needed for working headset button Jul 25 10:56:26 "We really want to make jackbias work as it is needed to make the headset button on the wired headphones function properly." Jul 25 10:56:31 That sound good?> Jul 25 10:56:45 and also if it is enabled in alsamixer, then some app (ohmd, alsaped, prolog, ...) disable it in some minutes Jul 25 10:56:51 freemangordon: remember it fails to build without one of my my go-faster flags, not sure which, might be -flto Jul 25 10:56:58 jonwil, ok Jul 25 10:57:35 what should I use as a subject line? Jul 25 10:58:07 Info about maemo sound policy Jul 25 10:58:28 or Enabling jackbias Jul 25 10:59:19 "Question about maemo sound policy/Open Hardware Manager" Jul 25 10:59:24 that sounds good Jul 25 10:59:30 ok Jul 25 10:59:37 freemangordon: I've put my qmake.conf up on my site Jul 25 10:59:48 email sent Jul 25 10:59:55 will let everyone know if I hear anything back Jul 25 11:01:06 ok, it bounced Jul 25 11:01:33 ----- The following addresses had permanent fatal errors ----- Jul 25 11:02:06 jonwil: not surprising Jul 25 11:02:55 pali, any ideas what to do next? Jul 25 11:03:03 find working new address Jul 25 11:03:08 I started googling... Jul 25 11:03:09 how? Jul 25 11:03:11 ok Jul 25 11:03:15 if you find one let me know Jul 25 11:03:25 https://gitorious.org/~hackgyver Jul 25 11:03:36 freemangordon: if you use -flto, remember you need to have the optimization options in LDFLAGS too Jul 25 11:03:53 ok Jul 25 11:04:03 but i am far from that :) Jul 25 11:05:01 freemangordon: well as said ^^ I've put my qmake.conf up on my site if that helps ;) Jul 25 11:05:07 should I send him a message through gitorius then? Jul 25 11:06:11 message sent Jul 25 11:07:35 now if I could just find matches-to-n900 packages for libxcb-xv0-dev, libxcb-randr0-dev and libsp-timestamp-dev... Jul 25 11:08:54 jonwil, here is @intel address: http://www.mail-archive.com/nemo-commits@lists.merproject.org/msg00594.html Jul 25 11:09:23 try this Jul 25 11:09:51 ok, will do Jul 25 11:11:23 BTW, having looked deeper at the prolog stuff, I am now fairly sure that its not GPL Jul 25 11:11:30 i.e. that Nokia are not required to share the source Jul 25 11:12:21 the policy-settings-basic is GPL Jul 25 11:12:39 but I see nothing in policy-settings-rx51 to suggest that its GPL Jul 25 11:12:46 or otherwise violating someones license Jul 25 11:12:55 FSCK U Google, fsck u internettablettalk, I get 121311211324323 hits for every shit you could imagine, all due to >>20 minutes ago – What we know about alsa-policy-enforcement and alsaped? (40). to Nokia N900 by pali - 16 mins ago. overclocking kernel modules + UI for ...<< in "active threads" of the page Jul 25 11:13:12 when googling for alsaped Jul 25 11:13:51 ok, from the intel address I get "I'm currently out of the office." Jul 25 11:13:55 autoreply Jul 25 11:13:59 freemangordon: I've cherry-picked the fixes from mozilla-central/inbound, shall I push? Jul 25 11:14:05 zeq1: ok, put it, i can use it for reference Jul 25 11:14:31 which means the address is valid :) Jul 25 11:14:50 zeq1: seems lots of activity today on m-c, and sunspider regression patches seems to be backed out Jul 25 11:15:21 freemangordon: I've pushed the cherry-picks to our git repo Jul 25 11:15:26 yes, push them, hopefully it won't bring us problems Jul 25 11:15:28 :D Jul 25 11:15:29 ok Jul 25 11:15:48 oh, were you looking for my qmake.conf? Jul 25 11:15:56 it should be in maemo on my site Jul 25 11:16:01 ok Jul 25 11:16:45 I need an index.html :P Jul 25 11:17:42 zeq1: and it works ok with -ftree-vectorize -ffast-math Jul 25 11:17:48 qt i mean Jul 25 11:17:59 yes Jul 25 11:18:16 at least, my testing hasn't resulting in any problems Jul 25 11:18:27 hmm, and what is -fgraphite Jul 25 11:19:36 http://gcc.gnu.org/wiki/Graphite Jul 25 11:19:46 thanks Jul 25 11:19:49 ok, I think we have done all we can regarding jackbias right now Jul 25 11:19:50 jonwil, can you try to send that email to others ohm/dres/prolog/.. devs? Jul 25 11:20:06 dont know of any other devs Jul 25 11:20:19 look at debian/changelog and debian/control Jul 25 11:22:10 jonwil, here is libxcb-xv0-dev and libxcb-randr0-dev: http://repository.maemo.org/pool/maemo5.0/free/libx/libxcb/ Jul 25 11:22:14 Pali: GAIA ADCIN2 / ECI_AD is direct analog voltage detect of micline/ring3 Jul 25 11:22:57 zeq1: what LDFLAGS for Qt? Jul 25 11:23:32 Doc, we alredy found it in N900 HW spec Jul 25 11:23:51 when we tried to find twl madc ADCIN* for bat temp Jul 25 11:24:20 Pali: N4007/ECI0 is comparator output to OMAP IRQ pin U3 GPMC_nBE1 Jul 25 11:24:36 freemangordon: it should be in my qmake.conf Jul 25 11:25:55 Doc, my idea about jackbias switch is that this is SW which which drop headset button events... Jul 25 11:26:06 :nod: Jul 25 11:27:25 hw is just fine as long as you enable N4040 CODEC B11 MICBIAS_B Jul 25 11:27:56 and of course switch mux to mic (not TVOUT) Jul 25 11:30:30 can't find libsp-timestamp-dev Jul 25 11:33:36 it is really needed for freamtnel version? Jul 25 11:34:04 yes, it is Jul 25 11:34:12 for ohm-plugins-misc Jul 25 11:34:30 try scratchbox $ apt-get source libsp-timestamp-dev Jul 25 11:35:20 that doesn't help Jul 25 11:35:23 http://harmattan-dev.nokia.com/pool/harmattan/free/s/sp-timestamp/ Jul 25 11:35:25 nothing more Jul 25 11:36:04 ohm-plugins-misc_0.6.65+0m5_armel.deb does not depend on that libsp- package Jul 25 11:36:11 no Jul 25 11:36:14 just the source package Jul 25 11:36:20 so I think dependency for ohm-plugins-misc is useless Jul 25 11:36:20 and some other package it builds Jul 25 11:37:02 try version from harmattan now Jul 25 11:37:53 library libsp-timestamp is not in maemo5, so none maemo package depends on it Jul 25 11:38:11 try removing that package from debian/control and built it Jul 25 11:39:21 yeah Jul 25 11:45:57 ok, that worked Jul 25 11:46:03 ohm-plugins-misc is building Jul 25 11:49:44 btw, that libep sounds nice for understanding dbus signals sent by ohm daemon Jul 25 11:50:45 BTW pali, I couldn't find anyone else listed in the ohm stuff worth sending email to Jul 25 11:52:33 freemangordon: romaxa's header order build fix in inbound wasn't sufficient, he'd neglected to move the GLdouble_defined along with the includes. I've pushed a fix to our tree. Jul 25 11:53:39 ok Jul 25 11:54:20 I am off for lunch Jul 25 11:55:49 me too :) Jul 25 11:56:54 jonwil, ok Jul 25 12:03:22 hey, cool, I didn't know osso-games-chess was open source... :) Jul 25 12:07:15 it is? Jul 25 12:08:10 port the fritz 12 engine to n900 :o (closed source? pfff) Jul 25 12:38:10 bah, my computer speakers broke :( Jul 25 12:49:40 jonwil: lrn2headphones Jul 25 12:50:51 headphones dont give me the sound I like (and have gotten from my last 3 or so sets of 2.1 PC speakers) Jul 25 12:51:09 I like a bit of bass :) Jul 25 12:54:03 i have some really nice bluetooth headphones that have a shitton of bass Jul 25 12:54:06 perhaps even too much Jul 25 12:55:20 I also dont particularly like the way headphones feel Jul 25 12:58:34 kerio: I bet jonwil has a 2.1 system with a 20kg subwoofer :D Jul 25 12:58:41 haha Jul 25 12:58:51 my broken set is a Logitec LS21 Jul 25 12:58:53 challenge: attach that to the n900 Jul 25 13:01:53 I wont be buying another LS21, didn't particularly like them Jul 25 13:11:15 what are some good games for the n900? Jul 25 13:11:30 so far i have aisleriot Jul 25 13:11:53 klondike is especially good in fullscreen Jul 25 13:19:48 kerio: pokerth Jul 25 13:20:32 texas holdem sucks, lrn2five-card