**** BEGIN LOGGING AT Tue Jul 31 02:59:59 2012 Jul 31 03:16:02 merlin1991: ping Jul 31 04:55:50 DocScrutinizer, freemangordon : dselect() is stable now, the test runs without problems and *MUCH* faster! To be expected I think. Jul 31 04:57:46 The locale problem is resolved, it stemmed from the configuration selecting -O3 optimization, which is much stronger with gcc-4.7. Jul 31 06:28:14 zeq1: sorry, too much noise in the channel logs, what was the preoblem? Jul 31 06:28:58 and it seems I was disconnected an hour ago Jul 31 06:30:10 Something in the locale code, possibly, although not necessarily related to Nokia's changes was miscompiling at -O3, which is the optimization default in the libc6 pkg. Jul 31 06:31:14 it was okay with gcc-42, but then that has a much weaker O3. Jul 31 06:31:28 aah, ok Jul 31 06:31:37 what now, -O2? Jul 31 06:31:44 yes Jul 31 06:31:56 I don't see why it should be O3 anyway Jul 31 06:31:57 yeah Jul 31 06:32:42 better to try to keep the code tight and in cache Jul 31 06:32:52 yes, sure Jul 31 06:33:53 anyway, the test ran without any problem, and much, much faster. Jul 31 06:37:18 freemangordon: I saw a fennec update in the repo, it seems to run okay. Jul 31 06:38:14 or rather HAM saw it :) Jul 31 06:42:41 zeq1: yeah, the one in the thumb repo is ok Jul 31 06:42:54 morning, freezeq :) Jul 31 06:43:10 morning Jul 31 06:43:17 morning Estel_ Jul 31 06:43:22 :) Jul 31 06:43:31 zeq1: BTW there is some development on the bug, going to read it Jul 31 06:43:57 hmm "Status: RESOLVED FIXED " Jul 31 06:44:00 fennec! sure, I knew there is something elft that I had to test, but forget Jul 31 06:44:22 freemangordon: can you paste the url? Jul 31 06:44:39 * Estel_ just finished testing usb Mode and sending Pali bugs found Jul 31 06:44:44 https://bugzilla.mozilla.org/show_bug.cgi?id=778580 Jul 31 06:44:46 04Bug 778580: was not found. Jul 31 06:45:22 Estel_: is the the new hostmode that lets you charge at the same time from external power? Jul 31 06:45:28 BTw, messing with MCe is bitchy. I did something wrong, and I had situation, where my screen was *always* off - even after reboot - and every use of sldie emchanism was like with TV-out and external screen - status bar disappearing for a while, then appearing again Jul 31 06:45:45 zeq, yea, although old hostmode allowed same thing Jul 31 06:45:57 ne wone is jsut properly integrate dinto kernel, and have properly written GUI Jul 31 06:46:02 it did? Jul 31 06:46:07 hackish way Jul 31 06:46:21 You needed to invoke "boostoff" from terminal, and then, invoke charging script Jul 31 06:46:28 ah Jul 31 06:46:36 say thanks to MohammadAG, who haven't found time for updating GUi in 3 years Jul 31 06:46:58 that is why i'm extremely happy with usb Mode, as Pali is going to update it when needed, for sure Jul 31 06:47:14 LOL, what a massive backout Jul 31 06:47:30 zeq1: will you pull m-c? Jul 31 06:47:35 sure Jul 31 06:48:16 BTW, improved booston script used in current HEN was available for 18 months before, but again, no one wanted to update GUI. At the time when I was learning how to upload things to repositories, I announced, that I'm going to start with providing new version of HEN. MohammadAG got so scarred that I'm going to make a testbed of it, that he released update ;P Jul 31 06:48:45 well, update was just like replacing 1 script in whole package, but still Jul 31 06:49:00 erm, enough history of Maemo, volume :535 Jul 31 06:49:30 as for MCE - finishing my short story, I needed to powercycle device, to bring it into working state Jul 31 06:50:01 unless no-backlight and no-lock + diode pattern going indefinitely is considered as working Jul 31 06:50:09 at least it didn't blow up :P Jul 31 06:50:13 I hope that we will be able, one day, to replace mce Jul 31 06:50:15 yea ;) Jul 31 06:50:39 good luck replacing MCE, its a very complex piece of code :P Jul 31 06:50:46 but iamgine my WTF when slide lock didn't locked device, but blinked statusbar only, disabled touchpad (but keyboard was still working and responding!)... Jul 31 06:50:51 trust me, I should know :) Jul 31 06:50:55 ...and run charging pattern via led Jul 31 06:51:06 jonwil i know Jul 31 06:51:07 freemangordon: any commits I need to pull before pushing? Jul 31 06:51:15 but once it was said that we will never be clsoe to replacing bme :P Jul 31 06:51:17 zeq1: no Jul 31 06:51:25 MCe is real biatch, so I just hope Jul 31 06:52:17 Well, time to stop generating "nosie in channel" (blinks to freemangordon) Jul 31 06:52:27 noise, ffs Jul 31 06:52:27 :nod: :P Jul 31 06:52:38 see ya, gentlemans! Jul 31 06:53:03 bye Estel_ Jul 31 06:53:21 didn't rebase cleanly, need to jiggle the patches Jul 31 06:53:38 hmm, why is that? Jul 31 06:53:46 aah, ! thingie Jul 31 06:53:58 maybe Jul 31 06:57:49 no, somebody changed nsnull to nullptr Jul 31 06:58:09 in widget/qt/nsWindow.cpp Jul 31 06:59:08 it was code surrounded by our #ifdef's so it lost context Jul 31 07:00:14 freemangordon: do we want the ! thingie patch in? Jul 31 07:00:24 Estel_: see ya Jul 31 07:02:32 zeq1: lets try without it, I could easily add it locally if needed for some testing Jul 31 07:02:48 I'm trying to isolate the flag causing the libc breakage, I think it might be -ftree-vectorize Jul 31 07:03:15 ok I'll reset HEAD^ Jul 31 07:03:15 hmm, it is auto-enabled on -O3 iirc Jul 31 07:03:21 yep Jul 31 07:03:27 zeq1: ok Jul 31 07:04:46 pushed Jul 31 07:05:03 freemangordon: ^ Jul 31 07:06:16 I now have a working thumb libc with kernel the missing syscalls working :) Jul 31 07:06:36 good, going to pull fennec Jul 31 07:06:46 -rwxr-xr-x 1 root root 796.2K Jul 31 07:57 /lib/libc-2.5.so Jul 31 07:07:13 -rw-r--r-- 1 root root 397.6K Jul 31 07:57 /lib/libm-2.5.so Jul 31 07:07:29 freemangordon: what was the libm size before? Jul 31 07:08:13 zeq1: libc 1160576 Jul 31 07:08:25 nice improvment :) Jul 31 07:08:26 libm 448316 Jul 31 07:08:39 yeah Jul 31 07:08:39 not bad :) Jul 31 07:09:15 debs? Jul 31 07:09:46 I might have to email them, my server is currently down for maintainence Jul 31 07:10:41 wut? when it will be up again, i was thinking to copy toolchain in merlin's server and to write a short howto page on wiki :( Jul 31 07:10:58 it might be up Jul 31 07:11:29 I lied, it is up Jul 31 07:11:34 :) Jul 31 07:11:35 ok Jul 31 07:11:47 but it is undergoing maintainence, all the same :) Jul 31 07:11:58 I'll put the debs up Jul 31 07:12:28 although, shouldn't we have a dependency flag like for the thumb errata? Jul 31 07:12:36 yeah Jul 31 07:12:45 for [armel] only Jul 31 07:13:05 zeq1: isn't it better to ARM buil fit Jul 31 07:13:12 *build it Jul 31 07:13:16 not thum Jul 31 07:13:35 that way it can be used outside thumb repo Jul 31 07:13:38 well for thumb repo thumb is better Jul 31 07:13:59 no reason I can't make a non-thumb version for cssu Jul 31 07:13:59 well, yeah, but then the version string should be libc-...-thumb0 Jul 31 07:14:23 still needs to depend on sigmask syscalls Jul 31 07:14:32 WTF? "Automatic merge failed; fix conflicts and then commit the result." Jul 31 07:14:44 i did git reset just before pull :( Jul 31 07:14:54 ? Jul 31 07:15:05 probably the null change Jul 31 07:17:32 but how im i supposed to pull ?!? Jul 31 07:20:19 you could locally resolve the conflict Jul 31 07:20:56 but I don't want to do that. anyway I rm -rf the directory, will clone from scratch Jul 31 07:21:01 the only other option would be to reclone, Jul 31 07:21:06 :) Jul 31 07:22:38 I'm building cssu kernel with new syscalls Jul 31 07:22:52 my testing has been with kp Jul 31 07:22:59 do you have .diff ? Jul 31 07:23:15 so I can push it on gitorious Jul 31 07:23:19 yes, shall I put it up Jul 31 07:23:32 on your server? yes Jul 31 07:23:52 I used quilt to apply it so you shouldn't have a problem with it Jul 31 07:25:18 in maemo/sigmask Jul 31 07:25:50 freemangordon: ^ Jul 31 07:27:08 zeq1: ok Jul 31 07:31:40 shall I put up my glibc diffs too? Jul 31 07:31:58 yes Jul 31 07:32:41 zeq1: hmm aren't they in the source package? Jul 31 07:33:14 they would be if I built one :) Jul 31 07:33:59 just put up the debs including source pkg then? Jul 31 07:34:22 dpkg-buildpackage -rfakeroot -sa -us -uc Jul 31 07:35:16 and put the original source package in the ../ dir, so it will make diff.tar.gz package Jul 31 07:35:47 zeq1: ^^^ Jul 31 07:36:29 fuck, clonning fennec takes ages Jul 31 07:36:58 ok will do Jul 31 07:37:40 zeq: you should rename the original source package to something like [libc].orig.tar.gz Jul 31 07:37:52 s/should/should also/ Jul 31 07:37:53 freemangordon meant: zeq: you should also rename the original source package to something like [libc].orig.tar.gz Jul 31 07:47:31 freemangordon: something to be aware of (very important) Jul 31 07:48:04 you need to have the linux headers installed before building glibc Jul 31 07:48:18 otherwise it doesn't obviously fail Jul 31 07:49:45 so I need to put in that feature dependency Jul 31 07:50:35 otherwise you end up with a libc completely lacking pselect()! Jul 31 07:50:59 zeq1: hehe Jul 31 07:51:17 not good! :) Jul 31 07:51:32 well, at least the symptoms will be quite unique then Jul 31 07:51:43 zeq1: I will add kernel-feature-libc-pselect. Jul 31 07:51:51 Pali: ^^^ how that sounds? Jul 31 07:52:30 umm, who'd doing pselect now? Jul 31 07:52:30 probably better to make it sigmask or something Jul 31 07:52:43 libc, or kernel? or none of both? Jul 31 07:52:48 kernel Jul 31 07:53:11 DocScrutinizer: it's much faster Jul 31 07:53:18 so it shouldn't be named kernel-feature-libc-pselect Jul 31 07:53:20 and doesn't hang anymore :) Jul 31 07:53:32 ok Jul 31 07:53:32 but this is for package *linux*-kernel-power-headers Jul 31 07:54:04 Pali: ok Jul 31 07:54:26 zeq1: what sigset? would you elaborate Jul 31 07:54:33 sigmask even Jul 31 07:54:44 see the patch name Jul 31 07:55:11 yeah, patch name is rather common best practice for feature name Jul 31 07:55:15 it's not just pselect(), but 3 syscalls related to the same thing Jul 31 07:56:36 I wonder how it's going to affect power consumption? Jul 31 07:56:59 how would it? Jul 31 07:57:13 dbus is a heavy user of pselect() Jul 31 07:57:22 HAH Jul 31 07:57:28 and the performance difference isn't slight Jul 31 07:57:49 ooh, there's a performance improvement from patched pselect() ? Jul 31 07:57:55 big Jul 31 07:58:12 WTF? what did they do in "old" lib? Jul 31 07:58:59 busy loop waiting for signals? or what Jul 31 07:59:11 I think so Jul 31 07:59:18 also several syscalls to emulate 1 Jul 31 07:59:27 that doesn't help Jul 31 07:59:44 lots of trips to and from kernel Jul 31 07:59:54 :nod: Jul 31 08:00:28 still pselect() usually results in process going sleep Jul 31 08:01:07 so yes, in a special testbed there might be improvements in CPU load Jul 31 08:01:19 or turnaround time of a tight loop Jul 31 08:01:31 it can't hurt Jul 31 08:01:38 definitely Jul 31 08:01:43 :-) Jul 31 08:09:07 DocScrutinizer05: so, what now, community kernel? Jul 31 08:09:26 for fixing pselect() ? Jul 31 08:09:35 and one more bug Jul 31 08:09:46 BT related Jul 31 08:09:59 if you could demonstrate the stock kernel has bug, then probably yes Jul 31 08:10:15 DocScrutinizer05: no need, it is CVE Jul 31 08:10:32 demonstrate as in "run the test code that summons the erratic behaviour" Jul 31 08:10:56 Dit is CVE, there is no need me to demonstrate it Jul 31 08:11:03 *it Jul 31 08:11:18 ping merlin1991 Jul 31 08:11:20 you asked for my notion. well that's it Jul 31 08:12:38 I'd expect you feeling happy about an answer way less negative than you probably expected Jul 31 08:12:49 ;-) Jul 31 08:15:11 well, TBH I didn't expect "prove that CVE is CVE". https://bugs.maemo.org/show_bug.cgi?id=12558 Jul 31 08:15:13 04Bug 12558: kernel/bluetooth: CVE-2010-1084: potential bad memory access with sysfs files Jul 31 08:15:42 DocScrutinizer05: ^^^ Jul 31 08:16:24 otherwise I perfectly know who the negatron man is :P Jul 31 08:16:36 I was commenting on the pselect() stuff Jul 31 08:16:54 DocScrutinizer05: I think zeq already demonstrate it Jul 31 08:17:02 zeq1: ^^^ ? Jul 31 08:17:53 freemangordon: we only need *one* valid rationale for cssu-kernel ;-D Jul 31 08:18:00 to be entirely fair, I haven't run a stock kernel for years Jul 31 08:18:18 yay stock kernel upgrade! Jul 31 08:18:21 to be entirely fair, I haven't run a stock kernel for years Jul 31 08:18:29 yay stock kernel upgrade! Jul 31 08:18:30 that said this is a well known problem in glibc Jul 31 08:18:30 zeq1: well, now you have thumb you can't run it Jul 31 08:18:41 indeed Jul 31 08:18:49 DocScrutinizer05: you are on stock, aren't you? Jul 31 08:18:57 yep Jul 31 08:19:00 freemangordon: by the way, i vote for it to be called "kernel" Jul 31 08:19:46 DocScrutinizer05: could you run the binary that hangs with buggy libc then Jul 31 08:19:55 because it's the kernel Jul 31 08:19:55 libc/kernel Jul 31 08:19:56 and stuff Jul 31 08:19:59 it doesn't hang Jul 31 08:20:05 D: Jul 31 08:20:06 or rather, shouldn't Jul 31 08:20:26 DocScrutinizer: I got it to hang Jul 31 08:20:49 yeah, and that's exactly what? a proof for nuttin Jul 31 08:20:58 it seemed to hang more quickly when the system was busy Jul 31 08:21:15 DocScrutinizer05: what Jul 31 08:21:20 ? Jul 31 08:21:36 it is a proof that fremantle is affected by this Jul 31 08:22:03 sorry, no Jul 31 08:22:13 Or I am totally unable to parse your statements Jul 31 08:22:15 it's an unavoidable problem with trying to emulate pselect in userspace Jul 31 08:22:22 you probably didn't understand what the program is supposed to do (or I didn't) Jul 31 08:22:42 zeq1: ^^^ ? Jul 31 08:23:23 DocScrutinizer05: did you read the problem description? Jul 31 08:23:31 doc there are some good concise articles on line giving the rationale for the kernel support Jul 31 08:23:33 on the mailinglist Jul 31 08:23:34 a proper demonstartion of a bug would usually