**** BEGIN LOGGING AT Mon Dec 05 03:00:00 2016 Dec 05 10:07:57 oink oink Dec 05 10:08:00 Wizzup oink Dec 05 10:08:04 hi! Dec 05 10:08:16 oink Dec 05 10:08:19 let me guess: you want grsec added in the mix? :P Dec 05 10:09:05 well, my friend tried installing grsec on debian and got a lot of problems with maintainers that don't care Dec 05 10:09:05 L29Ah: blogpost right? Dec 05 10:09:09 comparing to gentoo Dec 05 10:09:24 so i don't really expect it Dec 05 10:09:29 Wizzup: roadmap Dec 05 10:09:33 L29Ah: ack Dec 05 10:14:20 :] Dec 05 15:10:41 OBOA123 Dec 05 15:10:41  Dec 05 15:10:59 invalid password, please retry Dec 05 17:17:48 if an external dac works in android 4.0 or later will it work with the latest cssu stable? Dec 05 17:19:15 is support for it likely to be in the kernel Dec 05 20:15:01 Hello Dec 05 20:16:33 I am building keepassx 2.0.3 with FREMANTLE_ARMEL_GCC472 and I wonder how it may be published in extras-devel ? Dec 05 20:18:04 ymartin59: Hi! IIRC Maemo extras is not based on gcc 4.7 Dec 05 20:18:09 only gcc 4.2 Dec 05 20:18:33 only thumb-testing is based for gcc 4.7 Dec 05 20:18:54 http://wiki.maemo.org/Documentation Dec 05 20:29:18 OK but my first target is thumb I am running on device... Dec 05 20:29:42 I would try to "backport" to gcc 4.2 after Dec 05 20:29:54 maemo extras is not thumb based Dec 05 20:30:29 you must compile your application for non-thumb if you target maemo extras Dec 05 20:30:57 you can change scratchbox target from FREMANTLE_ARMEL_GCC472 to FREMANTLE_ARMEL (which is not thumb) Dec 05 20:34:47 you prolly want to upload a source package to extras* anyway, so ... Dec 05 20:35:09 submit to cssu-thumb maintainer? Dec 05 20:35:23 cssu-thumb is both: thumb AND 4.7.2 Dec 05 20:35:47 though if its not a system package.. Dec 05 20:35:48 cssu is not for additional applications Dec 05 20:35:58 it is only for system maemo packages Dec 05 20:35:58 yeah, just realised Dec 05 20:36:06 those which cannot be put into extras Dec 05 20:36:09 maybe we need extras-thumb ? Dec 05 20:36:16 yes :-) Dec 05 20:36:39 but I have no idea how hard is to configure such thing on maemo autobuilder... Dec 05 20:37:07 me neither, but idea is very nice, also there would be some priority issues Dec 05 20:37:42 and dependency problems Dec 05 20:38:41 Well I guess I can work harder to get this recent source code and build chain on ARMEL 4.2 with debhelper 7 Dec 05 20:44:05 actually we dont need extras-thumb Dec 05 20:44:15 we need another build variant Dec 05 20:45:02 (x86, armel, armelthumb) Dec 05 20:47:15 bencoh: no, it is still armel Dec 05 20:48:04 we can workaround a bit by depending on kernel-feature-errata-whateveritwas, but then you will pull a whole new kernel if you install from humb repo Dec 05 20:49:16 merlin1991: do you remember how often is stage synced to live? Dec 05 20:49:48 I know this is still armel stricly speaking, but ... how would you define it? Dec 05 20:50:07 it should be orthogonal to extras/testing/devel Dec 05 20:50:25 oops, -ECHAN Dec 05 20:50:33 bencoh: dunno Dec 05 20:55:34 I'd consider thumb another arch, like x86 also is another arch than armel Dec 05 20:55:41 extrasT, extrasT-devel ? Dec 05 20:55:45 strictly speaking it's another processor Dec 05 20:55:50 but Dec 05 20:56:05 if fmg and pali pull the mainline trick off, we will have real new arch, armhf Dec 05 20:56:06 :) Dec 05 20:56:17 which could be merged with thumb Dec 05 20:56:36 hehe, but not possible Dec 05 20:56:40 too many maemo blobs Dec 05 20:56:48 and armel != armhf, not compatible Dec 05 20:56:58 i know, that's why its another arch Dec 05 20:57:28 and