draw Jul 25 13:20:47 kerio: On the contrary. You suck. Jul 25 13:20:53 k Jul 25 13:20:55 :c Jul 25 13:20:58 anyway, apart from card games? Jul 25 13:21:10 kerio: NETHACK Jul 25 13:21:17 meh, good point Jul 25 13:21:27 except that vanilla nethack is horrible and unnethack is still at version 3.5.3 Jul 25 13:21:31 (version 4 was released a bit ago) Jul 25 13:21:52 (by vanilla nethack i mean the one from nethack.org as opposed to something like naohack) Jul 25 13:22:03 (which is still vanilla but the interface isn't absolute crap) Jul 25 13:22:11 I wish, for the love of god the punk would port doomrl to arm. Jul 25 13:22:20 I am fed up of his closed source horse shit. Jul 25 13:22:22 qemu it :D Jul 25 13:23:16 someone should crosscompile acehack for the n900 for me Jul 25 13:23:44 I tried to port penguin command and viruskiller. Jul 25 13:23:49 That was an utter fail. Jul 25 13:24:20 They both have hardcoded screen sizes...that are too big for the n900. Jul 25 13:24:44 vi__: the best you can do with osso-xterm is 79x25 :s Jul 25 13:25:35 also, osso-xterm considers brown as yellow :/ Jul 25 13:27:57 vi__: hm, acehack is surprisingly playable in 79x25 Jul 25 13:28:06 except that the last column of the status bar is cut :( Jul 25 13:29:11 kerio: install mrxvt, xterm etc. Jul 25 13:29:43 vi__: that sounds like too much effort though Jul 25 13:30:15 :3 Jul 25 13:44:04 vi__: what's your NAO handle? Jul 25 13:51:07 I literally have no idea what you are talking about Jul 25 13:54:43 your nickname on NAO Jul 25 13:54:50 or another public server Jul 25 13:54:53 (re:nethack) Jul 25 13:56:53 freemangordon: I've not looked into why yet, but for some reason our ifdefs aren't working anymore; moc_mozqwidget.cpp:74: undefined reference to `MozQWidget::orientationChanged()' Jul 25 13:57:10 vi__: if you never ascended on a nethack public server then you never ascended Jul 25 13:58:15 I have never ascended. I refuse to read the spoilers file and always die of quaffing from fountains at about level 6. Jul 25 13:58:35 well, maybe you should refrain from quaffing from fountains :) Jul 25 13:58:48 you'll never get anything good from them Jul 25 13:58:57 or, at least, the expected value of a fountain quaff is really, really bad Jul 25 14:01:34 Or this one time I used a stone to flesh spell on a boulder and kept eating it till I died. Jul 25 14:01:55 doing that is also bad, yes Jul 25 14:02:15 This one game I polymorphed my pet into a black dragon. Jul 25 14:02:19 By accident. Jul 25 14:02:37 so? they don't breathe when tame Jul 25 14:02:42 we just went around ransacking and looting everyshop. Jul 25 14:02:49 It was ace. Jul 25 14:03:02 oh, yeah, the extended AI patch Jul 25 14:03:06 Then I drank a potion or somthing retarded. Jul 25 14:03:10 and died. Jul 25 14:03:13 haha Jul 25 14:03:46 This one time I had a valkarie and gloves and I killed a cockatrice. Jul 25 14:03:58 and you went around while wielding the corpse? Jul 25 14:03:59 yeah, bad idea Jul 25 14:04:01 I was wielding that shit like a weapon, it was sweet. Jul 25 14:04:02 levitate or stay put Jul 25 14:04:17 Then I put on some cursed -4 boots. Jul 25 14:04:29 I tripped, fell over onto said cockatrice. Jul 25 14:04:31 ohoho, fumbling Jul 25 14:04:32 and died. Jul 25 14:04:32 neat Jul 25 14:04:48 that's one of the 47 ways you can die from a footrice Jul 25 14:06:41 zeq1: i've build qt with gcc 4.7, going to upload it Jul 25 14:06:54 ok Jul 25 14:07:14 freemangordon: thumby? Jul 25 14:07:29 I'm off to walk the dog now, bbl Jul 25 14:08:02 hm, about: is broken in microb Jul 25 14:08:27 kerio: No it isn't Jul 25 14:08:38 xml error, here Jul 25 14:09:28 kerio: yes Jul 25 14:09:34 freemangordon: ^_^ Jul 25 14:10:01 zeq: qtwebkit.so is 650k larger than one build with 4.6.2 :( Jul 25 14:11:02 freemangordon: which flags did you use in the end? Jul 25 14:11:35 won't upload it that way, will play a little with compiler switches Jul 25 14:11:40 zeq1: yours Jul 25 14:11:47 freemangordon: change the Ofast to Os Jul 25 14:11:56 yeah Jul 25 14:12:21 is that for Size or Speed? Jul 25 14:12:25 Os Jul 25 14:12:43 size, i think Jul 25 14:12:49 Os is size Jul 25 14:12:57 Ofast is maximum speed Jul 25 14:13:04 vi__: "about:" gives me an xml error Jul 25 14:13:04 moreso than O3 Jul 25 14:13:10 O2 fails with internal compiler error Jul 25 14:13:21 Ofast doesn't try to keep the size down Jul 25 14:13:24 at all Jul 25 14:13:40 I meant to change that, I was just testing :P Jul 25 14:13:57 :( Jul 25 14:14:02 yeah. but O2 fails Jul 25 14:14:04 re. ice Jul 25 14:14:54 but Os comes with the reduced speed AFAIK Jul 25 14:15:24 s/with the/with/ Jul 25 14:15:25 freemangordon meant: but Os comes with reduced speed AFAIK Jul 25 14:15:27 yes O2 is best compromise Jul 25 14:16:03 and it fails ;) Jul 25 14:16:16 does Os work? Jul 25 14:16:44 you changed both LFLAGS and CXXFLAGS? Jul 25 14:17:55 yes, both. what if I keep -Ofast in compiler, but set -O2 for linker? Jul 25 14:19:33 zeq1: what if I change -flto=4 to something like -flto=3 or whaterver? Jul 25 14:20:17 I think it's undefined if you mix Jul 25 14:20:42 lto=n where n breaks it up into smaller units Jul 25 14:21:00 so it can be parallelized like make -j Jul 25 14:21:21 otherwise the linking is a serial bottleneck Jul 25 14:21:27 freemangordon: isn't Os the whole point of thumb binaries? Jul 25 14:21:37 no Jul 25 14:21:48 it's a system that swaps heavily, having to grab less stuff from the swap means that it'll still be a lot faster Jul 25 14:21:56 you can only be sure of what's better by testing it Jul 25 14:21:56 again, no Jul 25 14:22:03 except that you'd have to benchmark the whole system .s Jul 25 14:22:04 :s Jul 25 14:22:10 kerio: that's generally true but some parts you want to run fast Jul 25 14:23:10 zeq1: i will try with -O2 without LTO Jul 25 14:23:11 freemangordon: is it the known 4.7 ICE? Jul 25 14:23:15 NFC Jul 25 14:23:31 I can't remember the bug number off-hand Jul 25 14:25:36 freemangordon: you could use Ofast and specifiy an unroll limit Jul 25 14:25:43 I mean inline limmit Jul 25 14:25:47 limit* Jul 25 14:26:01 sure, but that comes near to black magic :) Jul 25 14:26:29 you could reduce unrolling too probably Jul 25 14:27:05 anyway, i am on my way home, will check what has happened when i am there, bbl Jul 25 14:27:07 -fno-unroll-all-loops or something, I can't remember exactly which flags get set at Ofast Jul 25 14:27:28 -funroll-all-loops is definetely not considered a "fast" optimization Jul 25 14:27:36 most of the times it slows things down Jul 25 14:28:06 that's why I said no- :) Jul 25 14:33:23 zeq1: Dont you have a dog to walk? Jul 25 14:36:08 vi__: quite right, see you all later :) Jul 25 15:04:21 kerio: you're right I think, O3/fast doesn't enable unrolling. Jul 25 15:04:55 hrmpf, why is there no nc on the n900 Jul 25 15:05:01 this is bs, it's supposed to be unixy Jul 25 15:05:33 i can install socat but it's not the same thing! Jul 25 15:07:03 ugh, it's in the sdk repos Jul 25 15:08:52 will Bad Stuff happen if i enable it on the n900? Jul 25 15:08:55 not the sdk or the nokia-binaries one, mind you Jul 25 15:09:23 no problem to enable SDK and tools repos on the device Jul 25 15:09:32 the sdk one too? neat Jul 25 15:09:45 i thought that it wasn't a good idea to do so Jul 25 15:16:56 hrmpf, why is copying a file through wifi cpu-bound Jul 25 15:17:29 http://talk.maemo.org/showpost.php?p=1242804&postcount=44 Jul 25 15:17:48 kerio: csum? Jul 25 15:17:58 zeq1: csum for