exit with a printf saying "look here: the bug happened" Jul 31 08:23:44 hehe Jul 31 08:24:20 DocScrutinizer05: we're not doing some poerpoint presentation to management here Jul 31 08:24:25 *powerpoint Jul 31 08:24:35 in that code as I recall it that's been `printf("sigchld!") Jul 31 08:24:55 freemangordon: sorry you're going polemic again Jul 31 08:25:21 you can't try to run random code, say "look it doesn't run. that's a proof for XYZ" Jul 31 08:25:58 DocScrutinizer: the problem is very well known, and was fixed 5 years ago Jul 31 08:26:06 AIUI the code would print some msg when the pselect() fails Jul 31 08:26:21 well, it seems you didn't get what the test program does. AIUI it printfs until it hangs Jul 31 08:26:29 it locking means it doesn't work at all, and thus not proves anything Jul 31 08:26:53 well, I give up Jul 31 08:27:38 if you're not able to demonstrate a semi-conclusive proof of the bug, then your endurance is pretty flawed Jul 31 08:28:03 whether or not the test is really good, why do we need to prove a known defect? Jul 31 08:28:20 if the expected behaviour is that the program eventually hangs when error occurs, then explain why that happens and all OK Jul 31 08:28:57 are you asking for a technical explaination for pselect()? Jul 31 08:29:14 zeq1: because I don't support forcing a new kernel on all CSSU users to fix a bug that's not even showing up in stock kernel, for whatever reason Jul 31 08:29:27 the rationale for its existance? Jul 31 08:29:35 no Jul 31 08:29:41 people get hangs Jul 31 08:29:42 I understand the implications Jul 31 08:29:47 DocScrutinizer05: and how you are sure it does not show? Jul 31 08:29:51 I never got hangs Jul 31 08:29:58 sometime impossible to disagnose Jul 31 08:30:01 diagnose Jul 31 08:30:09 DocScrutinizer05: since when WFM is a proof? Jul 31 08:30:35 since when "doesn't work so it's a bug in kernel" was a valid rationale? Jul 31 08:30:55 but it's not a bug in the kernel Jul 31 08:31:05 show me the passage next to the test code that says "will eventually hang" and all id fine Jul 31 08:31:09 is* Jul 31 08:31:13 DocScrutinizer05: iirc it was already explained it is a bug in libc, solved in kernel Jul 31 08:31:29 freemangordon: so what? Jul 31 08:31:36 it's a multiple bugs in userspace where developers have used pselect() where itg doesn't exkist Jul 31 08:31:39 exist Jul 31 08:31:44 so what? Jul 31 08:31:59 typing too fast on N900 kbd :) Jul 31 08:32:07 demonstrate the bug happens on stock kernel Jul 31 08:32:21 DocScrutinizer05: that is what i asked you for Jul 31 08:32:48 to run the binary on your device, as you are on the stock kernel Jul 31 08:33:12 or if you want to code up something better that's cool Jul 31 08:33:39 I just grabbed that because it was there on the bug report Jul 31 08:33:52 fine Jul 31 08:34:24 and you said it hangs, while I don't see any indication anywhere that's the expected behaviour Jul 31 08:34:48 the expected behaviour is it prints some error msg Jul 31 08:34:50 in the code? Jul 31 08:34:51 AIUI Jul 31 08:35:09 perhaps I killed it too quickly Jul 31 08:35:32 zeq1: could you please paste the link again? Jul 31 08:35:50 I had the hildon this program has stopped responding Jul 31 08:35:53 so FIRST understand the code, SECOND explain to me so I understand it too, THIRD demonstrate it shows the expected behaviour Jul 31 08:35:58 and I killed it Jul 31 08:36:16 zeq1: nevermin, found it Jul 31 08:36:21 ok Jul 31 08:36:26 zeq1, how do you irc on n900? Jul 31 08:36:32 thats not asking for much, huh? Jul 31 08:36:34 CHAT Jul 31 08:36:38 XCHAT Jul 31 08:36:39 zeq1, xchat? Jul 31 08:36:41 oh Jul 31 08:36:51 oops wong button Jul 31 08:37:01 LOL Jul 31 08:37:05 wrong* Jul 31 08:37:14 LaoLang_cool: xchat on n900 is good Jul 31 08:37:33 nokia sans semibold 14 black-on-white Jul 31 08:37:40 I'm plugged into my wall charger and I'm afraid to get close to the plug! Jul 31 08:37:41 kerio, I'll try, I've tried it but not good in my memory Jul 31 08:38:19 LaoLang_cool: it works good Jul 31 08:38:59 DocScrutinizer05: read comment #6 here https://bugs.launchpad.net/ubuntu/+source/linux/+bug/319729 Jul 31 08:38:59 dman, I even asked you to run BOTH versions of the code, zeq1 Jul 31 08:39:09 one version is with simulated bug Jul 31 08:39:12 AIUI Jul 31 08:39:58 so the original author understood how to demontrate a bug Jul 31 08:39:59 DocScrutinizer05: do you want me to compile it and upload, so you can use it to test on stock kernel? Jul 31 08:41:27 sorry, I feel you're a bit confrontative regarding my ideas how to check bugs, so I don't think it helps when I run that stuff, you won't accept my results anyway as long as they are not in line with what you hope for Jul 31 08:41:39 and I got no time now, have to run Jul 31 08:42:13 DocScrutinizer05: what? you think it won't hang? Jul 31 08:42:23 and of course my basic statement stands Jul 31 08:42:26 [2012-07-31 10:35:53] so FIRST understand the code, SECOND explain to me so I understand it too, THIRD demonstrate it shows the expected behaviour Jul 31 08:42:37 naah, wont happen Jul 31 08:42:45 the FIRST point is most important one in here Jul 31 08:43:16 :shrug: Jul 31 08:43:36 you're not able to demonstrate there's any bug in stock kernel then Jul 31 08:43:45 why should I bother Jul 31 08:44:07 I really start to think I am losing too much time arguing with you. With no result but "you do it, I don't give a shit, we won't agree anyway" Jul 31 08:44:41 * LaoLang_coo_ is chating in xchat on N900 ;p seems ok Jul 31 08:44:50 freemangordon: sorry when I don't follow your inconsistent rationale Jul 31 08:45:00 np Jul 31 08:45:33 the problem is you don't follow others rationales. Upstream developers for example. Jul 31 08:46:16 look, if you can't prove there's a bug in kernel - even to me who's not completely averse and not complete idiot - how will you convince anybody else you did something worth it? Jul 31 08:47:15 I asked you for a quote that says "program locking is expected result when bug happens" you ignored Jul 31 08:47:34 "Hi guise, I fixed a bug that nobody ever found, and probably never will. Actually, the bug I found is part of a bit of ancient code slated for removal and marked deprecated since 1993. Can I apply bugfix?" Jul 31 08:47:43 I told you 2look buddy there's a #ifdef in code that *simulates* the bug" you don't listen or care Jul 31 08:48:32 The other problem is that you refuse to run a single binary on your device, which has a well defined behavior when run on an affected system. Jul 31 08:48:35 freemangordon: you're simply stating "it farts and that's my proof, go and check for yourself!" Jul 31 08:48:39 won't happen Jul 31 08:48:44 DocScrutinizer05: I don't have a system with stock kernel Jul 31 08:48:57 too bad then Jul 31 08:49:00 sorry dude Jul 31 08:49:11 freemangordon: just find someone else Jul 31 08:49:17 DocScrutinizer05: what? what's wrong with you man, I was asking for help, dammit. Jul 31 08:49:38 I already explained to you why I don't feel like *I* could help, when you're not even willing to convince me you understood the test code Jul 31 08:50:15 DocScrutinizer05: you want me to convince you to run some 5k binary? Jul 31 08:50:22 whatever result I get, for you it will be the proof that you're right and I'm wrong Jul 31 08:50:39 DocScrutinizer05: where did you get that from? Jul 31 08:51:47 freemangordon: And why it became personal? Jul 31 08:51:53 oops Jul 31 08:52:02 DocScrutinizer05: And why it became personal? Jul 31 08:58:42 freemangordon: that's enough. Jul 31 08:59:11 teotwaki: elaborate please, I didn't get what is enough Jul 31 08:59:18 you pestering doc. Jul 31 08:59:42 aah, sure. Jul 31 09:00:54 [2012-07-31 10:49:38] I already explained to you why I don't feel like *I* could help, when you're not even willing to convince me you understood the test code Jul 31 09:00:55 [2012-07-31 10:50:15] DocScrutinizer05: you want me to convince you to run some 5k binary? Jul 31 09:01:02 11:43 I asked you for a quote that says "program locking is expected result when bug happens" you ignored Jul 31 09:01:02 11:36 DocScrutinizer05: read comment #6 here https://bugs.launchpad.net/ubuntu/+source/linux/+bug/319729 Jul 31 09:01:04 get some coffe, dude! then reread Jul 31 09:01:14 Let's try a different tack. From wikipedia "Event Loop": Thus an obvious way to handle signals is for signal handlers to set a global flag and have the event loop check for the flag immediately before and after the select() call; if it is set, handle the signal in the same manner as with events on file descriptors. Unfortunately, this gives rise to a race condition: if a signal arrives immediately between checking the flag and calling Jul 31 09:01:14 select(), then it will not be handled until select() returns for some other reason (for example, being interrupted by a frustrated user). Jul 31 09:01:51 The solution arrived at by POSIX is the pselect call, which is similar to select() but takes an additional sigmask parameter, which describes a signal mask. This allows an application to mask signals in the main task, then remove the mask for the duration of the select() call, such that signal handlers are only called while the application is I/O bound. However, implementations of pselect() have only recently become reliable; versions Jul 31 09:01:51 of Linux prior to 2.6.16 do not have a pselect() system call, forcing glibc to emulate it via a method prone to the very same race condition pselect() is intended to avoid. Jul 31 09:02:04 zeq1: what the fuck are you on about? Jul 31 09:02:21 teotwaki: bug description Jul 31 09:02:32 no, he's just quoting wikipedia. Jul 31 09:02:49 pertinent quote from wikipedia Jul 31 09:02:50 and you can't really start a quote with "Thus" Jul 31 09:02:53 exactly, not useful Jul 31 09:03:08 which indicates a conclusion, or the midway point of a thought process Jul 31 09:03:29 use fucking English, and stop behaving like morons Jul 31 09:03:42 What's the bug? Jul 31 09:03:55 seriously, you're complaining about my editing or take a quote from wikipedia? Jul 31 09:03:59 yeaqh, finally some *decent* chan moderation - I missed that Jul 31 09:04:03 https://bugs.launchpad.net/ubuntu/+source/linux/+bug/319729 Jul 31 09:04:13 teotwaki: ^^^ Jul 31 09:04:15 afk Jul 31 09:04:34 How many people does it impact? , everyone (unlikely)> Jul 31 09:04:53 I quoted the pertinent paragraphs. Jul 31 09:05:16 teotwaki: a race condition in libc? everyone Jul 31 09:05:19 How often does it impact them? Jul 31 09:05:32 every app that thinks it's using pselect() Jul 31 09:05:44 freemangordon: no, I mean actual figures: You answer the question "how many people could be impacted" Jul 31 09:05:54 zeq1: thanks Jul 31 09:06:02 finally getting somewhere Jul 31 09:06:25 I'll tell you this though: I'm a massive select() user myself, hardly ever came across pselect(). Jul 31 09:06:59 with signals? Jul 31 09:07:16 sure, if your event loop is properly built. Jul 31 09:07:30 teotwaki: according to the bugreport, it is udev who uses pselect Jul 31 09:07:40 Ok Jul 31 09:07:41 well, one of the users Jul 31 09:08:27 teotwaki: read the full wikipedia entry Jul 31 09:08:37 "Event Loop" Jul 31 09:09:17 zeq1: yeah, I know what a race condition is mate, no worries there. Jul 31 09:09:52 zeq1: freemangordon: >>> < Keybuk> on a machine with a proper pselect(), you can emulate the hang by compiling -DUSE_SELECT <<< Jul 31 09:09:53 I wasn't disputing that, merely that there is one Jul 31 09:09:54 so... Jul 31 09:10:11 zeq1: no, there *can* be one if you write shitty code. Jul 31 09:10:18 ETX from my side, since you refuse to cooperate in a reasonable waqy Jul 31 09:11:29 there is one if you assume the glibc implementaion works Jul 31 09:11:30 when I can quote, why can't you? Jul 31 09:11:44 zeq1: I can assure you than using select() offers no problem whatsoever if your event loop is properly constructed: I have production code here using select a million times over, handling signals from the system constantly (I'm talking nearly 1M interrupts/s under load), and we have no issue whatsoever. Jul 31 09:12:09 anywa Jul 31 09:12:10 y Jul 31 09:12:13 regardless of that Jul 31 09:12:29 I can understand why the bugfix would be beneficial for the system/users by and large. Jul 31 09:12:36 Now, what's the issue? Why do you need someone to test? Jul 31 09:12:58 [2012-07-31 10:47:14] I asked you for a quote that says "program locking is expected result when bug happens" you ignored [2012-07-31 10:47:43] I told you 2look buddy there's a #ifdef in code that *simulates* the bug" you don't listen or care Jul 31 09:13:37 teotwaki: we don't run stock kernel, and simulating the issue on non-stock makes test flawed AIUI Jul 31 09:13:50 teotwaki: there are programatic ways of implementing event loops without pselect(). but using a broken pselect() isn't one of them. Jul 31 09:14:11 (teotwaki) I can understand why the bugfix would be beneficial for the system/users by and large. Jul 31 09:14:50 Well, no, it doesn't. Jul 31 09:15:01 First, you test it on your system, you use the compile flags to reproduce the issue Jul 31 09:15:25 teotwaki: zeq already did that Jul 31 09:15:28 Once you've reproduced the issue, you apply your patch, and validate that within your test environment, the bug is fixed. Jul 31 09:15:49 I did that too Jul 31 09:16:10 Ok, so what's the bloody problem then? Jul 31 09:16:23 at least the pselect() syscall is working as expected. Jul 31 09:16:58 and the broken glibc implementation is gone Jul 31 09:17:03 Excellent. Jul 31 09:17:07 So what's the problem? Jul 31 09:17:22 Just submit the patch for review and approval in whatever processes the CSSU has. Jul 31 09:17:37 teotwaki: just a minute. Jul 31 09:17:45 teotwaki: there's no process for kernel patches in cssu i think Jul 31 09:17:48 and don't forget