question is if we can ignore those blobs and still have wanted functionality Dec 05 20:57:37 no, cannot Dec 05 20:57:47 not on n900 at least Dec 05 20:57:52 it is not another arch Dec 05 20:58:00 it is still armel architecture Dec 05 20:58:16 just with optional thumb2 isa Dec 05 20:58:25 exactly Dec 05 20:58:31 you can still run non-thumb2 binaries Dec 05 20:58:43 but not backward compatible Dec 05 20:58:44 and you can mix thumb2 and non-thumb2 libraries Dec 05 20:58:50 so kind pentium vs pentium-mmx Dec 05 20:58:51 KotCzarny: it is Dec 05 20:58:58 no Dec 05 20:59:07 well, on 686 you can run 486 Dec 05 20:59:08 it is backward compatible Dec 05 20:59:23 exactly as DocScrutinizer05 said Dec 05 20:59:35 the whole problem is that CPU is buggy, not that it is anoter arch Dec 05 20:59:37 but you cant run 686 on 486 Dec 05 20:59:51 that's why i've said 'backward' Dec 05 21:00:01 yea, it is not forward-compatible Dec 05 21:00:12 but is backward Dec 05 21:00:15 anyway, sleepy time. bb! Dec 05 21:00:53 I'd consider armel-thumb a pretty good solution Dec 05 21:01:17 seamless integration into maemo.extras* Dec 05 21:01:32 DocScrutinizer05: I think it will be not that easy Dec 05 21:01:46 we need to teach debhelper about armel-thumb Dec 05 21:01:56 you cannot use new architecture name Dec 05 21:02:00 you need to do that anyway, no? Dec 05 21:02:04 no Dec 05 21:02:12 because in debian you cannot mix different architectures easily Dec 05 21:02:33 I don't get it Dec 05 21:02:38 the only workaround is to put depends in debian/control Dec 05 21:02:40 thumb2 compiled packages must always be "armel" Dec 05 21:02:55 DocScrutinizer05: armel is ABI Dec 05 21:03:21 well, for this topic it's a key in repo management, nothing else Dec 05 21:03:26 maybe we could modify debhelper/dpkg/dpkg-buildpackage to automatically put Depends when compiling for thumb? Dec 05 21:03:29 whatabout restrict "thumb" binaries into a specific "lib" and "bin" folders ? Dec 05 21:03:49 s/key/attribute/ Dec 05 21:03:50 DocScrutinizer05 meant: well, for this topic it's a attribute in repo management, nothing else Dec 05 21:03:54 like multi-arch debian with "lib-i686" and "lib-amd64" ? Dec 05 21:03:55 Pali: yes, this is what I was saying with "Depends:" workaroud Dec 05 21:04:12 ok Dec 05 21:04:56 ymartin59: the point there is that i386 and amd64 define different ABI, not that they are instruction set wise incompatible Dec 05 21:05:09 OK understood Dec 05 21:05:09 the same stands for armel vs armhf Dec 05 21:05:52 a question: where could I get ARM specifications ? Dec 05 21:05:55 a differing ABI is no mandatory requirement though Dec 05 21:06:50 ymartin59: on ARM site Dec 05 21:06:55 and actually the ABI differs between arm-w/o-thumb and full arm Dec 05 21:07:06 Pali: maybe promote u-oot-tools as well Dec 05 21:07:16 DocScrutinizer05: no Dec 05 21:07:21 ok, promote Dec 05 21:07:42 sure, in thumb libs you may have impair function adresses Dec 05 21:07:44 function arguments are still passed through r0-r4 Dec 05 21:07:51 this is the ABI Dec 05 21:08:02 thumb2 has same -abi as arm Dec 05 21:08:09 :nod: Dec 05 21:08:24 no, mere ARM doesn't allow odd function addr Dec 05 21:08:35 Pali: me to promote it? Dec 05 21:08:47 and that is as much ABI as registers for parameters Dec 05 21:08:50 please do it Dec 05 21:09:29 it doesn't have enough votes :) Dec 05 21:09:39 and seems flasher does not depend on it Dec 05 21:09:56 irrespective, when a new ABI means you need a new arch doesn't mean the reciprocal "new arch needs new ABI" Dec 05 21:10:32 you can have two arch with identical ABI Dec 05 21:11:09 