what? Jul 25 15:18:03 WEP? Jul 25 15:18:04 :s Jul 25 15:18:36 zeq1: still unable to build Qt Jul 25 15:18:55 freemangordon: does O3 work? Jul 25 15:19:17 didn't try it, will do it now Jul 25 15:19:22 oh god, 103mb of used swap Jul 25 15:20:23 kerio: that's not necessarily bad, unless it's things you need to swap in! :) Jul 25 15:20:54 freemangordon: I've been reading the gcc info Jul 25 15:21:07 freemangordon: there's lots of nice new optimization options Jul 25 15:21:16 especially from graphite Jul 25 15:21:36 sure, but how to find which one breaks compilation Jul 25 15:21:42 ok, we need the big guns for this Jul 25 15:21:49 can i export a file through mass storage mode? Jul 25 15:21:52 :D Jul 25 15:22:06 something between O2 and Ofast :) Jul 25 15:22:23 zeq1: yeah Jul 25 15:23:09 freemangordon: I'll try here as well, so we can cover more flags, but first I need to clean the bathroom! :P Jul 25 15:24:22 are the SDK repos enableable on the n900 then? Jul 25 15:25:18 freemangordon: info gcc does tell you which flags are enabled at each level Jul 25 15:25:40 so it should be too hard to find the one which works around the error Jul 25 15:26:00 hmm, seems it compiles with O3 Jul 25 15:26:40 17:16 < kerio> hrmpf, why is copying a file through wifi cpu-bound Jul 25 15:26:45 kerio: scp? Jul 25 15:26:51 if so, encryption Jul 25 15:27:12 StyXman_: i was using nc Jul 25 15:27:15 `-O3' turns on all optimizations specified by `-O2' and also turns on the `-finline-functions', `-funswitch-loops', `-fpredictive-commoning', `-fgcse-after-reload', `-ftree-vectorize' and `-fipa-cp-clone' options. Jul 25 15:27:21 so it's one of those Jul 25 15:27:29 1.7MB/s of raw transfer through wep shouldn't grind the system to a halt :( Jul 25 15:27:50 -ftree-vectorize is new there Jul 25 15:28:03 zeq1: yeah Jul 25 15:28:05 isn't it clang's -O4, pretty much? Jul 25 15:28:35 kerio: it the big auto-vectorize switch Jul 25 15:28:36 Pali: wooah! Jul 25 15:29:08 * kerio scoffs at these peasants using physical cords to play music Jul 25 15:29:10 jonwil, see post on tmo Jul 25 15:30:11 zeq1: i'm concerned about -ffast-math Jul 25 15:31:30 ok, thats good news Jul 25 15:31:46 sounds like the best way forward might be to hack the policy.,plc file somehow Jul 25 15:31:52 Kli delivers! Jul 25 15:32:04 * vi__ scoffs at kerios battery life. Jul 25 15:32:44 vi__: phone in my pocket, headphones around my neck, no cord between the two Jul 25 15:32:46 you can't explain that! Jul 25 15:33:35 kerio: Oh shi...my headset ran out of batteries! WTF am I to do?? Jul 25 15:34:16 vi__: look around the tram nervously because you're bored, then get home and charge them Jul 25 15:34:21 with lots of sadness :c Jul 25 15:34:31 If only the N900 supported the media buttons on the headset. That would be sweet. Jul 25 15:34:47 Unfortunatley all we have is this stupid jack bias PTT button. Jul 25 15:35:01 it's the engineers' fault Jul 25 15:35:14 for not producing infinite batteries Jul 25 15:35:19 jonwil, I will wait for answer if assert & retract is good idea Jul 25 15:36:27 jonwil, maybe you can link decompiled prolog code & predicates to that thread Jul 25 15:36:33 to have all in one place Jul 25 15:37:43 ok, posted Jul 25 15:39:40 Pali: was kli an N900 engineer? Jul 25 15:39:44 yes he was Jul 25 15:39:48 he did the ohm stuff Jul 25 15:39:53 is listed as maintainer for it Jul 25 15:40:33 waaaaaah why doesn't HAM respect my apt holds Jul 25 15:40:36 fucking asshole Jul 25 15:41:02 when I have time, I will copy decompiled code in policy-settings-basic to reconstruct policy-settings-rx51 Jul 25 15:41:19 one part is missing: dres decompiler Jul 25 15:42:09 well he gave us all the pointers to how to decompile a dresc file Jul 25 15:42:28 :nod: Jul 25 15:43:29 jonwil, write decompiler is real challenge :) Jul 25 15:43:41 freemangordon: so what are these big changes coming to cssu-thumb repo? Jul 25 15:43:52 vi__: gcc 4.7 Jul 25 15:43:59 yippee! Jul 25 15:44:14 actually I do not really know how that benefits. Jul 25 15:44:20 faster better code :) Jul 25 15:44:24 auto vectoriser Jul 25 15:44:26 w00t Jul 25 15:44:48 yeah, simd support without intrinstics Jul 25 15:45:06 I have an idea. Jul 25 15:45:32 If I stop modest from autostarting with dsme(?) will it die gracefully when I quit it? Jul 25 15:45:54 so yeah I think with what we have now it should be possible to come up with something functionally equivalent to the n900 policy.plc file Jul 25 15:45:55 brb, off to braek my n900 again. Jul 25 15:47:23 freemangordon: back Jul 25 15:47:48 freemangordon: help! Jul 25 15:47:55 Qt still compiles, i want to check binaries size with -O3 Jul 25 15:48:00 the sdk repos are making me update libgcc1 Jul 25 15:48:10 no, wait Jul 25 15:48:12 kerio: no way Jul 25 15:48:17 it is extras-devel Jul 25 15:48:26 or cssu-thumb Jul 25 15:48:45 community-thumb is at 4.6.2 Jul 25 15:48:52 this one is at 4.7.2 Jul 25 15:48:54 maybe -devel Jul 25 15:48:55 hmm Jul 25 15:48:56 re. fast-math, it does have its detractors because it simply doesn't work for code which depends upon strict ieee conformance. Jul 25 15:49:20 kerio: that is cssu-thumb, disable the repo ror now Jul 25 15:49:22 when it does work though, it's a big win, especially with neon Jul 25 15:49:38 i currently have the -thumb one installed Jul 25 15:49:42 zeq1: it is too risky, i think we should disable that Jul 25 15:50:09 most of the problematic programs check for it Jul 25 15:50:20 hmm, sure? Jul 25 15:50:24 well, ok Jul 25 15:50:26 sqlite for example Jul 25 15:51:06 does qt have conformance tests Jul 25 15:51:07 ? Jul 25 15:51:30 but it was me to find why CSSU Qt 4.7.2 raster engine is broken. It was because of -O3 and it was not fun Jul 25 15:51:47 zeq1: NFC Jul 25 15:51:48 we should try the tests Jul 25 15:52:05 make check Jul 25 15:52:22 can you try that? Jul 25 15:52:48 ugh, i typoed the token code for the nokia-binaries repo Jul 25 15:52:49 :s Jul 25 15:52:53 I'll need to build with O3 unless you discovered the flag? Jul 25 15:53:17 zeq1: why, you have it now with -Ofast Jul 25 15:53:27 which is -O3 aiui Jul 25 15:53:50 ok trying Ofast Jul 25 15:54:06 btw i think tests are disabled Jul 25 15:54:17 we should probably get into the habit of trying make check Jul 25 15:54:40 zeq1: https://bugzilla.mozilla.org/show_bug.cgi?id=777094 Jul 25 15:54:42 04Bug 777094: was not found. Jul 25 15:56:10 yeah, I fixed it differently Jul 25 15:56:26 yeah, but that is upstream :P Jul 25 15:56:47 I'll back out my fix Jul 25 16:03:07 freemangordon: no seriously, where does this libgcc1 come from D: Jul 25 16:03:18 cssu-thumb Jul 25 16:03:46 no, i installed that Jul 25 16:03:58 those are the 4.6.2 packages Jul 25 16:04:08 well, ok Jul 25 16:04:14 * freemangordon sighs Jul 25 16:04:41 use apt-cache show libgcc1 Jul 25 16:04:53 * vi__ passes FMG the bong. Jul 25 16:04:55 hold on, HAM is still apt-working Jul 25 16:05:08 fucking sonuvabitch Jul 25 16:05:11 :s Jul 25 16:05:38 hm, weird Jul 25 16:05:45 i did policy Jul 25 16:05:58 1:4.7.2-0+cssu0 on community-cssu Jul 25 16:06:06 unless i enabled cssu-d Jul 25 16:06:08 er Jul 25 16:06:11 community-thumb Jul 25 16:06:18 unless i enabled community-devel by mistake? Jul 25 16:06:51 no, there's no libgcc1 there Jul 25 16:07:03 freemangordon: seriously though, why is there no matching package in http://maemo.merlin1991.at/cssu/community-thumb/pool/free/libg/libgcc1/ ? Jul 25 16:07:03 :S Jul 25 16:07:36 Is there a stream of conscience coming out of your mouth