to submit the tests Jul 31 09:18:00 no, wait, it's a glibc problem, nvm Jul 31 09:18:16 DocScrutinizer05: to submit it to myself? Jul 31 09:18:45 freemangordon: a bug tracker is also for tracking bugs sometimes Jul 31 09:18:55 teotwaki: there are 2 people in CSSU related to kernel - me an Pali Jul 31 09:19:16 freemangordon: that's... not true, isn't it Jul 31 09:19:23 freemangordon: honestly, a " it hangs..." is a bit scarse for me, to even understand what that means or implicates Jul 31 09:19:27 there are 0 people in CSSU related to kernel Jul 31 09:19:48 because cssu doesn't deal with kernel, officially Jul 31 09:19:54 teotwaki: the problem is a political decision whether to have so called "community kernel" or not Jul 31 09:19:55 (yet?) Jul 31 09:20:22 (freemangordon) DocScrutinizer05: to submit it to myself? Jul 31 09:20:26 what does it fuckin' matter? Jul 31 09:20:42 I mean, whether Doc tests the crap, and says "GO or NOGO" Jul 31 09:20:51 guys, #maemo-ssu Jul 31 09:20:54 how can you know it's been properly tested? Jul 31 09:21:09 Either you use a good process, that has someone in the CSSU validate it Jul 31 09:21:17 And you submit the patch + testscases to them Jul 31 09:21:20 or you don't test it at all Jul 31 09:21:28 cuz from a truthiness point of view, it's all the same. Jul 31 09:21:40 freemangordon: I've put the libc on the server. Jul 31 09:21:46 zeq1: ok Jul 31 09:21:50 there's no more guarantee from having me, doc, or joeblogs test it. Jul 31 09:21:57 how the f*ck hard can it be to post some proper report to ML, like "this is the problem: dadadada.... \n this is the patch: c-code_c-code.... \n these is the test code and are the test results showing there's a problem (see line 56) and that we fix it (see line 124)" Jul 31 09:22:01 Submit it to the mailing lists, ask for feedback, give it 1-2 weeks time Jul 31 09:22:12 DocScrutinizer++ Jul 31 09:22:34 teotwaki: that is not how CSSU development works Jul 31 09:22:48 the proper test is just to test the new syscalls Jul 31 09:22:55 damn. Nokia failed the idea of merged contacts for unified communications miserably :-( and it is even worse in Harmattan and seems to actually have nothing with sync :-(( Jul 31 09:22:58 make sure they work to spec Jul 31 09:23:14 ArkanoiD_: works fine in fremantle, i think Jul 31 09:23:44 kerio, never worked for me, some contacts occasionally "exploded" all the way Jul 31 09:23:48 This is how I analyse and report bugs: http://lists.zeromq.org/pipermail/zeromq-dev/2012-July/017912.html http://lists.zeromq.org/pipermail/zeromq-dev/2012-July/017936.html Jul 31 09:24:01 ArkanoiD_: ouch Jul 31 09:24:03 i thought it is google sync to blame but actually it is something else Jul 31 09:24:05 Anyone on the list can double check what I said, so when I come with a patch/merge request, there'll be a trace. Jul 31 09:24:30 ooh, and "that is how the patch works: explain_explain, and that's why the test works: more_explain" Jul 31 09:24:34 because on n950 i disabled google sync completely and it did not get any better Jul 31 09:24:57 freemangordon: file a bug report, write a test case and a patch - then notice that someone filed a bug, wrote a test case and a patch, check them and push them Jul 31 09:25:08 s/them/the patch/ Jul 31 09:25:09 kerio meant: freemangordon: file a bug report, write a test case and a patch - then notice that someone filed a bug, wrote a test case and a patch, check the patch and push them Jul 31 09:25:18 hrmpf Jul 31 09:25:19 teotwaki: are you aware I am in CSSU team? Not kidding, asking seriouly. Jul 31 09:25:31 freemangordon: file a bug report, write a test case and a patch - then notice that someone filed a bug, wrote a test case and a patch, check them and push the patch Jul 31 09:25:38 freemangordon: yeah, I do, but fuck me, I couldn't tell by your attitude. Jul 31 09:25:43 freemangordon: having a trace is still good Jul 31 09:26:11 freemangordon: I'm guessing you didn't understand my rant of "go or nogo"? Jul 31 09:26:23 teotwaki: well, blame my english. Jul 31 09:26:53 freemangordon: what I meant was, you and pali or pati or whatever are basically the people who would approve/refuse this if it came to you from someone else, right? Jul 31 09:27:13 yep Jul 31 09:27:18 freemangordon: so, now you're trying to fix something, but asking someone else for approval of your work? Jul 31 09:27:30 So, either you shouldn't be approving other people's work in the first place Jul 31 09:27:42 or, you can simply submit it publicly, and approve it publicly. Jul 31 09:27:58 Which either validates the whole process, or invalidates the whole infrastructure. Jul 31 09:28:06 But I want doc on my side, as he is a part of CSSU for long time now, even not officialy Jul 31 09:28:24 Exactly, the keyword being "not officially", so it doesn't change anything to the whole political issue Jul 31 09:29:20 freemangordon: had you just sent an email to the mailing list explaining the issue, documenting your patch, and asking for review, I'm sure Doc would've been happy to test it Jul 31 09:29:35 indeed, we need a public statement why we need a new kernel in CSSU, with proper evaluation and rationale Jul 31 09:29:49 but going to him directly, and asking "yo bru, you've got a stock kernel right? Can you execute this binary for me?" Jul 31 09:29:52 that's just plain stupid Jul 31 09:30:06 And I'm not even talking about the security issue Jul 31 09:30:13 Just from a "due process" point of view Jul 31 09:30:40 freemangordon: do you get our position now? Jul 31 09:31:49 asking someone from CSSU team (even not officialy) to make a test for me is not stupid from my POV. Otherwise yes, I think I got it. Jul 31 09:31:52 a) I'd like to see at least a pastebin of "it hangs..." and b) we need to do this public, my "approval" in the end is worth nuttin Jul 31 09:32:36 and c) yes I think the rationale is basically correct, so that's my approval already Jul 31 09:32:47 teotwaki: the disagreement is technical over whether it's been proven that the glibc pselect() is racy. Doc has a quite reasonable objection to my testing since it was not conclusive proof. It did definitely show the pselect() syscall is working however. Jul 31 09:32:48 since you pointed me to comment #6 Jul 31 09:33:53 teotwaki: but in the end it seems it is not the correct channel to discuss CSSU issues, my apologies Jul 31 09:33:54 zeq1 & freemangordon: If you want to get anywhere, fire up a shared text editor, get on some form of VoIP, and draft an email that explains the issues, and offer some proposals Jul 31 09:34:01 On the other hand I'm not sure why I need to prove it when it's very well known defect, even making the quoted except fromw wikipedia. Jul 31 09:35:48 I'm now on #maemo-ssu if that a more appropriate channel. Jul 31 09:43:33 zeq: you for example need to prove that stock kernel not already has a similar patch (or other means to not be affected by the issue. E.G. by one of numerous ARM idiosyncrasies) Jul 31 12:05:13 ~Botsnack Jul 31 12:05:13 aw, gee, DocScrutinizer51 Jul 31 12:22:52 i installed a screen protector but i took too much time and now there's dust permanently on my screen :( Jul 31 12:23:43 it's only noticeable in direct light or when there are white areas on the screen, but it's still annoying Jul 31 12:23:53 and it was only one, so i can't try again either Jul 31 12:23:57 D: Jul 31 12:25:32 i could just not use it i suppose Jul 31 12:25:45 but it would piss off my father Jul 31 12:25:50 goddammit Jul 31 12:26:05 is there any way to clean the screen side of a screen protector? Jul 31 12:44:52 zeq: It is libgcc4.7.2 right? Jul 31 12:53:49 kerio piss of your father? Jul 31 12:54:23 why is that? I never used a screen protector. Sure, got some scratches because of that, but they're only noticeable under direct sun light Jul 31 12:54:57 RiD: because he doesn't approve of wasting money Jul 31 12:54:57 rightfully so Jul 31 12:55:06 but nevermind, cleaned it Jul 31 12:55:12 now it's perfect ^_____^ Jul 31 12:55:47 (for the record: clean with soap, put in alcohol, remove, shake off until only few drops remain, and put it on) Jul 31 13:02:50 Did you get your otterbox yet? Jul 31 13:03:25 What screen protector did you get? Jul 31 13:10:48 vi__: the otterbox should arrive tomorrow Jul 31 13:11:08 i got an atfolix protector Jul 31 13:13:33 DocScrutinizer05: ping Jul 31 13:19:59 who's the maemo.org repos maintainer? Jul 31 13:23:49 kerio: X-Fade why? Jul 31 13:42:04 zeq: ping Jul 31 13:43:34 hmm, backupmenu only handles a uSD when it's vfat on /dev/mmcblk1p1 apparently Jul 31 14:20:14 vi__: pong Jul 31 14:30:06 hey, you seen the forum? Jul 31 14:30:12 DocScrutinizer: ping Jul 31 14:30:16 doc|home: ping Jul 31 14:30:22 geeze, which one is real? Jul 31 14:30:32 zeq: You saw TMO? Jul 31 14:33:06 vi__: regarding what? Jul 31 14:33:57 kerio: we have raised enough monies to buy zeq an N900 for development purposes. Jul 31 14:33:59 :D Jul 31 14:34:01 sweet Jul 31 14:34:05 In less than an hour. Jul 31 14:34:06 damn, now i feel bad for not donating anything Jul 31 14:34:10 oh well Jul 31 14:34:21 kerio: and so you should. Jul 31 14:36:20 Pali: i keep loving you more and more ^_^ Jul 31 14:37:17 DocScrutinizer05 http://5z8.info/-OPEN-WEBCAM---START-RECORD--_q0i4ao_linked-in-of-sex Jul 31 14:42:35 vi__: fun fact: i even have a spare n900 i don't use because the usb port is broken Jul 31 14:48:37 I would have donated if I wasnt flat broke Jul 31 14:51:25 jonwil: thats ok jon. Your contributions of lines of code are worth way more than 20 euro. Jul 31 14:51:34 kerio: Then send it to me. Jul 31 14:51:39 vi__: hell naw Jul 31 14:51:51 fine, be like that. Jul 31 14:52:56 actually, I think my reverse engineering is by far the biggest contribution to the community Jul 31 14:55:39 vi__: great news Jul 31 14:57:46 vi__: which thread? Jul 31 15:11:16 don't you fucken hate it when you post a reply to a mailing list, then a follow-up the next day to provide a really thorough answer, just to get a reply "Hey, cool." Jul 31 15:24:29 teotwaki: hey cool Jul 31 15:24:54 I'm so tempted to ban you right now. Jul 31 15:26:41 scnr :) Jul 31 15:35:43 damn I have been spending too much money lately :( Jul 31 15:35:50 but its on things I need so I have to spend it Jul 31 15:37:07 freemangordon: [call to arms!] Jul 31 15:37:26 vi__: thanks Jul 31 15:37:31 freemangordon: if all the pledges come through, we have enough to buy one of a guy on the forum. Jul 31 15:37:40 He lives in the same city as me. Jul 31 15:37:58 So I can go see it, pay with cash if it is good. Jul 31 15:38:14 Sodler down the USB and ship that mutha off to zeqland. Jul 31 15:38:21 s/sodle/solder/ Jul 31 15:38:27 s/sodler/solder/ Jul 31 15:39:59 grea Jul 31 15:40:04 great even Jul 31 15:42:06 I just need DocScrutinizer to NOT buy one for 150 euros like he siad he might. Jul 31 16:54:45 ok Jul 31 16:55:52 then I'll not have to cry when I have to send out the baby, I can keep it after I find any :-) Jul 31 16:57:13 anybody educate vi that IRC is no drive_by medium? Jul 31 16:58:39 ~botsnack Jul 31 16:58:40 :), DocScrutinizer51 Jul 31 16:58:58 at least ONE is listening and online Jul 31 16:59:41 ~seen timeless Jul 31 16:59:48 timeless is currently on #maemo, last said: 'could someone please tell me what "inedi.ntcn" is in that file? :)'. Jul 31 17:00:19 timeless: still with RIM? Jul 31 17:09:42 Estel_, I read your constant whine about infobot, but you conveniently left out how YOU insulted community memebers by setting a negative factoid. Ironic in a way since your first spat here on IRC was over a negative factoid set by someone else. Yet another misrepresentation of Council... That said, your exit from Council has patched a lot of this, so I'm done talking about it. Just saddens me that I have to continue cleaning up you Jul 31 17:09:42 r messes even after you ran for the hills. Jul 31 17:14:16 DocScrutinizer05: ping Jul 31 17:19:49 brb, sprints. Jul 31 17:30:09 vi__: pong Jul 31 17:32:20 Woody14619: c'mon :-D meanwhile *everybody* has made his notion Jul 31 17:32:55 ?? :) Made his notion? Don't follow you. Jul 31 17:33:02 Woody14619: or is there any tmo rant ongoing which I haven't noticed? Jul 31 17:33:42 sorry for my french. I meant everybody knows estel and how to read his posts and rants Jul 31 17:33:50 ::: not at all Jul 31 17:34:36 No, I'm saying I find irony that he was all up at arms over the ~tmo factoing here being "insulting to the community", but then creates his own insulting factoid and get pissy when someone locks it to prevent him from changing it after he regrets it. Jul 31 17:34:51 I just improved my ignore/spamfilter. Got a better more finegrained dynamic version in wetware Jul 31 17:36:10 Lol... I tend to make my ignore filters at my "caring" level. Ignoring them electronicly just means you don't see what they're up to, not that they're not up to anything. Jul 31 17:36:10 Woody14619: after all that's been exactly my plan Jul 31 17:36:57 Woody14619: :nod: Jul 31 17:36:59 What's the old saying? "Keep your friends close and your enemies closer?" Better to know what they're up to than to ignore them. Jul 31 17:37:46 Estel_: my enemy? ohwell, too much honor I'd spend by that Jul 31 17:39:02 no, not my enemy, though I can't say if the inverse it true. Reguardless, someone I'd rather track than be supprised by later. ;) Jul 31 17:41:17 sure thing - double true when you got shanghai'd for chanop Jul 31 17:41:21 if you have "enemies" in the same open source community, you've got other things coming than "whatever they're planning". Jul 31 17:42:00 agreed teotwaki__, but that doesnt' stop some people from making them... Jul 31 17:42:20 Woody14619: what you didn't know... is that secretly... Jul 31 17:42:21 * Woody14619 nods to Doc Jul 31 17:42:22 I AM Jul 31 17:42:29 Mwhuahahahaha Jul 31 