DocScrutinizer05: anyway, we will have to teach debhelper and whatnot for this new arch Dec 05 21:11:23 which is mission impossible Dec 05 21:13:25 I can't comment on that - never looked into debhelper. But assuming it's FOSS why would it be "mission impossible"? Dec 05 21:14:27 By the way, I just got my keepassx_2.0.2-1~maemo1_armel.deb package built with FREMANTLE_ARMEL_GCC472 toolchain Dec 05 21:14:39 on non-thumb any debhelper only needs to ignore the unknown new arch. On thumb targets you can install a patched debhelper (if that thing even needed on tatget) Dec 05 21:14:42 because it will need resources we don;t have :), both manpower and experience Dec 05 21:14:53 aaah, oooh Dec 05 21:15:05 ymartin59: congrats Dec 05 21:15:59 when we don't have the manpower to implement thumb into maemo-extras repo environment, we're busted anyway Dec 05 21:16:47 DocScrutinizer05: me, personally, for example - I prefer to spent my time on porting h-d to gtk3 and allwinner, that to spend it trying to kep stuff which will die as soon as the last n900 dies. which will be not that far from now Dec 05 21:17:33 o.O Dec 05 21:18:06 and that's maemo related? Dec 05 21:18:16 particularly thumb related? Dec 05 21:18:17 My N900 dying ? not yet... and still have spares Dec 05 21:18:34 why surprised? for how much longer do you think there will be n900 users? when almost every day one or another service on the device stops working? Dec 05 21:19:03 ymartin59: not HW wise, but SW wise Dec 05 21:20:05 * DocScrutinizer05 can'T follow that rationale at all Dec 05 21:20:36 Anyway I found a quick fix to build with gcc 4.2... Dec 05 21:20:53 I really think we should focus on moving maemo to other HW platforms and newer distros. n900 will gain from that as well. Dec 05 21:22:03 I agree... I would invest again in my first dream... port dalvik, now ART to standard Linux kernel and GNU stack ! Dec 05 21:22:41 It has been tried by Ubuntu... maybe ART would be easier to port. Dec 05 21:22:49 Pali: ask merlin1991 for u-boot-tools promotion please, he is the repo maintainer, I don;t want to abuse my admin powers Dec 05 21:23:17 Pali: strictly speaking, that package is not ready to be promoted to extras Dec 05 21:23:45 merlin1991, can you promote that u-boot package to extras? ↑↑↑ Dec 05 21:25:10 why on earch ARM specs are not public ? Dec 05 21:25:41 which ARM specs arent public? Dec 05 21:25:50 ymartin59: they should be Dec 05 21:26:13 ymartin59: https://www.arm.com/products/processors/cortex-a/cortex-a8.php Dec 05 21:26:33 https://developer.arm.com/products/processors/cortex-a/cortex-a8 Dec 05 21:26:57 everything is there Dec 05 21:27:00 maybe she's talking about something specific? Dec 05 21:27:08 it asks me for account registration Dec 05 21:27:12 where? Dec 05 21:27:20 ymartin59: https://developer.arm.com/docs/ddi0344/k/1-introduction Dec 05 21:28:21 OK many thanks for that direct links... Dec 05 21:29:24 hmm, well, I just clicked on the relevant links on ARM website Dec 05 21:29:33 I don't have those bookmarked Dec 05 21:30:09 ymartin59: but, if you have some specific question, better ask than bury in the docs. They are endless Dec 05 21:30:17 probably I should have looked for "cortex-a8" instead of "arm" alone Dec 05 21:30:24 yeah Dec 05 21:30:41 I have always loved endless readings Dec 05 21:30:58 and I have probably too much questions Dec 05 21:30:59 seems you have endless free time then :) Dec 05 21:33:48 I learned other assembly languages (6809, z80, 680x0, pentium and HP Saturn) and really ARM lacks in my culture Dec 05 21:34:20 I started to like it more than x86 ASM :) Dec 05 21:34:54 for sure x86 is just a pile of dusts Dec 05 21:36:00 bye. see you soon Dec 05 21:44:47 23:16 < freemangordon> DocScrutinizer05: me, personally, for example - I prefer to spent my time on porting h-d to gtk3 and allwinner, that to spend it trying to kep stuff which will die as soon as the last n900 dies. which will be not that far from now <<-- not too surprised. i been meaning to ask actually for the last few days Dec 05 22:31:37 this simply means: don't use maemo-thumb, it's not going to get any more maintenance Dec 05 22:32:56 I tried for quite some time now to come up with any other interpretation Dec 05 22:34:22 btw not like we discussed this maemo extras thumb repo thing the first time today Dec 05 23:03:15 * Wizzup has all his n900s on thumb Dec 05 23:13:42 Post rejected. As a new member, you are only allowed to post a minimum number of links in your post. Hit the back button on your web browser and edit your post. Thanks. Dec 05 23:13:45 eh. Dec 05 23:13:50 joys of using a new account Dec 05 23:22:54 http://talk.maemo.org/showthread.php?p=1520074#post1520074 Dec 05 23:27:54 are the thumb issues present in mainline kernels? Dec 05 23:43:39 Wizzup: *new* account?? what the heck...? Dec 05 23:43:53 DocScrutinizer05: I couldn't figure out what address I used for my old one Dec 05 23:44:09 I had a username and password written in my password files, but they didn't work Dec 05 23:44:15 ugh, ask maintainer! Dec 05 23:44:16 So I gave up and didn't want to postpone writing the post Dec 05 23:44:23 Yeah, I will, but it's not top priority atm :) Dec 05 23:44:33 :-) Dec 05 23:44:50 kudos for post (only read first 2 lines so far) Dec 05 23:45:46 wouldn't that thread rather belong to http://talk.maemo.org/forumdisplay.php?f=58 ? Dec 05 23:45:55 I promised to write it some time ago, and there's still a lot to do, but it's a good start, I hope Dec 05 23:46:19 DocScrutinizer05: could be, but in the end it may also run on things other than the neo900 Dec 05 23:46:36 and right now it's kind of hard to test on the neo900 :) so we use allwinner tablets or n900 or other Dec 05 23:46:38 NeoFremantle is not about Neo900 Dec 05 23:46:42 ah, sorry. Dec 05 23:46:47 Yes, then probably that may be a better place. Dec 05 23:47:51 sorry I can't move threads *into* 'my' subfora Dec 05 23:48:37 I need to get some sleep, let's see if we can get it moved tomorrow or so (some redirect for a bit would be nice) Dec 05 23:49:27 nope you can't, only your own posts Dec 05 23:49:36 yeah, I meant asking someone Dec 05 23:49:54 redirect is always implied Dec 05 23:49:58 automatically Dec 05 23:50:34 for forum admins it's a matter of literally 3 clicks to move a thread around Dec 05 23:50:45 ack Dec 05 23:50:49 I need to get some sleep first :-) Dec 05 23:50:57 you may *absue' the 'report post' feature Dec 05 23:51:32 "please move this thread to http://talk.maemo.org/forumdisplay.php?f=58" - done Dec 05 23:51:42 ;-) Dec 05 23:52:18 will automatically show up on the task list of those who are in charge Dec 05 23:53:08 oooh, maybe you as a 'new user' would not have the "report post" button enabled? Dec 05 23:54:18 if there's no such button uper right of every post, I can do that on your behalf Dec 05 23:54:58 only will do so when I got sth to quote with a clear request to do so Dec 05 23:55:30 otherwise I'd receive bashing for egoistic motivations for that Dec 05 23:55:48 I'll try to do it tomorrow. Dec 05 23:55:59 np, just offering to help Dec 05 23:56:44 repost the line above and I'll copy it to the requester ;-) **** ENDING LOGGING AT Tue Dec 06 03:00:00 2016