IRL? Jul 25 16:07:38 qt still building Jul 25 16:07:55 cat /proc/swaps Jul 25 16:08:02 ... Jul 25 16:08:06 wrong xterm Jul 25 16:08:07 kerio: http://maemo.merlin1991.at/cssu/community-thumb/pool/free/g/gcc-4.7/libgcc1_4.7.2-0%2bcssu0_armel.deb Jul 25 16:08:19 oic Jul 25 16:08:22 upgrade or not? Jul 25 16:08:41 should not harm according to zeq1 Jul 25 16:08:52 zeq1: i trust you! Jul 25 16:09:07 same for libstdc++6? Jul 25 16:09:51 also, what's up with hildon-initscripts? the sdk repo has the same version but with a different revision Jul 25 16:10:42 I think if libgcc and libstdc++ didn't work one of the distros would have noticed by now :) Jul 25 16:11:14 zeq1: trying -O2 and no -ffast-math Jul 25 16:11:43 can someone with the sdk repos enabled on his n900 tell me if "upgrading" hildon-initscripts breaks stuff? Jul 25 16:12:01 i have them, no problem Jul 25 16:12:15 kerio: you have BM? Jul 25 16:12:26 freemangordon: yeah but my last backup is like a day old Jul 25 16:12:32 a day in which i did many things D: Jul 25 16:12:41 so, what you risk exactly? Jul 25 16:12:42 like installing most of cssu-thumb Jul 25 16:12:50 the repo is still there Jul 25 16:12:52 ...eeeeeh Jul 25 16:12:53 good point Jul 25 16:13:02 do anothe backup and play in your will ;) Jul 25 16:13:10 *on Jul 25 16:13:23 i just wonder why it's a different revision Jul 25 16:13:29 -dp1 instead of -0m5 Jul 25 16:13:43 zeq1: tests are disabled in our Qt build Jul 25 16:14:00 I'm going to run make check manually after it finishes Jul 25 16:14:18 or is that disabled somehow? Jul 25 16:14:32 ugh, "Small changes for the SDK" Jul 25 16:14:34 WHICH ONES Jul 25 16:14:42 zeq1: NFC Jul 25 16:15:40 we'll soon find out Jul 25 16:15:57 zeq1: but honestly, I wouldn't risk breaking stellariumobile and stuff like it for unknown performance increase Jul 25 16:16:38 ftree-vectorise should give us enough performance gain Jul 25 16:17:20 I've read it isn't really effective on Cortex-a8 without fast-math :/ Jul 25 16:17:46 if it passes the tests it should be fine Jul 25 16:17:50 I know that, but better a little bit slower and working code Jul 25 16:17:57 zeq1: ok Jul 25 16:18:17 it's what the tests are for afterall :) Jul 25 16:20:02 :) Jul 25 16:27:44 freemangordon: fennec rebased Jul 25 16:32:50 ok, can someone explain to me what apt-worker has to do for 9 minutes and 40 seconds after i close ham? Jul 25 16:32:56 9:40 and counting, i mean Jul 25 16:33:11 I JUST CLOSED IT I DON'T WANT IT TO KEEP DOING THINGS Jul 25 16:34:07 kerio: it's a very good question, I've often wondered Jul 25 16:34:12 10 minutes 55 seconds Jul 25 16:34:41 11 minutes 21 seconds Jul 25 16:35:07 i mean, i do have a lot of repos, but come on Jul 25 16:37:49 kerio: still going? Jul 25 16:37:55 done now Jul 25 16:38:00 :) Jul 25 16:38:09 12 minutes, AFTER closing every window Jul 25 16:38:13 I wonder what it did :) Jul 25 16:38:16 a normal user would have no idea it's doing stuff Jul 25 16:38:24 except for the fact that everything runs slooooooooooow Jul 25 16:46:43 zeq1: is there any chance gcc to have some patches in the meanwile? Jul 25 16:47:02 so... right now i have mr0, ssu, extras-devel, community-testing, community-thumb, sdk, sdk tools, sdk nokia-binaries enabled Jul 25 16:47:02 :3 Jul 25 16:47:04 no matter what i do i am unable to compile qt with -O2 Jul 25 16:47:24 there's still no excuse for HAM to take 12 minutes to *close* Jul 25 16:49:34 zeq1: hmm, it actually builds now, WTF? Jul 25 16:49:48 with -O3 -fno-inline-functions -fno-unswitch-loops -fno-predictive-commoning -fno-gcse-after-reload -fno-ipa-cp-clone Jul 25 16:50:22 seems there is another options -O3 is enabling Jul 25 17:01:31 completely unrelated things: why aren't the maemo repos using pdiffs for the Release file? Jul 25 17:26:47 freemangordon: the info gcc I used in the one on my laptop, which is for 4.7.1 Jul 25 17:27:12 yeah, that is the latest available documentation Jul 25 17:27:13 I looked at that since we don't have info pages installed Jul 25 17:27:25 on gcc site too Jul 25 17:27:28 ok Jul 25 17:27:59 well, i backported the XSync patch and trying to make the final build :) Jul 25 17:28:14 perhaps we should take a look in the source? although, how is the bin size with the above options? Jul 25 17:28:52 i hope it will give us that "Measured frame rate improvement of 20 - 60 % in an update stress testing the performance of partial updates." Jul 25 17:29:11 zeq1: still not built Jul 25 17:29:15 ok Jul 25 17:29:35 make check gave no errors btw Jul 25 17:29:53 well, i am compiling with -ffast-math anyway :D:D:D Jul 25 17:30:04 what tests did it execute? Jul 25 17:30:11 the full testsuite requires x11 Jul 25 17:30:42 aah. but you can run xephyr Jul 25 17:31:29 not sure! I'm building the full testsuite now :) Jul 25 17:31:51 yes, you can run xephyr in host OS Jul 25 17:31:55 as :1 Jul 25 17:31:59 ok Jul 25 17:32:17 or I suppose I could use a remote X server Jul 25 17:32:31 but,but it is just that Jul 25 17:32:51 xephyr is lightweight X server :D Jul 25 17:33:12 you can run in window Jul 25 17:33:14 ;) Jul 25 17:33:15 I mean the X server on my laptop :) Jul 25 17:33:23 hmm, yeah Jul 25 17:33:27 you could Jul 25 17:35:11 there are benchmarks too Jul 25 17:35:20 :) Jul 25 17:35:53 tests/benchmarks Jul 25 17:36:04 you can just run qmake; make Jul 25 17:43:06 I don't think they keep the testsuite quite up-to-date Jul 25 17:55:30 freemangordon: it doesn't *need* X Jul 25 17:56:04 obviously some of the tests do! Jul 25 17:56:04 zeq1: ? Jul 25 17:56:10 aah, yeah Jul 25 17:56:13 but it's running now without it Jul 25 17:56:45 do you run it in SB? Jul 25 17:57:03 I am running it in SB, some tests are quite slow to complete though Jul 25 17:57:14 run it on the device ;) Jul 25 17:57:14 is SB faster than the real thing? Jul 25 17:57:34 does not matter, you don;t have VFP/NEON Jul 25 17:57:43 I mean the real thing Jul 25 17:57:53 SB is waaay slower Jul 25 17:58:03 yeah Jul 25 17:58:11 qemu that is Jul 25 17:58:35 hmm, packaging ;) Jul 25 17:58:36 I'll copy it over once I put it on charge, my battery just died! Jul 25 17:59:34 we could package the testsuite too..? Jul 25 17:59:46 bad time to suggest it though! ;) Jul 25 17:59:51 yeah Jul 25 18:00:17 I don't see a reason why we can't, but yes, not now :) Jul 25 18:05:04 ok, Qt built, lets see if it works :D Jul 25 18:13:02 I need to install make Jul 25 18:13:52 and qmake... but it doesn't exist! :( Jul 25 18:14:02 yeah :( Jul 25 18:14:13 btw Qt works on the device, and is smaller Jul 25 18:14:14 what a pain Jul 25 18:14:20 ah good! Jul 25 18:14:32 I'd still like to get the testsuite running though Jul 25 18:14:32 that's my devel device, going to put it on primary Jul 25 18:14:42 zeq1: but of course Jul 25 18:14:48 I guess we should be able to build a qmake bin from the qt sources...? Jul 25 18:14:50 btw there is a -test package Jul 25 18:15:06 I *think* that's something else Jul 25 18:15:28 the QtTest lib Jul 25 18:15:32 whatever that's for Jul 25 18:15:42 not the same thing as the testsuite Jul 25 18:15:42 yeah Jul 25 18:19:35 running qmake; make in the qmake directory seems to work Jul 25 18:20:41 ok, going to put Qt in the repo Jul 25 18:20:46 you probably noticed the build produces an x86 qmake normally Jul 25 18:22:41 * zeq1 is going to eat now Jul 25 18:46:16 zeq1: stelaruimobile, qmltube and sociality work with updated Qt Jul 25 19:08:53 freemangordon: good news! Jul 25 19:09:21 yeah, PhoneME works too, as is fennec, I am