17:42:42 Nor do I care. :) Jul 31 17:42:48 lol ;) Jul 31 17:42:58 omgpupnik Jul 31 17:43:08 haha Jul 31 17:43:09 pupnik Jul 31 17:43:13 where's that crazy fuck gone? Jul 31 17:43:33 either died or got other interests now Jul 31 17:43:43 I defo hope the latter Jul 31 17:43:59 me too Jul 31 17:44:06 talking about bygones, I wonder how qwerty's doing Jul 31 17:44:12 he wasn't that bad Jul 31 17:44:22 I should shoot him another email one of these days, see what he's up to. Jul 31 17:44:39 :nod: Jul 31 17:44:53 greetings from me Jul 31 17:44:58 will do Jul 31 17:45:06 by the way Jul 31 17:45:11 can you imagine that two years ago Jul 31 17:45:23 there were some 200 more people in here? Jul 31 17:45:36 I *know* Jul 31 17:45:53 Heck, 3 years ago, 3 months from now, there were 400 more people in here. Jul 31 17:45:57 even 250 more at times Jul 31 17:46:09 (amsterdam summit, n900 distribution) Jul 31 17:46:09 duh Jul 31 17:46:30 that's been slightly before my time Jul 31 17:46:42 you weren't here back then? Jul 31 17:47:09 but we're stable at 272 +- a few Jul 31 17:47:14 fuck me Jul 31 17:47:22 (back then) nope Jul 31 17:47:22 I've been idling in this channel for nearly 6 years Jul 31 17:48:02 DocScrutinizer51: seriously? I always thought you were here before me. Jul 31 17:48:03 Oh well Jul 31 17:48:15 I lived rather autistic wrt maemo community with my N810 Jul 31 17:48:32 I need to run off quickly Jul 31 17:48:41 won't be back tonight though, people coming over for dinner Jul 31 17:48:53 laters Jul 31 17:48:55 cya Jul 31 17:49:07 Oh... and since even a stopped clock is right twice a day... Jul 31 17:49:09 enjoy Jul 31 17:49:12 ~h-e-n Jul 31 17:49:13 hmm... hen is hostmode-easy-now, or http://talk.maemo.org/showthread.php?t=65232 Jul 31 17:49:24 Oh good.. that got fixed. :) Jul 31 17:49:39 Must have missed that in the scroll back. :) Jul 31 17:49:44 woops Jul 31 17:49:50 sorry, just wanted to post this real quick Jul 31 17:49:54 Woody14619: hehe Jul 31 17:49:54 http://firestarter.wedrop.it Jul 31 17:50:07 It's mah project :) Jul 31 17:50:15 ooooh Jul 31 17:50:23 Just a placeholder, but it's served by the actual project already :) Jul 31 17:51:38 that page is rendered by two lines of code :) Jul 31 17:52:10 auto & vl = this.layout(); Jul 31 17:52:19 mhm Jul 31 17:52:30 vl.addChild().size(400).type("F"); Jul 31 17:53:03 vl.addChild().text("firestarter"); Jul 31 17:53:54 (not quite true, there's a bit more manual styling, but you get the idea) Jul 31 17:54:16 anyway Jul 31 17:54:18 'latah Jul 31 18:02:26 Woody14619: hey! Jul 31 18:02:46 DocScrutinizer51: we found an n900. I hope you did not acquire one yet... Jul 31 18:02:58 Was it hiding under you bed? :) Jul 31 18:03:15 no! Jul 31 18:04:37 within an hour of starting the donations drive we had 110 euros in pledges to buy an n900 for zeq. A guy from my city (gregoranderson on tmo) has one for sale. Jul 31 18:05:06 TMO has some really cool people on it. Jul 31 18:05:36 like me Jul 31 18:06:10 you are a cool person? Jul 31 18:06:30 24x7 Jul 31 18:06:52 M TO THE MUTHA FUKIN FO MAH NIZZLE. Jul 31 18:07:17 Woody14619: also, what is this offensive factoid? Jul 31 18:08:09 Doc took care of them already... They were the ones around h-e-n that Estel made a while back apparently... Jul 31 18:08:56 Hello ! Jul 31 18:09:07 ~h-e-n Jul 31 18:09:08 from memory, hen is hostmode-easy-now, or http://talk.maemo.org/showthread.php?t=65232 Jul 31 18:09:15 ~hen Jul 31 18:09:16 [hen] hostmode-easy-now, or http://talk.maemo.org/showthread.php?t=65232 Jul 31 18:09:21 lame Jul 31 18:09:30 ~tmo! Jul 31 18:09:31 extra, extra, read all about it, tmo! is trolls, morons and ogres. Jul 31 18:09:31 Better than it was two days ago. Jul 31 18:09:37 lols Jul 31 18:10:13 I'm wondering: when in a hildon-desktop plugin, where do all those "g_printf" I'm trying to output ? I don't see them even in the console in my scratchbox... Are they supposed to show up there ? Is there a log file ? Jul 31 18:11:26 ~hildon-desktop Jul 31 18:11:32 (worth trying...) Jul 31 18:17:43 vi__: [2012-07-31 18:55:52] then I'll not have to cry when I have to send out the baby, I can keep it after I find any :-) Jul 31 18:21:08 Tofe: I'd guess the go to stdout of H-D Jul 31 18:21:18 whatever that might be Jul 31 18:21:31 probably same as stdout of dsme Jul 31 18:21:49 and that one probably has no stdout Jul 31 18:21:52 well, I do see other outputs (trackerd, for instance), but not mine Jul 31 18:22:16 maybe my compilation options are to be blamed, I quite don't know Jul 31 18:22:59 DocScrutinizer05: wtf does that mean? Jul 31 18:37:53 * vi__ tsssshhk Jul 31 18:48:48 vi__: does what mean? Jul 31 18:49:18 my answer to your ping, given when you weren't in? Jul 31 18:51:04 oohfsck! 20:50 3652 22 22 -192 249 249 249 65535 78 Jul 31 19:24:11 78 minutes? :) Jul 31 19:29:56 Woody14619, honestly, instead of Your constant conspiracy theories and seeking enemies, I would like to hear more about Council's work on things that seems stalled for 3 weeks already, you're chair, after all Jul 31 19:30:04 Other things are in fact too funny to even comment on them Jul 31 19:30:48 freemangordon, can I nominate You for Council?:P you got awesome patioence. I've went through Your discussion with DocScrutinizer, and honestly, I'm amazed by ammount of time You spenmd on persuading people that are like south park's "We won't listen" Jul 31 19:31:18 It seems that You have much bigger "soft skills" (tm) than You think You have Jul 31 19:31:25 Estel_: chill the fuck out. Have you manaed ED on n950 yet? Jul 31 19:32:02 I, for one, would already stop discussing with someone who isn't CSSu mmaintainer, but just lieks to sound like one ;) That';s why I'm not in Council, probably. Jul 31 19:32:10 vi__, sure, qole managed it ages ago Jul 31 19:32:24 I'm fighting one buggy blocker thing that makes Ed unusable on N950 Jul 31 19:32:29 bug bigger than anything else Jul 31 19:32:34 biggest design flaw in N950/N9 Jul 31 19:32:34 what? Jul 31 19:32:38 capacitive screen ;P Jul 31 19:32:39 aegis. Jul 31 19:32:52 noope, aegis is noob compared to screen that can't detec anything smaller than 10mm in diameter Jul 31 19:33:17 cap stylus?t Jul 31 19:33:20 I'm constructing capacitive stylus with 1mm precision, despite need for10mm touch surface., blah! Jul 31 19:33:28 normal cap styluses are worse than clicking with fingers Jul 31 19:33:35 precise ones cost 1/3 price of N9 Jul 31 19:33:46 I'm doing one for 1$ ;) Jul 31 19:33:56 how? Jul 31 19:34:04 actually for free, using replacement stylus for N900 and anti-ESD foil from Rpi package Jul 31 19:34:23 ? Jul 31 19:34:39 cutting a irregular shape out of foil, that makes detected touch zone *out* of place covered by "stylus" Jul 31 19:34:44 something like "virtual crosshair" Jul 31 19:34:50 You know where it will click Jul 31 19:34:53 but You see this point Jul 31 19:35:04 = You can draw line and expect where it will end ;p Jul 31 19:35:31 will post tutorial after finishing, i must just think about way to make this foil less elastic Jul 31 19:35:34 glue something to it Jul 31 19:35:35 oh for a multitouch resistive screen. Jul 31 19:35:39 yea Jul 31 19:35:48 even resistive witohut multitouch, I have m-touch in ass Jul 31 19:35:52 sur,e it helps in games Jul 31 19:36:05 but honestly, I take resistive screen with pressure sensitivity, over it, any day Jul 31 19:36:07 captouch is F'ed if you want to really do stuff Jul 31 19:36:13 yes Jul 31 19:36:33 but, you know, from my experiments, it is possible to make stylus which will, at least, make possibole to hit small x buttoin in easy debian reliably Jul 31 19:36:37 otherwise it's kinda unusable Jul 31 19:36:59 until you xbindkeys it via keyboard shortcuts to death (and still, you can't expect to prepare every possible shortcut that may be needed) Jul 31 19:37:03 so you have installed ED? Jul 31 19:37:20 as per qole's instructions. BTW, qwazix also did it Jul 31 19:37:28 there is threa don TMO, somewhere about ed in harm Jul 31 19:37:54 still, I want to use wheezy there ;P Jul 31 19:38:17 actually, I'm focused in upgrtading N90 0version to wheezy, then, i'll use same image on N9 Jul 31 19:38:26 n oidea about pulseaudio problems, though, and such things Jul 31 19:39:52 vi__, have You read backscroll as requested by freemangordon, about community donated n900 for zeq? Jul 31 19:40:41 given how cash strapped N is... wonder if we can get enough donations to get N to make one more N900 run Jul 31 19:40:54 Estel_: what about it? Jul 31 19:40:56 no. Jul 31 19:41:06 vi__, basically, zeq needs development device Jul 31 19:41:27 his first impresiov was - which I'm proud to hear, and little ashamed :) about buying device from my "n900" service ;) Jul 31 19:41:39 as a way of donation, I would sell it under costs Jul 31 19:41:44 vbut, freemangordon got even better idea Jul 31 19:41:48 Estel_: I have already started the fundraiser. Jul 31 19:41:52 he think if You could mimic what was done with DavyP Jul 31 19:41:56 ah Jul 31 19:41:57 nice4 Jul 31 19:42:01 where? Jul 31 19:42:12 Estel_: we received 110 euros of pledges within the hour. Jul 31 19:42:25 awesome and nice to hear :) Jul 31 19:42:32 where is thread/annoucement about it? Jul 31 19:42:36 Estel_: I found a guy in my city who wants to sell an n900. Jul 31 19:42:56 great, I was about to aks if You're able to find bad condition supercheap N900 Jul 31 19:43:12 I just have to wait for the pledges to be fulfilled, I buy it, solder down usb. shippit to zeq. Jul 31 19:43:25 nice :) how expensice the device is? Jul 31 19:43:29 expensive* Jul 31 19:43:35 100 euro. Jul 31 19:43:43 second hand. Jul 31 19:43:51 great. Haven't You said, that You have 110 already? Jul 31 19:44:33 it's much cheaper that I would be able to buy Jul 31 19:44:43 so i'm glad You did it Jul 31 19:44:54 Estel_: we have 110 promised, just have to wait fot them to pay up. Jul 31 19:44:58 (ok as in "thumb up") Jul 31 19:45:15 nice. but, again, where the f is donation thread? Or it's closed already? Jul 31 19:45:20 Estel_: just a case of the right place at the right time. Jul 31 19:45:32 Estel_: buried! Jul 31 19:46:09 :) Jul 31 19:46:15 Estel_: it was called [call to arms!] Jul 31 19:46:43 * Estel_ nods Jul 31 19:47:29 You know that you've basically organized what Council would be supposed to organize? Yet, you didn't wanted to candidate for Council, when I've asked You to :) Jul 31 19:48:51 it's quite a pity, that people who have skills/moral/merit to become councilors doesn't want to (you, freemangordon, would find some other names to...) and instead, people that *want* become councilors, end up candidating and being elected Jul 31 19:49:03 paradox of human nature, eh? Jul 31 19:49:24 :\ Jul 31 19:50:30 Estel_: you need to curb talking that shit estel. it just pisses people off. Jul 31 19:50:59 vi, who's saying that? i always though you're the one who says things when needed to be said, not carring much about "like/dislike" Jul 31 19:51:13 anyway, I think that last thought is quite calm and constructive Jul 31 19:51:23 its not, its called trolling Jul 31 19:51:59 you know what we ended with, and it could be even worse (arie also candidated :P ). We have freemangordon, You, Pali, and other people that would be absolutely perfect candidates for Council re skills, bot soft and hard, both coding and social :) Jul 31 19:52:03 * Sicelo agrees with vi, beford. Jul 31 19:52:12 beford, what isn't called trolling these days, eh? Jul 31 19:52:27 sharing thoughts in civil way? *shrugs* Jul 31 19:52:53 you know that irc have nice little button with "X" label, if You preffer places, where people think only like you think? Jul 31 19:53:35 so now, please GTFO, when i'mtrying to convince someone to candidate for Council during next election, due to meritocratic things (like organising donation of devices for devs) :) beford, that's for You :* Jul 31 19:53:50 wtf? Jul 31 19:53:57 it's also where You can kiss me, when You want to call something trolling, just becouse it's trending word. Fine? Jul 31 19:54:07 thanks for You time. Jul 31 19:54:22 shutup women, get on my horse Jul 31 20:00:31 Sicelo, after this moment of understandable silence, You still agree with beford? Or are You going to find less childish "twin souls"? ;) Jul 31 20:00:49 stop trolling Estel_ Jul 31 20:01:15 you attack me and espect me to be quiet? Jul 31 20:01:27 it's up to you, but I would feel a little ashamed, due to fact of agreeing with such msierable creature Jul 31 20:01:34 you ask me to GTFO, and ask me to quit the channel Jul 31 20:01:46 yea, right. Jul 31 20:01:59 beford, Your words are your choose, If all you have to tell us is some childish, 12-years old like (at best) sexist... I don't know, "gag"? Jul 31 20:02:42 I udnerstand that immature doesnt-mean-anything lines of sexist comparisions are common in Your natural enviromnent ;) but we're on #maemo, ya know. Jul 31 20:02:52 surprise muthefucka. this is the internet. Jul 31 20:03:10 well, internet got exact same ratio of miserable morons than real life, I suppose ;) Jul 31 20:03:36 just stop attacking me. stfu. please. Jul 31 20:03:51 I've seen manyu different people here - some liking to argue jsut for the sake of it, some thinking that they're fighting for better good,. but You first one to land on my IRc ignore liost, due to being completely miserable creature, not worth my time. Congratulations. Jul 31 20:04:16 -_- Jul 31 20:04:20 and take Your time, aftyer all, You need to have at least one place where You can talk about Your view on sex without being laughed off, yep? Jul 31 20:04:27 we're not *so* cruel. Jul 31 20:09:49 Estel_: have you ever tripped on hallucinogenic mushrooms? Jul 31 20:11:29 vi__, mushrooms are for kids, Datura is a way to go! Jul 31 20:12:04 But after Datura I wouldn't see you as candidate for councilor - you