out of ideas what to try next Jul 25 19:11:49 are any of the geolocation apps qt based? Jul 25 19:14:10 zeq1: can't remember Jul 25 19:20:20 my status menu has 11 items Jul 25 19:20:28 it fills the screen nicely in portrait Jul 25 19:43:50 Estel_: ping Jul 25 19:45:59 is anyone here familiar with the easy debian chroot scripts? Jul 25 19:53:36 zeq1: still unable to build fennec, those stupt GLdouble errors :( Jul 25 19:54:08 freemangordon: shall I fix them again? Jul 25 19:54:17 can you ? Jul 25 19:54:19 I'll try rebasing again first Jul 25 19:54:23 ok Jul 25 19:54:28 I can fix them Jul 25 19:54:57 i/m afk for a while Jul 25 19:55:24 I've not been able to get a sufficiently working qmake to run the testsuite on my N900 Jul 25 19:57:02 I can't install libqt4-dev because it's too big (and not optified) Jul 25 20:05:47 freemangordon: I've not tested it, but I think I've fixed it correctly. Jul 25 20:06:08 the rebase didn't help Jul 25 20:06:58 I'm building fennec now Jul 25 20:07:03 it's already pushed Jul 25 20:12:01 for fuck's sake, why is binutils in -devel still not fixed Jul 25 20:14:34 kerio: fixed? what? Jul 25 20:15:11 DocScrutinizer51: the Provides: line for some version of binutils doesn't follow the specs Jul 25 20:15:34 you can't Provide a particular version, which is what the maintainer wants to do, apparently Jul 25 20:16:09 or was it that you can't Provide a range of versions? either way, it's not something that apt likes Jul 25 20:16:15 hmmm Jul 25 20:16:17 you get a warning if you do apt-get update Jul 25 20:17:55 Installed: 2.18.50.20070820-4+0m5 Jul 25 20:18:29 no warning Jul 25 20:18:54 maybe it's because you installed it? idk Jul 25 20:20:00 maybe it's because I usually don't have extras-dsevel enabled? Jul 25 20:21:11 maybe Jul 25 20:21:14 :c Jul 25 20:21:28 but that's just because you're a masochist who uses HAM :P Jul 25 20:21:40 apt-get has no problems with tons of huge repos Jul 25 20:23:32 I'm actually using apt-get most of the time. That's because I _know_ when I better use HAM Jul 25 20:24:36 and I'm not as silly as doing an apt-get upgrade ;-P Jul 25 20:25:02 do you mean dist-upgrade? Jul 25 20:25:06 I prefer to check each single pkg upgrade Jul 25 20:25:13 oh, you're actually crazy, i see Jul 25 20:25:18 it's ok i still love you <3 Jul 25 20:25:54 it's not like i'd see more than one group of packages upgraded at the same time, considering how often i play with this Jul 25 20:25:57 for the rest there's HAM :-P Jul 25 20:26:26 which actually doesn't annoy me a bit for upgrades Jul 25 20:26:44 as it's just fire and forget Jul 25 20:27:19 DocScrutinizer51: well, HAM took 12 minutes to properly close for me, today Jul 25 20:27:29 i might have a bit too many repos though Jul 25 20:27:50 eventually I close HAM, and I don't give a sh* if it takes 30 min Jul 25 20:28:10 though never takes more than 5 for me Jul 25 20:28:30 i have mr0, ssu, extras-devel, community-testing, community-thumb, sdk, sdk tools, sdk nokia-binaries Jul 25 20:28:47 it's a miracle it ended, to be fair Jul 25 20:29:01 i just wonder... *what* does it do, after every window is closed? Jul 25 20:29:08 can't it just, you know, stop? Jul 25 20:29:24 update db? Jul 25 20:29:30 what for? Jul 25 20:30:02 errr, maybe to write back any changes done? nfc Jul 25 20:31:44 strace will tell you ;-) Jul 25 20:40:22 is there a way to enable both mass storage *and* usb networking? Jul 25 20:40:30 or would that require a much more complicated gadget module? Jul 25 20:41:52 hrmpf, why is usb mass storage so fast? Jul 25 20:53:10 hi Jul 25 20:53:47 I just purchased an n900 on eBay, wanted to get the "computer in your pocket" feeling I got from my n810 that I was missing for all this time. :) Jul 25 20:55:19 hmm, this channel seems dead now, it seemed more alive at 3am in the morning. ;-) Jul 25 20:55:43 brb Jul 25 20:55:52 * RST38h sacrifices a hamster to the Tentacled One Jul 25 21:03:49 RST38h, hi Jul 25 21:03:49 EdLin: neat! Jul 25 21:04:05 EdLin: file down the plugs on the usb cable and the wallcharger *right now* Jul 25 21:04:18 then flash both COMBINED and VANILLA as explained in the wiki Jul 25 21:04:20 ~flashing Jul 25 21:04:20 kerio, I think it's going to be neat, there are some complaints about the seller on talk. Jul 25 21:04:21 maemo-flashing is, like, http://wiki.maemo.org/Updating_the_tablet_firmware Jul 25 21:04:29 or there Jul 25 21:04:38 oh, you paid for it but you still don't have it Jul 25 21:04:39 i see Jul 25 21:04:40 :s Jul 25 21:04:41 I should have checked before I pulled the trigger. Jul 25 21:04:46 how much did you pay? Jul 25 21:05:02 $162.50, plus I ordered $9.99 expidited shipping. Jul 25 21:05:25 not bad, even if it really is refurbished without a stylus included as reported. Jul 25 21:05:37 eeh, 133€ Jul 25 21:05:40 i paid about 150€ for mine Jul 25 21:05:45 but it was almost in mint condition Jul 25 21:06:01 barely used, everything cleaned up, only a minor scratch on the screen Jul 25 21:06:02 everything working Jul 25 21:06:07 $150 was used, I thought this was new... Jul 25 21:06:17 but apparently its refurb Jul 25 21:06:25 there's no such thing as a new n900 Jul 25 21:06:28 not anymore Jul 25 21:06:39 so I gather. Jul 25 21:06:59 * kerio shakes his fist at Elop Jul 25 21:07:02 damn youuuuuuu Jul 25 21:07:44 Nokia was really taken over by Microsoft, only reason why they didn't finish the job instead of just give them a brain transplant with key executives is because of Finland's foreign ownership laws. Jul 25 21:08:46 and like Microsoft does to everyone they borg, they're screwing them over. Jul 25 21:10:27 any of you know a good source for N900 stylusii? Since mine is bound not to come with one apparently. Jul 25 21:14:23 hmm, google shopping is turning up big capacitive stylii. Wrong! Jul 25 21:14:56 EdLin: find a tree Jul 25 21:15:01 snap a twig Jul 25 21:15:04 use the twig Jul 25 21:16:11 kerio, thanks. But there are no trees in Brooklyn. ;-) Jul 25 21:16:17 madness Jul 25 21:16:29 actually, there are, just not enough except in big parks. Jul 25 21:16:42 or neighborhoods more classy than mine Jul 25 21:17:43 looking on Amazon, I found a poorly reviewed chinese 5800 stylus knock-off that didn't fit the hole in even the device was advertised for, and a gaggle of capacitive styluses. Jul 25 21:17:44 ok, now I shot my ears with superlux HD668B :-S - maybe now it's time to drive them in for a week, allegedly the treble and hiss goes away a bit by doing so Jul 25 21:18:10 I'm going to be pissed if I have to buy it from the same folks that sold it to me without it. Jul 25 21:18:29 assuming talk.mameo.org is right about cellphonesforever. Jul 25 21:19:50 maemo I mean Jul 25 21:20:20 how do you pronounce Maemo, btw? :) Jul 25 21:20:40 I've had an n810 for years, always wanted to know. ;-) Jul 25 21:20:55 * EdLin is talking to himself Jul 25 21:21:05 one does not pronounce Maemo Jul 25 21:21:18 kerio, I see, its the ineffable name. Gotcha. Jul 25 21:23:25 it was a word made up by a password generator :D Jul 25 21:25:51 may-mo Jul 25 21:26:11 EdLin: You bought your n900 through ebay? Jul 25 21:26:17 vi_, that's the way I've always been pronoucing it Jul 25 21:26:20 vi_, yes Jul 25 21:26:23 good. Jul 25 21:26:47 as the buyer you hold all the cards. Jul 25 21:26:58 they also rated me good already. :) Jul 25 21:27:10 ebay is rigged to totally fuck over the seller if anything goes wrong. Jul 25 21:27:43 1. Your device is NOT refurbished. It is second hand. Jul 25 21:28:09 it was listed as new. :( Jul 25 21:28:14 2. As soon as you receive it check it over. COMPLETELY. Jul 25 21:28:24 EdLin: ebay sellers are scum. Jul 25 21:28:40 especially ones in HK, I should have known. Jul 25 21:28:58 start a thread on TMO asking about what you need to check out. Jul 25 21:29:25 vi_, can't you tell me that here? Already started one thread there about buying it. Jul 25 21:29:28 I cannot be bothered listng, I am typing on my n900 and it is not conducive to long sentances. Jul 25 21:29:31 ah Jul 25 21:30:00 the biggest things to look out for are USB port damage. Jul 25 21:30:07 and slider damage. Jul 25 21:30:29 the USB port on the n900 is shit. Jul 25 21:30:31 how do I tell if they're damaged? Beyond the obvious (i.e. not working) Jul 25 21:30:54 you cannot really tell :( Jul 25 21:31:16 sigh, guess I should keep my backup phone instead of hocking it then. Jul 25 21:31:24 The usual fi is to open it up and solder it down to the main board. Jul 25 21:31:30 ^fix Jul 25 21:31:39 anyway. Jul 25 21:31:54 I'm not very handy with a soldering iron, as I found out by majoring in electronics engineering tech. Jul 25 21:31:56 assuming it is all there and works fine then happy days. Jul 25 21:32:14 EdLin: then it sounds like you are over qualified. Jul 25 21:32:18 lol Jul 25 21:32:25 it is easy, a chimp could do it. Jul 25 21:32:34 yeah, I learned how to do NASA-spec joints. Jul 25 21:32:52 the community college served an apprentiship program at NASA. Jul 25 21:32:58 if it is missing the stylus just say the item is not as described. Jul 25 21:33:04 through ebay. Jul 25 21:33:20 ask the seller to send you one. Jul 25 21:34:08 I'll do that I reckon. Hopefully they won't give me a hard time. Jul 25 21:34:13 Refurbished can mean anything from a brand new case, to wiped over with a damp rag. Jul 25 21:34:20 if he offers to sell you one tell him it is a standard acessoriy and you expecteded it to be included (unless the auction specifically says it was not included.) Jul 25 21:34:26 SpeedEvil: yup. Jul 25 21:35:23 basically you can make life hell and cost the seller a lot of money. Jul 25 21:35:57 you can demand a refund, get him to agree then make him pay for the shipping. Jul 25 21:36:18 ebay is stacked in the buyers favour. Jul 25 21:36:31 we have a saying in the motorcycle world. Jul 25 21:36:57 'don't sell your shit on ebay unless you want to get raped by cunts' Jul 25 21:38:03 vi_, I sold stuff in craigslist, twice. It managed to work out OK, though I hear some people get screwed. Jul 25 21:38:39 asked to meet in a public place, only accepted cash, no shipped items, etc. Jul 25 21:39:05 and I'm in Brooklyn, where the scammers are probably among the worst. Jul 25 21:39:34 I prefer to sell stuff to people I know though. Jul 25 21:44:10 Ebay is unfortunately fail for local collection. Jul 25 21:44:45 Then again - I did have a seller that shipped me 2*1800mm*54mm glass tubes. Jul 25 21:44:55 In three layers of bubblewrap only. Jul 25 21:45:05 It went back in a rather smaller box. Jul 25 21:45:21 They claimed it was 'industry standard' packaging. Jul 25 21:57:55 freemangordon: Lemme know if you need some testing help with near-future thumbed libs again. Jul 25 22:10:42 Raimu: feel free to apt-get upgrade ;) Jul 25 22:11:02 just make sure you don't have extras-devel enabled :P Jul 25 22:11:14 Heh, what's in extras-devel now? :D Jul 25 22:11:20 NFC Jul 25 22:12:06 just in case :) Jul 25 22:12:15 Huh. Well, I'll backup and see what's new in the repo. Jul 25 22:12:41 Raimu: i've already upgraded my primary device, so I assume it safe Jul 25 22:12:58 but yeah, you'd better make a backup Jul 25 22:45:31 SpeedEvil: lol Jul 25 22:48:27 SpeedEvil: btw wtf are you doing with 2 pcs 2m glass tube Jul 25 22:49:13 thermal sun collectors, neon signs, laser? Jul 25 22:49:34 solar vacuum tube collectors. Jul 25 22:49:49 :nod: Jul 25 22:50:02 I broke two prior to properly installing on roof. Jul 25 22:50:04 there's been a reason it was first in my list Jul 25 22:50:44 :P Jul 25 22:51:12 hell!!! either amarok is a pile of s**t, or this lenovo T500 has broken audio amp Jul 25 22:53:12 I'm temped to plug the headphone out to the mic/line-in and check a sinewave playback with audacity recording Jul 25 23:09:39 dnag, it must be phonon, or (surprise surprise) PA, that introduces terrible clipping even when master volume is turned down to 10% of max Jul 25 23:10:05 or actually amarok Jul 25 23:37:10 ok, was amarok's equalizer Jul 25 23:37:27 chinese. Jul 25 23:37:36 I have to restart irc client again, *sigh* Jul 25 23:48:17 vi__, pong Jul 25 23:48:23 what do you want from ed chroot?:) Jul 25 23:49:07 DocScrutinizer, you won't belive it, but I actually re-used this IR diode, and after 1,5 hour of carefull preparing (needed to re-do plate, as one on N900 board was fubar gone), and soldering it, it works liek a charm Jul 25 23:49:12 s/liek/like/ Jul 25 23:49:16 Estel_ meant: DocScrutinizer, you won't belive it, but I actually re-used this IR diode, and after 1,5 hour of carefull preparing (needed to re-do plate, as one on N900 board was fubar gone), and soldering it, it works like a charm Jul 25 23:49:50 of course, it required even more scratching on surface, now I really have 2mm of ground-less space aroun pad, and pad look like island. Jul 25 23:50:14 funny facty - IR diode seems to be lighing with more power (comapred with replacement N900 that got shipped today) Jul 25 23:50:19 compared* Jul 25 23:50:34 lolo Jul 25 23:50:37 (i.e. looking via digital camera - repaired N900's IR looks considerably brighter) Jul 25 23:50:39 stronger distance to turn off TVs and grief your friends Jul 25 23:50:51 or in waiting longues Jul 25 23:50:57 always msot shitty tv shows are ON there Jul 25 23:51:03 http://www.youtube.com/watch?v=9bZkp7q19f0&feature=youtu.be Jul 25 23:51:05 I love TVBGone! widget Jul 25 23:51:05 real music ^ Jul 25 23:51:08 me too Jul 25 23:51:18 doesn't work that much tho, on my country Jul 25 23:51:18 :\ Jul 25 23:51:24 That music is going to make you jealous Jul 25 23:51:58 well, check this ;) http://www.youtube.com/watch?v=NhYKex-j4_0 Jul 25 23:52:18 well, as for TVBGone! widget for n900, I have yet to see TV that it doesn't turn off. although, I would love to update it with fresh IR codes Jul 25 23:52:43 how far can you turn them off? Jul 25 23:53:38 it depens on rpecise aiming, N900 diode generates highly tunelled beam Jul 25 23:53:55 but if You don't want to play sniper without a scope, comfortable range is few meters, unfortunately Jul 25 23:54:03 IDK now, after repair, I don't have TV to test on Jul 25 23:54:39 once I thought about putting difussing lens there Jul 25 23:54:44 took from some remote Jul 25 23:54:54 most of them have difussing lenses, generating wide beam Jul 25 23:55:21 well, on my parent's house, there is Ir Tv and remote, that works even when aimed exactly backward. No idea why, probably, through reflections Jul 25 23:55:33 i have a remote like that Jul 25 23:55:34 too Jul 25 23:56:03 you can point the remote ir light down your chin Jul 25 23:56:15 so your skull will somewhat work as a, erm, I don't know how it is called, but it spreads the signal Jul 25 23:56:24 never tried to do that with my phone and I would look plain stupid Jul 25 23:56:28 well, actually, it would be fun mod to attach powerful high pwoer IR diode to N900, just controlled via same transistor as current diode (but with other power source) Jul 25 23:56:45 ;P Jul 