would be candidate for Pope in my eyes! Jul 31 20:12:52 Estel_: I know what datura is. I have grown it, I have consumed it. The original question still stands. Jul 31 20:13:08 consumed? I've actually added it to tea (no joke) Jul 31 20:13:08 Hello Jul 31 20:13:22 muchsroms, hm, let me think Jul 31 20:13:26 I think no. Jul 31 20:13:42 hi, anyone here managed to run power kernel along nemo Jul 31 20:13:53 have You tried Juares? It's nice addition for Sunday chicken date Jul 31 20:14:04 sobczyk, yea, many people, see TMO Jul 31 20:14:09 You should try them. They will let you see the world in a whole new light! Jul 31 20:14:12 s/sea/search on/ Jul 31 20:14:33 pink light spectrum instead of white? Jul 31 20:14:55 better answer if You can be convinced to candidate for Council, as this questions stays, too Jul 31 20:15:48 Just a quick question. Can anyone use routing in maemo-mapper or CloudGPS? Jul 31 20:16:32 cloudgps has a weird bug for me, the "search" text box displays text rotated by π/2 Jul 31 20:17:44 mrover: AFAIK maemo-mapper is abandoned. Jul 31 20:17:51 gnuite.com seems to be down. Just wanted to know if anybody is experiencing same problem Jul 31 20:18:15 mrover, try modrana or Marble Jul 31 20:18:20 Modrana seems like a way to go for You Jul 31 20:18:38 kerio: dwarz said it would be fixed in the next version. That was a while ago. Jul 31 20:18:50 I think he has moved on :/ Jul 31 20:18:52 vi__, question about You as candidate for Council still stand Jul 31 20:19:14 modrana is the only activly maintained map application left. Jul 31 20:19:23 Well, thank you for answers. Cheers! Jul 31 20:19:29 what about maep? Jul 31 20:19:45 Estel_: No chance. I neither have the time nor the patience. Jul 31 20:20:00 Besides, I rather like the current councillers. Jul 31 20:20:08 Understood, but it's a pity, as being 8no* councilor You do councils tasks many times Jul 31 20:20:14 like explaining things, maintaining communication, or donating Jul 31 20:20:26 so You proven already, that you have both time and patience :P Jul 31 20:21:06 [rant]It's quite tragic, that people who *want8 to become councilors candidate and get there, instead of people that *should* become. It never leads to good things[/rant] Jul 31 21:09:28 Estel_: it doesn't prove anything, as most people here donated, helped others, communicated... Jul 31 21:09:49 donated != organized donations Jul 31 21:10:26 there are people who did that too Jul 31 21:10:28 organized donations = moved own ass to find thing required, organize gathering funds, pick it up using money,s end to recipent, and being respected enough to actually get donations Jul 31 21:10:39 Sure, and almost none of them candidate for council, sadly :( Jul 31 21:10:52 so? Jul 31 21:10:52 too busy for politics, probably. and that is problem - Council is, suprisingly, not about politics Jul 31 21:11:02 well, it's a pity for community, that's all Jul 31 21:11:10 man, you stop right there. Jul 31 21:11:30 when we end up with people "wanting" to become councilors, cause they're not occupied by actually contributing as organisers, it end up with politicians in Council Jul 31 21:11:34 You asked Jul 31 21:11:55 if You ask without need for getting answer, go away and don't use my time, I answered honestly and verbosely. Jul 31 21:12:00 but you are excatly type of person you rant about Jul 31 21:12:07 Can't respect thjat someone use own time to answer your question for obvious things? gtfo. Jul 31 21:12:19 lol Jul 31 21:12:20 so you should like mne, as i'm not in Council anymore due to my decision Jul 31 21:12:33 and i try to convince people that actually organsie things to candidate Jul 31 21:12:33 no i don't Jul 31 21:12:46 you just want to get back a piece of cake Jul 31 21:12:49 that's all Jul 31 21:12:55 just live with it Jul 31 21:13:04 fine, but it is entirely your problem. Have something valuable to say, or You just going to try to prove me, that organisers like freemangordon or vi_ ar enot good for council, and You are? Jul 31 21:13:28 i never sad i'm good for council and never asked for that Jul 31 21:13:35 it's not me or someone else Jul 31 21:13:38 honestly - I can be eprfectly honest, as i'm not p[olitician - kiss my ass. Your contributions ammount is clsoe to zero, but You like to accuse others about "Wanting a piece of cake". Jul 31 21:13:53 go, do something useful Yourself. Jul 31 21:14:15 Estel_: or maybe you don't know my contribution? Jul 31 21:14:38 not to mention that You're quite borderline - one day You're whinning here how lfie is unfair and people bad because Your personal preoblems with work, and other day You argue jsut for sake of arguing, when someone try to convince *others* to candidate for Council Jul 31 21:15:02 have problem with me convincing vi_? Sure, don't vote for him, if he agrees (which seems unlikely) Jul 31 21:15:18 Estel_: lol Jul 31 21:15:25 you ranting about how others are bad Jul 31 21:15:26 now You can go and *** yourself with stun baton, I'm off, no need to waste time with you Jul 31 21:15:34 and not only promoting vi Jul 31 21:15:48 i think he can decide for himself(c) your quote several times Jul 31 21:16:50 Estel_: i hope you didn't start to drink when you quit your "job". have a good time... and btw contributing is not if you council or not, or even was kicked out of... Jul 31 21:17:03 * DocScrutinizer05 offers +q for free Jul 31 21:17:14 wanna check how my app looks ZogG_laptop? :P Jul 31 21:17:43 i think the pal here need timel to chill out, and stop making revolutions Jul 31 21:17:53 befr0d: why not Jul 31 21:18:14 next one with an unfriendly word, notion, attitude, whatever, or the wrong md5 across his nick, earns first +q Jul 31 21:18:44 befr0d: but you need to make t-shirt "ZogG is da best contributor ever - he tested how my app looks like" :P Jul 31 21:18:52 md5 is insecure :p Jul 31 21:19:05 lol Jul 31 21:19:28 befr0d: so i wanna test Jul 31 21:19:32 honestly guys, one whole screen full of bitchingt Jul 31 21:19:48 and I haven't even scrolled back yet, to find more Jul 31 21:20:00 ZogG_laptop: hold on let me upload it Jul 31 21:20:05 DocScrutinizer: don't read from the begining before judging Jul 31 21:20:32 what is +q? Jul 31 21:20:42 vi__: it means quite Jul 31 21:20:45 +q == STFU Jul 31 21:20:57 or STFU in Doc's language Jul 31 21:21:11 =) Jul 31 21:24:34 http://freenode.net/using_the_network.shtml Jul 31 21:29:48 ZogG_laptop: I'm not the judge, I'm the ATF Jul 31 22:21:14 ~popcorn Jul 31 22:21:14 somebody said popcorn was http://www.youtube.com/watch?v=T0f2YL_MThI, which is epic. Jul 31 22:34:12 ~pornstar Jul 31 22:34:20 ~seen pornstar Jul 31 22:34:25 sfdsdgfsd: i haven't seen 'pornstar' Jul 31 22:37:22 I must say, that Pali's hostmode is awesome Jul 31 22:37:52 without any disregard for H-E-N and it's authors (and considering, that Pali's hostmode is derived from this code, so hats off), it made H-E-N virtually obsolete Jul 31 22:37:58 and it's close to make it rpactically obsolete Jul 31 22:38:18 autodetection of speed works like a charm, worked 7/70 attempts with different devices, for now Jul 31 22:38:26 and during different conditions, I tried to stress it Jul 31 22:38:35 even found some small bugs, like with HUBs Jul 31 22:38:54 (trivial to fix, need to extend time before switchign speed during guessing, and then it works OK) Jul 31 22:39:13 and to this proper integration with system, and we have obvious candidate for cssu-thumb Jul 31 22:40:10 BTw, IDK what Pali did, but high-speed became completely reliable Jul 31 22:40:26 no more bitching with enumerating it - I jsut fire it up with USB HD or pendrive, and forget Jul 31 22:40:41 after few seconds (well, sometimes up to 15, but that's prive for auto-guessing) it's enumerated Jul 31 22:41:29 maybe he use better time in script that I did manually, or twaked something during kernel integration - most important thing is that it works as it should, never expected it to happen so quickly Jul 31 22:45:35 Estel_: i had to increase the times a bit Jul 31 22:45:46 but yeah, it works great for me too Jul 31 22:47:11 what are Your time values? Jul 31 22:47:20 times between attempts to change speed? Jul 31 22:56:08 btw see #maemo and my question Jul 31 22:57:02 well, I'm starting upgrading ED to wheezy :) Jul 31 23:11:07 Estel_: yeah, i put them to 2 seconds between boost and enum, and 4 seconds before checking successful enumeration Jul 31 23:11:12 perhaps they could be tightened a bit, but meh Jul 31 23:11:33 between boost and enum is probably not needed, but before checking enumeration is good idea Jul 31 23:11:46 which line fo script it is in? i haven't traced it yet Jul 31 23:11:48 Estel_: i think it waits 2 seconds by default Jul 31 23:11:51 and no need, if You traced it already ;) Jul 31 23:12:07 search for usb_enum Jul 31 23:12:10 or "if usb_attached" Jul 31 23:12:18 there are two sleep calls Jul 31 23:12:21 thanks Jul 31 23:12:36 i put them as 2 and 4, not sure what the defaults are Jul 31 23:14:06 damn, Pali told me sysfs lcoation fro altering current used during charging, but it was on IRc, now need to search through logs... silly me should write it instantly Jul 31 23:16:46 ~pali Jul 31 23:16:46 somebody said pali was http://atrey.karlin.mff.cuni.cz/~pali/ Jul 31 23:17:14 /sys/class/power_supply/bqwhatever Jul 31 23:17:23 probably Jul 31 23:20:00 yea, found it Jul 31 23:20:02 w-wut? Jul 31 23:20:41 cat /sys/class/power_supply/bq24150-0/charge_current Jul 31 23:20:44 1750 Jul 31 23:20:48 dafuq? 0_o? Jul 31 23:21:05 da very, very fuck? Jul 31 23:21:10 oh wow Jul 31 23:21:12 neat Jul 31 23:21:18 your battery must be really good Jul 31 23:21:25 maybe it calculates it from my battery capacity Jul 31 23:21:27 well, it's not calibrated yet Jul 31 23:21:35 it's 3Ah, and now calibrated as 2,6Ah only Jul 31 23:21:47 still, charging chip will accept 1250 max Jul 31 23:21:55 and given such value jsut use default 950 i'm afraid Jul 31 23:23:00 cat /sys/class/power_supply/bq24150-0/current_limit Jul 31 23:23:01 1800 Jul 31 23:23:02 suuure Jul 31 23:24:04 Estel_: try putting it in the 500mA mode Jul 31 23:25:15 cat /sys/class/power_supply/bq24150-0/termination_current Jul 31 23:25:16 6650 Jul 31 23:25:26 wut? Are the divided by smth, or what? Jul 31 23:25:36 s/divided/to be divided/ Jul 31 23:25:37 Estel_ meant: wut? Are the to be divided by smth, or what? Jul 31 23:27:58 DocScrutinizer, any comment on those values? Do you have idea, how they should be "consumed"? Jul 31 23:28:12 here is source code that implement those entries Jul 31 23:28:13 https://garage.maemo.org/plugins/ggit/browse.php/?p=kernel-power;a=blob;f=kernel-power-2.6.28/debian/patches/bq2415x_charger.patch;hb=HEAD#l1319 Jul 31 23:55:22 it seems, that 1750 is 950 mAh Jul 31 23:55:43 300 more than 1750 is 2050, which is what sysfs want to accept as max, so it's 1250, which is also 300 more than 950 Jul 31 23:56:05 lwoest possible value is 1350, which is 400 less than 1750 (default value), jsut like 550 is 400 less than 950 Jul 31 23:56:12 and 450 is lowest possible value for chip as in mA Jul 31 23:56:20 I wonder, why it translates like that iun sysfs entries Jul 31 23:56:35 no idea about current_limit that is 1800 by default, and doesn't want to set itself higher Jul 31 23:56:46 or termiantion_current which is 6650, no idea how to translate it to real numbers Jul 31 23:56:55 maybe the same way as charge_current ;) Jul 31 23:57:37 Well, erm, check the bq27xx datasheet? Jul 31 23:57:55 Then, setup an i2C call snooper Aug 01 00:18:41 why i2c call snooper? Pali already included charging module in kernel Aug 01 00:18:51 it's just about editing syfs entries for it Aug 01 00:19:02 In that case, you've got the source Aug 01 00:19:08 BTW, i've checked charge.sh script used in backupmenu, and it's indeed outdated one, not the latest and greates shadowjk prepared Aug 01 00:19:18 sure: Aug 01 00:19:19 https://garage.maemo.org/plugins/ggit/browse.php/?p=kernel-power;a=blob;f=kernel-power-2.6.28/debian/patches/bq2415x_charger.patch;hb=HEAD#l1319 Aug 01 00:19:30 now, do you have idea how values are calculated re mA values? :) Aug 01 00:20:22 charge_current is for sur elike I described it, but termination_current doesn't want to play nicely Aug 01 00:20:30 it's 6550 by default, which seems to be 100 mA Aug 01 00:20:48 as lowest is 6500, which would be 50 mA, lowest accepted by chip Aug 01 00:21:03 but, if You change it, you can't change it back to 6550, it jsut doesn't maccept it (!) despite it was default value Aug 01 00:21:22 then, higher current accepted by chip is 400 mA, which would be 6850. But, it accept 6800 as max Aug 01 00:21:44 and generally, no matter what You set, it changes itself to *either* 6800 or 6500, doesn;'t accept anything between, even default value of 6550. I wonde,r if it's bug Aug 01 00:22:06 after all, default value of sysfs entry should be re-aplicable later, too? Aug 01 00:30:28 robbiethe1st, I've updated charge script in backupmenu Aug 01 00:30:51 also, I've included things needed for lzma comprfession in extrafiles and, where applicable in filelist.txt (list to copy) Aug 01 00:31:09 lzma compression is much more effective than gzip for things we have in rootfs and optfs Aug 01 00:31:25 and people who used compressed backup - like me - do it overnight, usually, so increased time shouldn't be a concern Aug 01 00:31:36 OK Aug 01 00:31:38 as a result, my rootfs.tar.lzma file from backupmenu is ~60 MB (!) Aug 01 00:32:07 wanna me to push update to extras-devel? Aug 01 00:32:21 Out of curiosity, can we use lzma compression in ubifs? Aug 01 00:32:44 erm? how so? we're lzma'ing tar with rootfs content Aug 01 00:32:53 jsut like we're gzip'ing tar with rootfs content, before Aug 01 00:33:00 Not BM wisr Aug 01 00:33:04 ah Aug 01 00:33:10 I mean, in the FS itself? Aug 01 00:33:13 no