25 23:57:08 have anyone tried updates iptables from extras devel? Jul 25 23:57:18 Nah nah nah. USB controlled IR! Jul 25 23:57:21 they sit there for a month, without changelog or TMO post/thread about WTF Jul 25 23:57:29 robbiethe1st, sure, that would be best Jul 25 23:57:35 although USB control for IR is kinda bitchy Jul 25 23:57:42 everything I know works through serial Jul 25 23:57:43 ftdi Jul 25 23:57:48 ^^ Jul 25 23:58:01 as for iptables, i'm usually cautious about updates without changelogs or annoucements Jul 25 23:58:09 http://www.huitsing.nl/irftdi/ Jul 25 23:58:11 And usb->serial chips do work with the N900 Jul 25 23:58:14 nox-, what is it? Jul 25 23:58:34 Speaking of which, I've run a PL2302-based usb->serial adaptor from my N900; worked great. Jul 25 23:58:35 robbiethe1st, but DIY IKR things doesn't work via usb->serial adaptorsa, most of the time Jul 25 23:58:38 an usb serial chip with bitbang mode needed for ir Jul 25 23:58:48 nox, ah Jul 25 23:58:50 sounds nice Jul 25 23:59:01 Uh, why not? Jul 25 23:59:15 Estel_ one REALLY powerful then Jul 25 23:59:18 nox-, so using ftdi based adaptor, most things, including IR diodes and receivers (!) should work? Jul 25 23:59:27 Estel_, yes Jul 25 23:59:32 lircd -Hftdi Jul 25 23:59:32 it would be nice, it would allow to make N900 also a "learning remote" Jul 25 23:59:40 (i.e. recording Ir patterns from remotes) Jul 25 23:59:51 Should. Anything that can have a serial input/output will work with a serial->usb adaptor and the N900 Jul 25 23:59:51 or you could use an mceusb device i guess Jul 26 00:00:25 robbiethe1st, serial<->usb without bitbang mode is useless for ir remote stuff tho Jul 26 00:00:41 nox-, sounds very interesting, where could I read more about it? any examples of how to detect ftdi based serial adaptor, when buying via net, or maybe it is DIY'able? Jul 26 00:01:04 well, most cheap serial->usb adaptors I've seen lacked bitbang Jul 26 00:01:05 Digitus DA-70156 Jul 26 00:01:14 Uh, why? Jul 26 00:01:17 is one Jul 26 00:01:39 You have hardware signaling lines, you can always bitbang them? Jul 26 00:01:45 robbiethe1st, not via usb Jul 26 00:01:48 wuuuut, 20 usd? Jul 26 00:01:56 Uh, why not? Jul 26 00:01:57 no cheaper chinese things? Jul 26 00:02:12 I /know/ they're there, I /know/ they work... not fast enough? Jul 26 00:02:17 not that it is bad, if it's good quality Jul 26 00:02:35 but I wonder, if chinese copycats also use this chip? Jul 26 00:02:50 FTDI? yes. Jul 26 00:03:04 nox-, do you know anything about parallel adaptors for USb and their usage with N900? Jul 26 00:03:14 Or the common rs232 chip, the PL2302. Look for Arduino programming adaptors. Jul 26 00:03:39 PL2302 also works OK with Ir things etc? Jul 26 00:03:41 you can also look for chinese mceusb clones Jul 26 00:03:41 (pl2302 puts out rs232 voltages, versus fdti which puts out 0v/3-5v instead Jul 26 00:04:03 but only those that dont just emulate a keyboard Jul 26 00:04:04 Estel_, no clue. I know it works with the Fadal CNC Mill we have at work Jul 26 00:04:17 And that's old, relatively standards-complient serial. Jul 26 00:04:40 PL2302 doesnt have bitbang mode i think Jul 26 00:05:14 this might be mceusb: http://www.ebay.ca/itm/Brand-NEW-Original-Genuine-HP-USB-MCE-IR-Wireless-Receiver-Win7-Vista-US-fast-/270929871250?pt=US_Remote_Controls_Pointers&hash=item3f14adc192 Jul 26 00:05:15 Which begs the question: why do you *need* bitbang mode? Jul 26 00:05:41 robbiethe1st, normal usb serial protocol is too slow Jul 26 00:06:10 for bitbanging the control lines Jul 26 00:06:31 I'm guessing you're dealing with a 'dumb' IR device? Jul 26 00:06:37 yes Jul 26 00:06:47 ir reciever chip or ir diode Jul 26 00:06:52 versus something that can be sent a couple of characters, and send out the bit-stream based on that. Jul 26 00:07:02 yes sure thats something else Jul 26 00:08:17 Estel_, never used parallel usb but i suppose they're also too slow for bitbanging ir Jul 26 00:09:03 nox-, actually, I though about on-the-field mainteintance of CNC machines that use parallel and linuxCNC Jul 26 00:09:04 How fast are we dealing with? what sort of frequency are these IR signals transmitted at? Jul 26 00:09:04 software Jul 26 00:09:28 drivers compatible with linuxCNC use parallel, most of the times Jul 26 00:09:58 I know N900 wouldn't be super-hyper beast at generating steps for drivers, but for debugging/mainteintance it should suffice Jul 26 00:10:14 Well, if you can get a driver for it(possibly by compiling the module from general Linux source), I'm sure it'd work Jul 26 00:10:57 nox, what do you think about that? Jul 26 00:10:58 http://allegro.pl/konwerter-usb-rs232-com-na-ft232-rl-ftdi-i2494850791.html Jul 26 00:11:02 don't worry about polish language Jul 26 00:11:10 photos and parameters are universal Jul 26 00:11:23 I'm little lsot amongst many types of ftdi chips Jul 26 00:11:30 this one looks like slower than digitus (same price) Jul 26 00:11:36 at leats reading poarameters Jul 26 00:11:50 digitus claims up to !mbps compatibility, this one 450 kbps, I don't know if it matters really Jul 26 00:11:57 Estel_, as long as its really ftdi it'll probably work Jul 26 00:12:23 For serial, most things aren't faster than 115200 baud, 115.2kbps Jul 26 00:13:53 robbiethe1st, 36 kHz carrier or similar Jul 26 00:15:01 For IR? Yeah, I'd want a 'smart' chip there. Jul 26 00:15:50 https://en.wikipedia.org/wiki/RC-5 Jul 26 00:15:55 one popular protocol Jul 26 00:19:33 nox-, sorry for bugging with direct question, but something not-completely-assembled like that: Jul 26 00:19:34 http://allegro.pl/konwerter-usb-rs232-ttl-interfejs-ft232-rl-kabel-i2493207148.html Jul 26 00:20:00 how it's compared to digitus? i.e. digitus is "open", or chip is covered in unpenetrable black something? Jul 26 00:20:48 Also, I wonder if linux compatibility is no problem as long as it's FTDI, or it also depends on assembled unit architecture Jul 26 00:21:02 I'm absolutely green on such adaptors, so again, sorry for pestering :) Jul 26 00:21:04 digitus is just a normal usb-to-serial black covered dongle that happens to use an ftdi inside Jul 26 00:21:15 * Estel_ nods Jul 26 00:21:42 i _think_ ftdi all have the needed bitbang mode... Jul 26 00:21:47 thing I've linked is cheaper and it's not dongl;e, + *claims* to have much more possibilities, no idea how to read between the lines properly, when it comes to such things Jul 26 00:21:59 so that thing you liked should do as well Jul 26 00:22:23 of course it require putting flex and serial port there, I suppose ;) Jul 26 00:22:23 lol Jul 26 00:23:07 well this one uses ttl levels not rs232 levels but ofc if you want to connect an ir chip/ir diode thats better Jul 26 00:24:11 if you just want something thats ready made for ir get an mceusb but they maybe a bit more pricey Jul 26 00:26:39 no no, i would like multi-purpose thing, expandable to whatever may come to my mind latter Jul 26 00:26:45 the more universal it is, the better Jul 26 00:27:05 erm, what is "better" generally, not only for IR diode, ttl levels or rs232 levels? Jul 26 00:27:30 also, any experiences with ATmega8U2 chips? Arduino switched from using ftdi to this one, programmed as USB->serial converter Jul 26 00:27:36 no idea about reasoning Jul 26 00:27:40 rs232 levels if you want to connect to `nromal' serial ports Jul 26 00:27:45 normal even Jul 26 00:28:20 i think theres a popular chip to convert ttl to rs232 levels called max232 Jul 26 00:28:36 Yup Jul 26 00:28:45 Several