idea, honestly Aug 01 00:33:20 doesn't see why not, though Aug 01 00:33:27 you can put there lzma file as any other Aug 01 00:33:34 but, we have compression already there, transparent one Aug 01 00:33:51 IDK if putting lzma compressed file wouldn't result in file becoming *bigger* after ubifs compression on it Aug 01 00:33:51 I mean to use LZMA as that transparent compression, instead of gzip or w/e Aug 01 00:33:55 no. Aug 01 00:33:56 we can't Aug 01 00:34:01 and it would be GODDAMN slow Aug 01 00:34:14 For writes, reads or both? Aug 01 00:34:21 we're using gzip as ubifs transparent compression? I think that it's not true Aug 01 00:34:23 BM wise, send me a copy of the tarball please Aug 01 00:34:43 I recall that there's two types of transparent compression in ubifs Aug 01 00:35:01 robbiethe1st, sure, another thing - in my modification, compression is done via lzma only, no option to gz it (I jsut replaced what was needed), You may want to allow 2 types of comnpression Aug 01 00:35:05 One which is used for new files, and one which was used on /some/ of the files in the rootfs image Aug 01 00:35:14 still, I think that if someone compress it, he want msot compression, and don't care about time Aug 01 00:35:25 Probably, yeah Aug 01 00:35:34 robbiethe1st, unfortunately, I have no idea as for ubifs compression :( Aug 01 00:35:52 I wonder where it is documented, for our case of N900 rootfs Aug 01 00:36:28 http://en.wikipedia.org/wiki/UBIFS - zlib(deflate) or LZO Aug 01 00:38:23 but no idea about our possible duality of compression types? Aug 01 00:38:45 BTW, i'm all for - when we will have everything possible thumb2-compiled - optifying rest if needed, and disabling compression at all Aug 01 00:38:52 should give some ncie speed boost to out lovely N900 Aug 01 00:38:57 s/ncie/nice/ Aug 01 00:38:58 Estel_ meant: should give some nice speed boost to out lovely N900 Aug 01 00:39:20 even now I have 70 MB free rootfs, with everything thumb2'ed, it will be more - then some optification, and we should fit rootfs there Aug 01 00:39:36 then, it would be really good to put things that need to be executed quickly on rootfs Aug 01 00:39:56 as now practical action is to move things that we need to execute very fast OUT of rootfs Aug 01 00:40:02 like qt libs Aug 01 00:40:31 freemangordon tried with putting them on rootfs, due to increased space thansk to thumb2, and measurement clearly highlighted when and how much compression hurts performance Aug 01 00:40:48 ..and that as result, libs are executing faster from optfs Aug 01 00:41:18 I wonder; could you raid together a portion of the eMMC and your SD card(raid 0, obviously) for better performance? Aug 01 00:41:31 naaah, those media have different speeds Aug 01 00:41:35 and raid on flash is pathetic Aug 01 00:41:45 there was even some funny article about raid on usb sticks ;) Aug 01 00:42:00 theoretically it should be double speed of slowest medium, i.e. emmc Aug 01 00:42:05 Yeah Aug 01 00:42:06 practical benefit is close to 0 Aug 01 00:42:21 swap on uncompressed rootfs would be very fast, but, it's wear&tear Aug 01 00:42:46 I preffer to sue microSd for swap. It's pity, that they don't produce oens with superfast random write speed :) Aug 01 00:43:05 or ones with 2 partitions, one being RAM :P Aug 01 00:43:09 Why not look into more practical handling of memory? use zswap or w/e, possibly optimize the swap algorithms to only write in blocks etc. Aug 01 00:43:13 2 psychical ones, of course Aug 01 00:43:26 Also, possibly change memory page size from what... 1MB down to 4K? Aug 01 00:43:28 compache is sued already, and still sucks, unfortunately Aug 01 00:43:47 Then the page size that gets swapped out is smaller? Aug 01 00:43:49 optimizations are done, i use optimized settings, freemangordon too, and it makes lfie easier, but Aug 01 00:44:01 rewriting swpa alghoritms to introduce proper garbage collectors is quite hard Aug 01 00:44:06 possible, but no trivial task Aug 01 00:44:16 ShadowJK have idea how it *should* look Aug 01 00:44:33 freemangordon, after discussion with him, agreed that without it, we don't have much possibilities for improving it Aug 01 00:44:52 BTW, for now, refreshing swap as in package ereswap is a way to go Aug 01 00:45:03 I sue 2Gb swap partition on microSd, and refreshing it isn't bad Aug 01 00:45:14 but, it never ever crerates something comparable to more RAM, obviously Aug 01 00:45:35 refreshjing swap, we already have benefits that we would have with rewriting swap alghoritms (it would just eliminate need for refreshing) Aug 01 00:45:50 as initially, swap is written in pure linear form, from beginning of partition to the end Aug 01 00:46:05 problems start after filling whole available space and re-using previously freed portions Aug 01 00:46:17 ereswap deals with that Aug 01 00:46:25 by disabling and enablign swap :P Aug 01 00:46:45 + introducing safety mechanisms, like backup swap to be able to boot device in case of microSd card missing/malfunctioning, etc Aug 01 00:46:46 I wonder... what if you simply got a small(256mb or so) ram chip-board attached to some sort of controller. Attach it to the back, connect to one of the IO pins internally Aug 01 00:46:54 Use it as a block device with custom driver. Aug 01 00:47:06 good luck with connecting it to internal pins, RAM is on OneNAND chip Aug 01 00:47:17 No, not to the ram. Aug 01 00:47:21 ah, IO chips./ Well, i think they're too slow Aug 01 00:47:26 no idea, though Aug 01 00:47:34 and "some kind of controller" is the tricky part, probably Aug 01 00:47:39 Likely Aug 01 00:47:50 I would be first one to use 1GB notebook-sized ram in mugen cover :) Aug 01 00:48:00 You might be able to do it over USB, though; at >20mb/sec random reads/writes Aug 01 00:48:25 less, our hostmode allow 4MB/s speeds only, even in high Aug 01 00:48:28 speed mode Aug 01 00:48:43 honestly, over wifi would be faster ;) Aug 01 00:49:05 but it's rather faster swap than ram Aug 01 00:49:25 haha swapping over nfs over wifi? Aug 01 00:49:33 ;P Aug 01 00:49:38 That'd be cool Aug 01 00:49:55 ?I suspect wifi ping would kill the idea Aug 01 00:50:00 probably, USB one too Aug 01 00:50:04 probably Aug 01 00:50:21 * nox- gets 30ms roundtrip times over wifi iirc Aug 01 00:50:26 Actually, though, the best idea is simply to reduce the memory & code footprint of your applications Aug 01 00:50:27 i.e. it would, with much effort, prove to be beneficial, but by 10 or 15%, and using megatons of battery power :P Aug 01 00:50:38 robbiethe1st, say it to fennec developers Aug 01 00:50:45 Yeah :\ Aug 01 00:50:51 but, we're doing it via thumb :) Aug 01 00:51:12 Some will not admit it, but cssu-thumb is already a fork of CSSU, due to awesome slowness on adapting good things on the latter Aug 01 00:51:12 Which I'm skeptical about, as that reduces CPU effeciency Aug 01 00:51:19 robbiethe1st, it doesn't, in fact Aug 01 00:51:26 theoretically, yes, rpactically, no Aug 01 00:51:35 and due they introduced better compiler with thumb... Aug 01 00:51:45 practical output is faster, and many placebo-prone people admit it Aug 01 00:51:47 including me Aug 01 00:51:59 i.e. with old compiler, there was no detectable slowdown, even by biased ones Aug 01 00:52:01 Couldn't we just use a better compiler with ARM, then? Aug 01 00:52:06 with new compiler, everyone You ask says it's faster Aug 01 00:52:25 probably, but why no with thumb, as speed loses are theoretical only, and memory advantages are huge Aug 01 00:52:56 Well, the most important thing then: Can we get (most of) MicroB recompiled? Aug 01 00:53:06 engine, for sure Aug 01 00:53:08 UI, no chances Aug 01 00:53:15 Mainly the rendering engine, which should the part that takes up the CPU Aug 01 00:53:21 freemangordon and zeq are fighting with fennec to make it usable, as first step Aug 01 00:53:32 allowing using it's gecko engine with microB Ui is another step Aug 01 00:53:41 rendering engine - absolutely yes Aug 01 00:53:55 Isn't that what MicroB already has? An old version of Gecko? Aug 01 00:54:09 idela solution would be to use fennec gui with tabs, extensions etc for everyday use, and microB with same engine, for situations when you need microB specific features, like cursor mode Aug 01 00:54:14 yes Aug 01 00:54:23 but putting new gecko isn't swap-in thing Aug 01 00:54:45 well, even with new engine, microB won't be comaptible with addons, sadly Aug 01 00:55:18 I'm happy that freemangordon and zeq took fighting with gecko and fennec, but OTOH sometimes I wonder if proper port of chromium wouldn't be more beneficial, in long run Aug 01 00:55:24 it have cursor mode out of box... Aug 01 00:55:25 We already have a couple of addons(like flash click-to-play) Aug 01 00:55:30 extensions compatibility Aug 01 00:55:34 and isn't as much of memory hog as fennec Aug 01 00:55:45 robbiethe1st, yea, but not most important ones :P and no tab browsing Aug 01 00:55:47 Couldn't we implement a wrapper for "proper" addons? Aug 01 00:55:55 nah, every addon need own re-write Aug 01 00:56:04 their work on engine level, but not in Ui one Aug 01 00:56:07 blame Nokia Aug 01 00:56:11 Tab browsing, though, really isn't a priority for me - Maemo's switcher is better than the tabs I've seen Aug 01 00:56:25 fennec tabsd are more convenient, but I agree that it isn't a priority Aug 01 00:56:36 it's jsut fuckin irritating that You can't open new tab in background, without switching to it Aug 01 00:56:40 in microB Aug 01 00:56:51 new tab = auto-switch, minimizing, going back to where You've been reading Aug 01 00:56:56 7x more time spent Aug 01 00:57:16 on chromium I just click a link wioth 3th mosue button modifier (shift + fn in my case) Aug 01 00:57:24 and it gently opens in background, as another tab Aug 01 00:57:44 chromium via easy debian - I suspect, that if it would get same treatment as fennec got from freemangordon and zeq, it would work quite fast Aug 01 00:57:46 Chromium port would be useful; I've got the 'desktop' chromium port which would be perfectly good Aug 01 00:57:57 with spell checking, things like lazarus form recovery etc installable Aug 01 00:58:20 well, i use chromium via easy debian, and it's OK - especially, when You have more RAM available, thanksw to thumb2 Aug 01 00:58:26 but proper port would be even better Aug 01 00:58:59 when we're speaking about browsers, I finally must test this unreleased, end-product quality flash 10 plugin, discovered not so long ago Aug 01 00:59:10 Ooh, link~! Aug 01 00:59:22 no idea, it was quite a "conspiracy" for some time Aug 01 00:59:32 it started with some random dude discovering Nokia's internal repo Aug 01 00:59:35 available for public Aug 01 00:59:40 idiot started to send links to many people Aug 01 00:59:47 wanting only flash 10 from it, as he found it there Aug 01 01:00:01 Pali, me and other guys decided to not announce it unless repos will be mirrored Aug 01 01:00:05 and domain searched for other repos Aug 01 01:00:12 AFAIk no additional ones were found Aug 01 01:00:36 it was mirrored weeks ago, though, and still no one announced it officialy, despite it's already spoken topic, on IRC Aug 01 01:00:44 (that's why i'm not afraid of talking about it) Aug 01 01:00:51 no chances of uploading it to cSSU, though Aug 01 01:00:57 stupid Nokians, it was finished in 2010 Aug 01 01:00:59 Ah. Just give me a link. Aug 01 01:01:01 never released, though Aug 01 01:01:14 Dun care if it's released, I want to mess with it. ;) Aug 01 01:01:17 Don't have it handy, ask freemangordon or Pali in normal hours Aug 01 01:01:24 i msut do same thing Aug 01 01:01:28 I have this link "somewhere" Aug 01 01:01:33 can't recall now, though Aug 01 01:01:37 k Aug 01 01:01:48 which will end in me asking for link too ;) Aug 01 01:05:08 how large a repo? Aug 01 01:10:47 SpeedEvil, it was quite big IIRC Aug 01 01:10:58 although, haven't heard of anything useful beyond flash 10 found there Aug 01 01:11:18 BTw I wonder if this "random guy" sending link haven't got it from some Nokian, that didn't wanted it to waste itself when they pull the plug Aug 01 01:11:24 who knows Aug 01 01:11:50 pester Pali and freemangordon for it, it's due time to announce something in public :) Aug 01 01:12:20 You can skip pestering Council - it's good if they have any idea about it, actually. when I was councilor I've sent a mail about it to Council mailing list, but no one ever replied Aug 01 01:13:17 interest was close to 0, if You don't count Woody14_something being angry that he haven't been considered "worthy" to receive link from said "one who found it" ;) Aug 01 01:13:33 nah, lets stop her,e until I'll say too much about Council internals :) Aug 01 01:13:57 It's udnerstandable that such file, as being not nokia copyrighted, won't be available for installation oficially, from repos Aug 01 01:14:10 although, wget script setting it up and downloading + torrent should be applicable Aug 01 01:14:21 I checked the repository file of this repo vs the normal repo and didn't find anything interesting except the Flash plugin Aug 01 01:14:27 at best, private repo, hosted by someone in country where sharing such thing in PRIVATE repo isn't forbidden Aug 01 01:14:38 jonwil, exactly, that was what I knew Aug 01 01:14:51 jonwil, btw, have You link handy? Aug 01 01:14:55 for flash? Aug 01 01:14:58 that I dont Aug 01 01:15:05 robbiethe1st, iI've finished code cleanup for backupmenu Aug 01 01:15:10 sending it Aug 01 01:15:13 I have no interest in Flash on the N900, never had a reason to use it Aug 01 01:15:29 although, you know me, it's not path, just updated version of scripts and extrafiles.tar.gz, everything bundled together :) Aug 01 01:15:30 For a while, it worked very well with Iplayer Aug 01 01:15:41 jonwil, same here, mostly Aug 01 01:15:43 Flash might work well to turn the N900 into a hand warmer app Aug 01 01:15:45 fullscreen