versions, infact. Jul 26 00:28:52 heh Jul 26 00:29:08 max232? I have 10 of such chips lying around Jul 26 00:29:20 :) Jul 26 00:29:29 once wanted (and still want, just need to find time) to do serial mod to my wrt54gl, or was it usb 1.1 port? I don't remember Jul 26 00:29:47 maxim is very kind to send free samples to literally anyone willing Jul 26 00:30:21 requiment - you must order at least 50 of them and declare 5000 units produced per year, or they may think You're fake ;p Jul 26 00:31:04 robbiethe1st, as for using parallel to usb converter for linuxCNC machine, main concern is that 99% adaptors emulate only things needed for printers Jul 26 00:31:13 and linuxCNC use, in fact, all pins of parallel Jul 26 00:31:27 linuxCNC and CNC drivers compatible with it Jul 26 00:32:03 nah, max232 was definitely needed for USB mod in wrt54gl Jul 26 00:35:57 usb doesnt use rs232 levels... Jul 26 00:36:06 so must be serial connection Jul 26 00:36:43 see you all :) Jul 26 00:45:33 nox-, IIRC, in wrt54g;, some things for usb 1.1 mod require same pins as used for serial Jul 26 00:45:38 or I've mixed everything up :) Jul 26 00:46:04 might be, but you only need rs232 levels if you want to connect as serial Jul 26 01:02:31 ShadowJK, when using charge21.sh and bq27200.sh 5, at the same time, their results are mutually exclusive Jul 26 01:02:54 I mean that charge21.sh shows charging current quite high ammount off, of what bq27200.sh shows (the latter seems to be real) Jul 26 01:03:07 I don't mean only system usage calculation, but whole ammount Jul 26 01:03:23 is it normal? Jul 26 01:03:35 charge21.sh acts like it's guessworking no only system load, but whole current balance Jul 26 01:03:55 it's "generally" ok, but up to 150 mA wrong, sometimes (more in extreme cases) Jul 26 01:05:50 What's charge21.sh? Jul 26 01:06:37 I know I've got a working charge.sh in BackupMenu; just lets the bq2700 do it's thing with reasonable defaults and seems to work. Jul 26 01:07:36 charge21.sh seems like updated version of charge.sh Jul 26 01:07:40 ShadowJK, I see what the problem is Jul 26 01:08:13 charge21.sh shows both "Rate" and "system" basing it on assumption, that it's chargin on full declared ammount (1250 mA, in my case) Jul 26 01:08:33 still, it's wrong, as for a life of mine, it want to charge only with 950 mA max Jul 26 01:08:48 well, 1050 at beginning of charging, for few minutes Jul 26 01:08:58 then, both "Rate" and "System" is going fubar Jul 26 01:09:46 consideriong, that charge SH know real relative current (i.e. one going to battery, = whole current - used by system) couldn't it use it as base for calculations,. and show it as "Rate"... Jul 26 01:10:04 then, try to calculate system load, and show it as "System" (-some_value) Jul 26 01:10:22 so user would knwo that real charging value is "Rate" - "System"? Jul 26 01:10:33 or even 3th value, as a result of this calculation, for slow math ones? Jul 26 01:10:53 now, when it is guessworking both rate and system, whole calculation is quite useless, as it's wrong 99% of time (at least in my case) Jul 26 01:11:27 oh nice, MS offers Windows 7 ISOs Jul 26 01:11:37 i.e. it always try to have "Rate" + "System" as = 1250, so, as a result, both "Rate" and "system" are wrong, with the latter being wrong totally (showing 300 mA, when real usage is 8 mA, for example) Jul 26 01:11:41 * MohammadAG closes some piracy website Jul 26 01:12:08 don't worry, in next version they will force You to use beecons, MohammadAG ;) Jul 26 01:12:25 You though that windows mobile was bad? Now You will have it in Your desktop too ;P Jul 26 01:12:25 if anyone needs the links http://www.digital-kaos.co.uk/forums/f94/original-untouched-windows-7-sp1-iso-32-bit-64-bit-220877/ Jul 26 01:12:43 this is actually a huge step Estel_ Jul 26 01:12:50 sure, backward one Jul 26 01:12:54 now I don't need to pirate ISOs to install a legal copy Jul 26 01:12:56 nope Jul 26 01:13:05 no no Jul 26 01:13:09 we're talking about two different things Jul 26 01:13:19 I tyalk about desktop in "winbdows 8" or whatever it will be called Jul 26 01:13:22 you're talking about a different thing :P Jul 26 01:13:33 as for iso's, I couldn't care less :D Jul 26 01:13:41 don't worry, in next version they will force You to use beecons, MohammadAG ;) Jul 26 01:13:59 well, considering this is #maemo channel, we both talk about "different" thing ;) Jul 26 01:14:13 I need to punch my ISP in the balls Jul 26 01:14:20 10MBps speed isn't 1000kbps Jul 26 01:15:02 It remind me, that I must finally sign contract for 50 mbps net, instead of leeching neighbour's !mbps one Jul 26 01:15:03 Sure, it is. In ISP-speak Jul 26 01:15:14 s/!mbps/1mbps/ Jul 26 01:15:15 Estel_ meant: It remind me, that I must finally sign contract for 50 mbps net, instead of leeching neighbour's 1mbps one Jul 26 01:15:31 robbiethe1st, I expect 1.4MB/s downloads speeds Jul 26 01:15:34 not 1.0 Jul 26 01:15:41 yes, 0.4 makes a difference :P Jul 26 01:15:52 heh Jul 26 01:16:04 1000 kbps would be 128 KB/s download speed Jul 26 01:16:17 so You need to punch them in the balls, but not so hard Jul 26 01:16:47 some german cable isps match for `speedtest' in urls and drop limiting... Jul 26 01:17:06 well, when You've written: Jul 26 01:17:10 MohammadAG> 10MBps speed isn't 1000kbps Jul 26 01:17:16 You though You actually talk about 80 mbps Jul 26 01:17:17 1000kb/s Jul 26 01:17:27 it's about B and b Jul 26 01:17:44 bytes/bits Jul 26 01:17:50 bite Your ISP, anyway ;) Jul 26 01:17:52 nox-: Cool... just need to proxy everything through something with speedtest in the name Jul 26 01:18:02 robbiethe1st, hehe Jul 26 01:18:12 robbiethe1st, lol Jul 26 01:19:01 (i.e. ppl get the full speed only for speedtests and lower otherwise) Jul 26 01:19:04 robbiethe1st, make Your cheap p;ayware proxy with such name, and You're a millionare Jul 26 01:19:10 heh Jul 26 01:19:12 hahaha Jul 26 01:19:13 do You imagine, how many ISP are doing that? Jul 26 01:19:25 probably not few... Jul 26 01:19:30 at leats until they introduce burts/packet size limits Jul 26 01:19:40 so You can upgrade your proxy too Jul 26 01:19:51 and business goes on :) Jul 26 01:20:20 better than payware softporn site Jul 26 01:20:32 although actual "coding" is much less pleasant Jul 26 01:23:47 uhh Jul 26 01:23:52 ...o_o Jul 26 02:49:29 Estel_; yes I changed that to 1250 when you wanted to test charging at 1250. It's usually vaguely correct as long as charger is able to provide requested current, even if charger chip is charging at lower rathe than that Jul 26 02:51:14 Estel_; oh also the RS value might be different from your charge21.sh and bq27200.sh version. I use RS=21 normally Jul 26 02:51:18 (as that was the empirically derived value by some MeeGo kernel hackers back in the day) Jul 26 02:52:49 although, the "parasitic resistance" as I call it, might be actually higher than 132 when charging at >950 Jul 26 02:53:55 (CR parameter) Jul 26 02:53:55 btw, rate is net rate in/out of battery, as measured by bq27200 Jul 26 02:55:55 sys and ch are made up. sys guesstimates current used by system, ch guesstimates current provided by charger. If the calculation shows a ch rate higher than the set rate, it gets clamped to set rate. If charger for some reason is unable the source enough current from attached dc adapter, the calculation starts showing that deficit as system use, wrongly Jul 26 02:56:11 * RST38h moos at ShadowJK Jul 26 02:56:29 ok, time for work **** ENDING LOGGING AT Thu Jul 26 02:59:58 2012