video worked well Aug 01 01:16:43 I think the only other thing in this repo that was even remotely interesting was some extra feedservice plugins like feedservice-plugin-cnn and feedservice-plugin-reuters Aug 01 01:16:53 jonwil, although sometimes, I run animated films for my son, via things like youtube Aug 01 01:17:01 hm Aug 01 01:18:25 Ha ha ha ha http://www.theregister.co.uk/2012/08/01/nexus_q_preorders_halted/ Aug 01 01:18:42 * RST38h cackles Aug 01 01:19:46 You ordered one? Aug 01 01:21:05 nope Aug 01 01:21:06 why? Aug 01 01:21:10 free beer Aug 01 01:21:37 robbiethe1st: http://lorienart.pl/backupmenu_estel-patch.tar.lzma Aug 01 01:22:48 ty Aug 01 01:24:04 np, it works for me and changes are not dramatic - updated charger and .lzma compression instead of tar, as said (+ required libraries copied via filelist.txt and lzma added to extrafiles), but as always You may want to check it on own device Aug 01 01:24:21 What charger improvements? Aug 01 01:24:30 afaik it's based on backupmenu 1.1-1_all.deb Aug 01 01:24:38 no idea, as script was almsot totally re-written - m Aug 01 01:24:46 for sure betetr sense resistance of chip Aug 01 01:24:48 21 mOhm Aug 01 01:25:01 charging current adjusted Aug 01 01:25:07 other than that, ask ShadowJK Aug 01 01:25:45 robbiethe1st, BTW, maybe in future, for KP suers, You would want to use kernel charging module :) Aug 01 01:25:46 ok Aug 01 01:25:53 s/suers/users/ Aug 01 01:25:53 Estel_ meant: robbiethe1st, BTW, maybe in future, for KP users, You would want to use kernel charging module :) Aug 01 01:26:02 There's a kernel charging module? How does that work? Aug 01 01:26:25 well, see kp51r1 changelog. It's base don same logic as charge.sh script, but properly implemente dinto kernel Aug 01 01:26:31 we're getting bme replacement, it's first step Aug 01 01:26:33 huge step Aug 01 01:26:46 well, now it works like that - you stop bme and load module, then connect charger ;) Aug 01 01:26:50 it jsut works, charge Aug 01 01:27:04 current and other things controllable via sysfs, I was writing about it when you arrived Aug 01 01:27:12 it detects plugging/unplugging Aug 01 01:27:13 Ah Aug 01 01:27:15 works during hostmode Aug 01 01:27:23 i.e. hostmdoe + charging without script Aug 01 01:27:28 generally, eliminates need for scripts Aug 01 01:27:33 So all I have to do is load a specific kernel module and it does it all? sweet! Aug 01 01:27:39 next step - userland things like charging diode, low battery warnings, etc Aug 01 01:27:45 robbiethe1st, yes Aug 01 01:27:56 I suspect it will work in such early stage as backupmenu? Aug 01 01:28:02 it's called bq2415x_charger Aug 01 01:28:04 module Aug 01 01:28:34 side effect - for now, bright yellow diode like emergency charging is ON (at least in Maemo), but it charges OK Aug 01 01:28:40 Well, you could simply replace the 'charge.sh' line in bm.item with modprobe bq2415x_charger or w/e... Aug 01 01:28:54 also, remember that it's for kp users only - but I think that kp users are only one that You should care for, anyway :) Aug 01 01:29:02 robbiethe1st, yes Aug 01 01:29:22 I uploaded You version with charge script, that uses i2c, as backupmenu isn't dependent on kernel-power? Aug 01 01:29:23 So, just make a test if the .ko file exist, if it does use that, otherwise use the script Aug 01 01:29:32 also, there are older versions of KP Aug 01 01:29:34 good idea Aug 01 01:30:10 so, new version of backupmenu is soon to arrive?:) Aug 01 01:30:15 Hm... Do you know how to implement conditional package requirements? like KP v 49 OR i2c_tools etc... Aug 01 01:31:04 oh god, no idea, but such ocmbination seems impossible. It's rather depending on something like "kernel_feature_charging", and then, every variation that have it, specify "Provides: kernel_feature_charging" Aug 01 01:31:25 but nothing bad in depending on i2ctools Aug 01 01:31:32 You may want to wait for kp51r1 entering extras Aug 01 01:31:40 and then, drop support for non-kernel-power at all Aug 01 01:31:45 Is i2ctools in extras? Aug 01 01:32:01 sure, but no idea if in -testing- or -devel or -main Aug 01 01:32:04 but it is Aug 01 01:32:08 Nah, I won't do that. Mainly because BM is very useful as you can use flasher to load the stock kernel back again. Aug 01 01:32:14 my bnf package depenjds on it Aug 01 01:32:27 robbiethe1st, true Aug 01 01:32:33 I'd be looking at main/testing; devel we all know has it Aug 01 01:32:57 robbiethe1st, but backups from kernel power doesn't work with stock anyeway, due to modules Aug 01 01:33:06 so someone who want load stock kernel msut start from scratch anyway Aug 01 01:33:07 Eh. ot dpes/ Aug 01 01:33:11 It still does Aug 01 01:33:22 Because you still have the stock modules, unless you delete them Aug 01 01:33:24 bah, this eapd daemon is turning out to be a pain in the ass to reverse engineer :( Aug 01 01:33:38 When BM starts, it copies the currently running kernel's modules over Aug 01 01:33:39 Doesn't help that it appears to have been written in C++ Aug 01 01:33:39 robbiethe1st, AFAIK it doesn't from practice, see psots in Your thread, but Aug 01 01:33:47 kernel with thumb2 support indeed deloete stock one Aug 01 01:33:49 and it's modules Aug 01 01:33:55 and you know as good as me that it's the future :) Aug 01 01:34:07 no wait, not kernel with thumb support Aug 01 01:34:08 I've done it several times. What doesn't work is a backup with stock modules and a PK kernel partition Aug 01 01:34:10 cssu-thumb deletes it Aug 01 01:34:16 yes yes, agreed Aug 01 01:34:42 what i mean is that cssu-thumb deletes stock kernel (as it's useless for thumb system, anyway, it won't work reliably with stock kernel) Aug 01 01:35:05 (due to requiment for thumb silicon erratum) Aug 01 01:35:09 But as it is now, thumb is a very... specialized thing. Only for expert tweakers Aug 01 01:35:33 frankly? Everyone with single brain cell use it, sometime,s when noobs take it, with dramatic results ;) Aug 01 01:35:41 but I agree, you may as well wait with it until it becomes stable Aug 01 01:35:53 I am not using thumb yet :) Aug 01 01:36:04 heck, I dont even use CSSU yet Aug 01 01:36:07 or for community kernel, that will happen, sooner or later anyway Aug 01 01:36:12 in this or another form Aug 01 01:36:29 Same. I'm using PK, because it's very important(Need OC), but CSSU, so far... Aug 01 01:36:58 Mainly because I'd likely need to make a new system image, start from scratch etc. Aug 01 01:37:04 IMHO, it depends on "how soon freemangordon witll lost patience for useless discussions with DocScrutinizer" and either convince CSSu maintainers to community kernel, or continue to work on thumb fork of CSSU Aug 01 01:37:27 robbiethe1st, why yopu need OC? i'm using it too, but I wonder Aug 01 01:37:35 kernel-power have many things even more important bundled in Aug 01 01:37:39 tons of bugfiuxes Aug 01 01:37:44 Makes the system actually usable. Aug 01 01:37:52 just like CSSU ;) Aug 01 01:37:54 600mhz is just way too slow. 950mhz is actually decent. Aug 01 01:37:57 cssu-thumb, to be precise Aug 01 01:38:12 why so? 900 is great, and smart reflex works thanks to freemangordon Aug 01 01:38:22 Even with that, Slashdot brings the system to a crawl when loading a page. It can take 30s to load. Aug 01 01:38:25 950 or 1000 is psychological achi9evment, not practical one Aug 01 01:38:26 (with CPU pegged) Aug 01 01:38:34 well, blame slashdot Aug 01 01:38:50 Yeah, but the whole /purpose/ of the device is to do what I need, right? Aug 01 01:39:05 it depends, if You need to fly to mars sitting on it, then, no ;p Aug 01 01:39:11 seriously though, sure Aug 01 01:39:19 SR with 900 mhz have muuuch better battery life, though Aug 01 01:39:25 If I didn't need it's webbrowser and such, I could just as soon use my N950, or heck, N770. Aug 01 01:39:28 950 may be unstable with both SR ON Aug 01 01:39:47 no no, I'm rather allergic to bloated useles pages like slashdot or faceshit ;P Aug 01 01:39:49 My current setup is 100% stable, at full load for long periods of time. Aug 01 01:39:55 of course everyone is free to use whatever feel fancy Aug 01 01:40:02 are You using SR? Aug 01 01:40:08 No Aug 01 01:40:20 so it's tradeoff between 50 mhz and *much* battery life Aug 01 01:40:23 Custom voltages, slightly above the crashing level at each speed. Aug 01 01:40:25 your choice, of course Aug 01 01:40:33 no need, Sr does it better, anyway Aug 01 01:40:41 (custom voltages) Aug 01 01:40:49 Considering that 90% of my battery life is used when the device is sleeping, I don't really notice it. Aug 01 01:40:52 That and the display. Aug 01 01:40:59 huh? why so? Aug 01 01:41:03 probably, because lack of SR Aug 01 01:41:16 my device uses 3-4 mA per hour, when idle Aug 01 01:41:22 4-7 with GSM on Aug 01 01:41:29 5-9 with GSM + wifi Aug 01 01:41:39 It lasts about a day, possibly two with no use for me. Aug 01 01:41:46 divice 3000 mAh, and You will see how much standby time it is Aug 01 01:42:00 But I've also got a stock battery Aug 01 01:42:19 so i'm even more surprised, that You waste standby battery life, to get 50 mhz more Aug 01 01:42:26 as Sr would undervolt it as hell during standby Aug 01 01:42:34 rising voltage to Your current levels only during heavy usage Aug 01 01:42:44 as Your static voltage settings are OK for heavy usage, doesn't crash Aug 01 01:42:46 I've already done that. Major manual undervolting at low speeds Aug 01 01:42:52 so for idle they're overkill too huge Aug 01 01:43:06 not even half as effective a smartreflex, though Aug 01 01:43:16 not to mention risk of filesystem corruption Aug 01 01:43:20 It's still tiny, compared to, say, the screen. Aug 01 01:43:24 + thing that monitors it running constantly Aug 01 01:43:28 eating more power than you ghain Aug 01 01:43:39 well, You've said that most power is sued when idle, though Aug 01 01:43:44 used, damn* Aug 01 01:43:59 well, do whatever You want, but i'm sure You're wasting at least whole day of power availability Aug 01 01:44:07 and gain from 50 mhz is neglible, if any Aug 01 01:44:13 Well, I know it's better than stock/ Aug 01 01:44:27 but worse than it could be, without any effort required :) Aug 01 01:44:29 Plus, erm, why would that 50mhz hurt? it should never get to that speed when idle Aug 01 01:44:39 Shouldn't get above 500. Aug 01 01:44:42 ever.\ Aug 01 01:44:51 because You need voltage set up for it for 500, 600, 720, 800, 850 mhz too Aug 01 01:45:07 as your custom undervolting is OK only for lowest states, probably Aug 01 01:45:17 also,monitor frequencies and undervolting then eats more power than You gain Aug 01 01:45:26 it's kinda batterypatch-like way of doing things Aug 01 01:45:43 (it did same thing - undervolting on loiw speed, + additionaly, custom speeds when idle) Aug 01 01:45:55 = you're wasting more power for sole act of monitoring - it goes even when device is idle... Aug 01 01:45:57 than you gain Aug 01 01:45:57 I just use the kernel-settings thing. Aug 01 01:46:00 trust me on that (tm) Aug 01 01:46:09 ah, this kind of undervolting Aug 01 01:46:11 no way Aug 01 01:46:18 it's static undervolting, then Aug 01 01:46:19 And locked each speed, then undervolted till it crashed, upped it by three notches. Aug 01 01:46:25 Sr udnervolts not only basing on speed, but also on load Aug 01 01:46:37 Your settings are what is required for max load on such speed Aug 01 01:46:41 so you avoid crash Aug 01 01:46:41 So I have settings for everything between 250 and 1150mhz. Aug 01 01:46:49 SR does it constantly, and does it good, after freemangordon fixed it Aug 01 01:46:52 Which seems importantly. Aug 01 01:47:02 I.e. no risk of crashing. Aug 01 01:47:11 = for every speed, You must have constant HIGHEST load value, to avoid crash Aug 01 01:47:17 Yes Aug 01 01:47:29 well, Sr uses LOWEST values when approriate, highest when approriate, and everything in between the same way Aug 01 01:47:32 without risk of crash ;) Aug 01 01:47:58 at the cost of 50 mhz? BTW, this single step from 900 to 950 mhz stress Your CPU much more than you gain Aug 01 01:48:05 900 is considered max "sane" frequency Aug 01 01:48:11 (that's why Sr is calculated only up to 900 mhz) Aug 01 01:48:27 (it could be higher, but no real gain on using more than 900 mhz, and much risk) Aug 01 01:48:28 I am not using SR, and my settings work perfectly nicely. Next best thing: Reducing the amount of time my CPU spends in any state other than 0mhz at idle. Aug 01 01:48:53 Honestly, I'd be using 1.15ghz, if it didn't heat up so much... even 1ghz makes the device warm under max load. Aug 01 01:48:54 not likely, as your device seems to eat lotta shit of power when idle Aug 01 01:49:21 well, if you burtn your device out and wouldnb't be happy with android shit, remember, I have N900's with reinforced USB port for sale, 180 euro one :) Aug 01 01:49:47 look, on power-save savyy device, You should get - with worn battery... Aug 01 01:50:03 1000 mAh / 10 mAh Aug 01 01:50:05 with GSM on Aug 01 01:50:07 wifi ON Aug 01 01:50:09 24/7 Aug 01 01:50:17 + some idiotic widget that east power too Aug 01 01:50:21 it's 100 hours of idle time Aug 01 01:50:30 more than 4 days Aug 01 01:50:32 Well, lets see... 19 hours went from 1210mAh down to 720mAh. That's idle, with wifi on, no cell or bluetooth Aug 01 01:50:33 on worn battery Aug 01 01:50:45 well, i counted with cell on Aug 01 01:50:50 Estel_:you talk too much. Aug 01 01:50:54 + rounded it up in + for ease of calculations Aug 01 01:51:05 RST38h - IRc have nice little "close" button on top right corner Aug 01 01:51:15 thumbs up for using it, when You don't like it Aug 01 01:51:47 RST38h, sadly, you're right. But meh. Aug 01 01:51:53 robbiethe1st, after calculations, your device used 26-30 mA per hour Aug 01 01:51:54 on idle Aug 01 01:52:04 it's ~3x too much Aug 01 01:52:09 Yes. And I believe it's due to apps running when they shouldn't. Aug 01 01:52:23 well, robbiethe1st, ythe same apply to you - I've already told you that it's Your device, and You're free to do whatever You wabt Aug 01 01:52:30 You still try to convince me, that it's power savy, though Aug 01 01:52:52 quite possible, but at least half of it is due to unoptimized settings like lack of Sr and stupid max frequency Aug 01 01:53:08 anyway, it's true that I have better things to do, so see ya Aug 01 01:53:16 Can you run powertop for me? Aug 01 01:53:19 I'm curious. Aug 01 01:53:30 robbiethe1st, sure Aug 01 01:53:34 I have 97% time in C4 Aug 01 01:53:50 I'm very oriented toward power saving, so I use it quite often ;) Aug 01 01:54:02 any specific data You're interested in? Aug 01 01:54:04 That's likely my problem then... 93% of time in C4 Aug 01 01:54:09 20-30mA is approx my idle with the usual stuff i have open Aug 01 01:54:17 3.1 and 3.7% in c2-3 Aug 01 01:54:27 ShadowJK, sure, but for robbiethe1st, it's 26 mA with only wifi on, no celullar, no things Aug 01 01:54:37 robbiethe1st, 93% in C4 is good too Aug 01 01:54:54 and for sure doesn't explain why You're wasting 4x more power than should (when I said 3x, I forget about cellular being off) Aug 01 01:55:14 robbiethe1st, your times are in normal averages Aug 01 01:55:40 I don't understand why You insist, that Your settins re frequency and SR effects in this power wastage Aug 01 01:55:49 ShadowJK, what kernel are you using, and are you using SR? Aug 01 01:55:49 just curious, as it's well proved and researched fact Aug 01 01:55:55 I wonder when channel in general will lose patience with useless estel Aug 01 01:56:20 DocScrutinizer, overreactive again? Sorry if You don't like it, but i'm free to express my opinion about Your stane re community kernel Aug 01 01:56:41 like You're free to rise your ego via attacking me due chat with Woody14_something Aug 01 01:56:47 deal? Aug 01 01:57:00 robbiethe1st; stock Aug 01 01:57:19 ShadowJK, and no OC? Aug 01 01:57:44 I'm learning ssh to control n900 on pc, from the wiki http://wiki.maemo.org/SSH, I need to run 'ssh root@tablet ip' on pc, how to know what my n900 was called? Aug 01 01:57:45 but, DocScrutinizer, actually, you could be very useful here - explain robbiethe1st why 950 mhz isn't healthy frequency ;) Aug 01 01:57:52 nope Aug 01 01:58:07 You're good in arguing just for sake of arguing, and i'll finish dist-upgrading easy debian to wheezy, in meantime Aug 01 01:58:10 He's tried. Aug 01 01:58:47 Once things get a bit more tested, I might try out SR if I get my second N900 back running again(broken USB) Aug 01 01:58:48 oc/sr is less relevant for idle Aug 01 01:58:48 because in idle it's 0MHz 0V anyway :P Aug 01 01:58:52 Mourning, ShadowJK Aug 01 01:59:13 I know the question is naive Aug 01 01:59:35 ShadowJK, yea, but You know that this idle isn't perfect, and higher voltage during it wastes power anyway (during wakeups) Aug 01 01:59:39 But how to? Aug 01 01:59:49 especially, if it's highest voltage required to run 250 without crash Aug 01 01:59:56 ShadowJK, yeah. But he at least thinks he's getting away with 4ma at idle Aug 01 02:00:04 LaoLang_cool, install qnetman from repositories Aug 01 02:00:14 it have widget that tells You IP ;) Aug 01 02:00:23 much better and faster ways to Aug 01 02:00:26 but this one is convenient Aug 01 02:00:33 ...I wonder if having BatteryEye installed adds a noticible reduction in power? Aug 01 02:00:37 Estel_, Estel_ I know how to get my ip, but I don't know how to get my n900's name Aug 01 02:00:48 robbiethe1st, unless You have it open all the time, no Aug 01 02:00:56 LaoLang_cool, you don't need name for ssh Aug 01 02:00:58 How does it track, then? Aug 01 02:01:01 hey sweethearts Aug 01 02:01:02 if You have openssh installed Aug 01 02:01:15 robbiethe1st, it doesn't it just suxx info from hal, like bme Aug 01 02:01:26 Estel_, ah, thank you, learning it :) Aug 01 02:01:36 no problem Aug 01 02:01:37 just use Aug 01 02:01:44 ssh root@ip Aug 01 02:01:49 no need for name :) Aug 01 02:01:56 BTw, remember, password authentication is quite unsecure Aug 01 02:02:01 Now at 125mA powerdraw. 20 chans irc on 2g, gpodder through bluetooth a2dp Aug 01 02:02:06 after learning basics, You should switch to public key authentication, as in wiki Aug 01 02:02:22 Hello RiD Aug 01 02:02:22 i got an old infobot running on my phone. Doesn't seem to be writing the factoids, even when as root Aug 01 02:02:27 remind me, which timezone You're in? Aug 01 02:02:35 GMT. Aug 01 02:02:38 London/Lisbon Aug 01 02:02:45 RiD, once I tried to make infobot running from my wrt54gl router ;) Aug 01 02:02:53 RiD, go back to bed ;) Aug 01 02:02:53 lol Aug 01 02:03:25 Estel_, thank you! Learning it follows wiki instruction :) Aug 01 02:03:43 no problem, ask if You have other concerns :) Aug 01 02:03:56 Estel_: a router where you can run perl? Aug 01 02:04:04 RiD, sure Aug 01 02:04:10 you can run debian on it, if willing to Aug 01 02:04:14 its wrt54g* Aug 01 02:04:17 family Aug 01 02:04:24 lol, crazy routers then Aug 01 02:04:24 actually, I run openwrt on it Aug 01 02:04:30 20$ used Aug 01 02:04:32 ;) Aug 01 02:04:33 mine is the dumbest there is Aug 01 02:04:36 well, maybe 30 Aug 01 02:04:56 well, my router runs dircproxy (irc bouncer) for me Aug 01 02:05:09 anyways, a (super) noobish question. How to set file write/read permissions? Aug 01 02:05:10 I plan to install voip PCBX on it Aug 01 02:05:33 chmod Aug 01 02:05:58 and file owners - chown Aug 01 02:06:40 Estel_, wow, I'm using n900 via my pc, feel exciting! Aug 01 02:06:53 great thing, i know this feeling :) Aug 01 02:07:10 wait until you will use Your Pc from your N900 via vnc through ssh tunnel ;) Aug 01 02:07:28 it's a bit slow for that Aug 01 02:07:35 or n900 from Pc with visual interface via x11vnc on N900 and VNC viewer on PC, also through ssh tunnel Aug 01 02:07:43 RiD, with correct settings it isn't bad Aug 01 02:07:51 I use TigerVNC on desktop Aug 01 02:07:55 well, there are not miracles Aug 01 02:07:57 and presence on device Aug 01 02:08:02 yep Aug 01 02:08:12 rdesktop is faster overall, but no 3d accel or w/e Aug 01 02:08:17 so you're limited Aug 01 02:08:27 I've controlled my desktop from my N900... via straight SSH Aug 01 02:08:28 interestingl;y, with properly configured x11vnc on N900 and tigervnc viewer on computer, it's so fast that tux racer is playable from PC ;) Aug 01 02:08:34 who needs GUI? Aug 01 02:08:35 anyways, you told me to go and I will...out of battery Aug 01 02:08:43 robbiethe1st, GUI programs need gui ;) Aug 01 02:08:49 Estel_: how many fps would you say? Aug 01 02:08:55 RiD, 5? Aug 01 02:09:00 smth like that, maybe more Aug 01 02:09:04 or not Aug 01 02:09:12 as for vnc, it was quite good ;P Aug 01 02:09:13 try on 320x240 Aug 01 02:09:14 :D Aug 01 02:09:25 would need to put device in such mode, too much hassle Aug 01 02:09:27 I don't want to use all of my n900's function via pc, just to edit file or something like that :) Aug 01 02:09:35 you mean pc? Aug 01 02:09:36 LaoLang_cool, via vi? Aug 01 02:09:40 Estel_, yes Aug 01 02:09:43 RiD, no, N900 Aug 01 02:09:44 vim Aug 01 02:09:58 LaoLang_cool, sure, why not Aug 01 02:10:04 oh, you're talking about controlling n900 from pc? Aug 01 02:10:09 RiD, yes Aug 01 02:10:15 RiD, yes :) Aug 01 02:10:17 i misunderstood it Aug 01 02:10:20 in this scenario tux racer was playable from n90 visible on desktop Aug 01 02:10:25 other way, it was slower Aug 01 02:10:25 i was thinking the other way Aug 01 02:10:48 well, enough to play heroes 3 in multiplayer, on desktop 1000 miles away Aug 01 02:10:48 but maybe the other way while forcing a terrible resolution on pc it might not be so slow Aug 01 02:10:55 possible Aug 01 02:11:10 with nvidia cards (idk with ati) you can add custom resolutions Aug 01 02:11:11 A question, I'm using putty on windows to connect n900, can I copy some files from n900 to my pc? Aug 01 02:11:13 it was funny to play Heroes with someone sitting on PC, in qwuasi-locally setup Aug 01 02:11:18 being connected to same desktop Aug 01 02:11:32 LaoLang_cool, using scp Aug 01 02:11:32 only for fullscreen programs though. For desktop i guess you'll have to override the drivers Aug 01 02:11:39 I want to backup some files under /root and /home/user Aug 01 02:11:45 LaoLang_cool, but most conveinent way is to use WinSCP on windows Aug 01 02:11:46 for that Aug 01 02:11:47 Estel_, need another tools?... Aug 01 02:11:57 LaoLang_cool, no, SCP is bundled in SSH Aug 01 02:12:02 anyways, out of battery so I'll leave sooner than i thought Aug 01 02:12:05 I thought putty can do it :( Aug 01 02:12:09 but much, much more conveinent way is to use WinSCP program, which is another tool Aug 01 02:12:13 (sharing code with putty, btw) Aug 01 02:12:14 3...2...1... power off Aug 01 02:12:19 bb rid Aug 01 02:12:28 LaoLang_cool, as said, using command scp, You can Aug 01 02:12:39 * LaoLang_cool will download winscp Aug 01 02:12:40 but there is program on top of it, winSCPO, that makes it very conveinent Aug 01 02:12:49 WinSCP* Aug 01 02:13:05 Maemo wiki have some info about winscp IIRC Aug 01 02:14:03 Estel_, yes, under the page about ssh, it tells benifit of wiscp and the download link Aug 01 02:16:46 Estel_, I'm using winscp, what protocol I should use? sftp, scp, ftp Aug 01 02:16:56 sftp Aug 01 02:17:46 Estel_, I'm using it, seems easy to use :) Aug 01 02:18:01 yea, it's great Aug 01 02:18:40 Estel_, Is there a console version like it? actually, I'd like to use rsync on n900, and rsync directories on n900 with directories on pc Aug 01 02:19:15 rsync on n900 can be used over network as ios, IIRC Aug 01 02:19:25 and for console like, well, you have ssh client and scp command, as said Aug 01 02:19:45 WinSCP just combines smart using of LS with commands to transfer files Aug 01 02:19:48 ls to list them Aug 01 02:19:55 + it uses sftp as more powerful protocol than scp Aug 01 02:20:02 but You can use scp command in ssh too Aug 01 02:20:16 although you need to manually write locations etc, see scp --help Aug 01 02:20:22 winscp is much more conveinent Aug 01 02:22:06 Estel_, hard to understand :( what I want is, I can run rsync in n900, and n900 can access my pc filesystem, for example, on /media/windows, then I can 'rsync /root /media/windows/backup/root' Aug 01 02:23:01 for that see rsync manual, it can work voer network Aug 01 02:23:05 but I haven't used it much Aug 01 02:23:12 LaoLang_cool, do you have a Linux desktop PC? Aug 01 02:23:15 so can't help here, search for rsync articles on wiki or TMO Aug 01 02:23:20 windoze Aug 01 02:23:25 Both of pc and n900 are connecting to the internet Aug 01 02:23:32 Create a windows share Aug 01 02:23:36 on your desktop Aug 01 02:23:45 robbiethe1st, I have only windows :( Aug 01 02:23:53 Estel_, thank you all the same! Aug 01 02:24:31 If it's not so easy to set up, then I stick to winscp Aug 01 02:24:32 What you'll need to do is mount the windows network share as a directory on your N900, then you can just rsync to it easily. Aug 01 02:25:47 check http://maemo.org/packages/view/wizard-mounter/ Aug 01 02:26:09 See if that will allow you to mount it. Aug 01 02:26:36 Once it's mounted, you can just type 'mount' in the N900's terminal to see where it's mounted, then rsync to that directory Aug 01 02:27:24 robbiethe1st, thank you, will read it. Aug 01 02:28:22 oh, samba, nfs, too advanced concept, I think it takes sometime to understand them, I will read them later Aug 01 02:29:02 Samba on n900 is quite easy, actually Aug 01 02:30:13 Estel_, I need to set up samba on windows firstly, then let n900 know samba, then mount the samba, then use it, right? Aug 01 02:30:14 Estel, help him please Aug 01 02:30:28 Isn't samba just a Windows share? Aug 01 02:31:14 no idea, never used it Aug 01 02:31:24 but I remember reading about it and it was easy + popular Aug 01 02:32:24 oh, will learn it later, use winscp at present, thank you Estel_ and robbiethe1st ! Aug 01 02:32:45 no problem, have fun :) Aug 01 02:32:51 robbiethe1st, lame question Aug 01 02:32:58 I have package set on hold, via Aug 01 02:32:59 echo libpulse0 "hold" | dpkg --set-selections Aug 01 02:33:07 could You remind me command to unset hold for it? Aug 01 02:33:10 i.e. unlock it? Aug 01 02:33:16 :) Aug 01 02:33:18 Absolutely no clue. Aug 01 02:33:22 never messed with apt mor dpkg selections too much Aug 01 02:33:29 huh, thanks anyway Aug 01 02:33:32 Try "release" Aug 01 02:33:39 OK'ish Aug 01 02:35:13 release is unknown state :P Aug 01 02:35:57 "unlock"? Aug 01 02:36:13 "unhold". Aug 01 02:36:18 does maemo have a repo for oovoo for the n900? Aug 01 02:36:21 Oh, "install" to unlock. Aug 01 02:38:31 robbiethe1st, thanks Aug 01 02:41:34 dist-upgrade to wheezy - 825 MB to download, 725 more will be used Aug 01 02:41:36 0_o Aug 01 02:42:15 xd Aug 01 02:51:29 I'm learning phone control via dbus at http://wiki.maemo.org/Phone_control#Start_vibrating, a question, why I need to tweak vibrating for each pattern? Aug 01 02:51:54 do I need: Aug 01 02:51:56 dbus-send --system --print-reply --dest=com.nokia.mce /com/nokia/mce/request com.nokia.mce.request.req_vibrator_pattern_activate string:PatternIncomingCall Aug 01 02:52:03 dbus-send --system --print-reply --dest=com.nokia.mce /com/nokia/mce/request com.nokia.mce.request.req_vibrator_pattern_activate string:PatternCommunicationSMS Aug 01 02:52:06 ? Aug 01 02:52:50 vibrating is global option, I can't find where I can set it for call, sms, im respectively Aug 01 02:55:14 The section "Set a profile value" at http://wiki.maemo.org/Phone_control#Set_a_profile_value , there is another way to start vibrating: dbus-send --type=method_call --dest=com.nokia.profiled /com/nokia/profiled com.nokia.profiled.set_value string:"general" string:"vibrating.alert.enabled" string:"On" Aug 01 02:55:22 I'm confused with them **** ENDING LOGGING AT Wed Aug 01 02:59:59 2012