**** BEGIN LOGGING AT Sun Jan 15 03:00:01 2017 Jan 15 03:00:18 What is bootmenu? Part of u-boot? Jan 15 03:00:31 Do you have another N900 to experiment with beforehand? Jan 15 03:00:40 U-boot-flasher says it will replace existing kernel Jan 15 03:00:48 Not at hand Jan 15 03:02:17 It's a menu in u-boot, so when you turn the device on, you can choose to boot the appended kernel or run a script from the first partition of the SD card, etc Jan 15 03:02:49 I just want a u-boot which will not try to confli Jan 15 03:03:02 conflict with kernel-power Jan 15 03:03:53 Yeah, so you'd probably want one that has the version of kernel-power you want included. Jan 15 03:05:04 Though it might be better ultimately to have a boot partition on eMMC that holds your kernels, so you don't have to reflash anything when you want to upgrade. Jan 15 03:05:16 u-boot-for-power-kernel sounds like it constructs the image, not like it pre-includes it Jan 15 03:09:06 Installing bootmenu Jan 15 03:09:36 Installing bootmenu Jan 15 03:16:29 Changed battery. Bootmenu installed. It doesn't conflict kernel-power, right? How to check whether it auto-detects SD card to boot from? Jan 15 03:19:48 Leds are still not working, all the same. Wonder what the problem is... Jan 15 03:20:52 When I say "bootmenu", I mean the one in u-boot, not the userland "bootmenu" that exists in some Maemo package. Jan 15 03:21:54 Does `lsmod` show `leds_lp5523`? Jan 15 03:22:40 lsmod | grep lp leds_lp5523 10048 0 led_class 3460 2 leds_lp5523,leds_twl4030_vibra Jan 15 03:30:49 Same on your side? Any way to see errors of kernel modules? Because kernel modules for lp5523 exist, but lp5523 do not appear in /sys/class/leds Jan 15 03:31:22 Yes, same here. Jan 15 03:32:00 At least, vibrator works, that's good. Jan 15 04:34:32 It's all confusing. u-boot (lots of packages), backupmenu (potentially useful, auto-notices SD card?), multiboot (not good with kernel-power), recovery-boot, nitdroid-installer (not needed atm?), bootmenu and bootmenu-n900 Jan 15 06:56:08 Bagged the n9 for $10.50. :) Will pick-up during the week. Can a sim card be tested without the door? Jan 15 07:05:23 Michael_a380: you _could_, definitely do not recommend it though. you'll end up having to fish it out of the SIM cage if you do it wrong Jan 15 07:12:51 Better just get the SIM tray. Shipping may cost, though. Jan 15 07:14:49 Unless you want to drill the SIM card and put a thread through it, to make it easier to get SIM card out... But margin for error is small, make sure to not damage the SIM card. Jan 15 07:17:58 Doesn't seem *that* small, given that practically any SIM card you buy is just a nano-SIM inside a micro-SIM inside a mini-SIM inside a full size SIM. Jan 15 07:18:19 Looks like there's a decent amount of room in the mini-SIM part to drill a small hole. Jan 15 07:20:42 (the relevance of the nano-SIM thing is that there is obviously no circuitry around the nano-SIM portion) Jan 15 07:21:44 Don't remember size of SIM for N9, but it's smaller than size of SIM for N900. Jan 15 07:22:14 thanks Jan 15 07:22:26 It's a micro-SIM. Jan 15 07:23:07 er, I meant there's a decent amount of room in the micro-SIM part, not the mini-SIM part. Jan 15 07:23:15 don't think you could fit a mini-SIM in the slot. Jan 15 07:24:18 though if you can figure out the dimensions, I imagine you could sand down a mini-SIM so it fits properly in the slot, probably with something behind it to wedge it in place. Jan 15 07:24:30 * Maxdamantus doesn't want to recommend it. Jan 15 07:26:28 Yes, microsim does still have some good space. Jan 15 07:36:37 Very helpful, thanks. I probably not risk it. Jan 15 07:36:51 *I'll Jan 15 08:34:51 freemangordon, Russia. As for ads, sorry, mine just silently blocked them all :) Jan 15 08:41:58 freemangordon: wasn't gitorious supposed to be fully archived by archive.org at some point? Jan 15 08:48:35 'Oksanaa > At least, vibrator works, that's good.' Jan 15 08:48:42 * KotCzarny snickers Jan 15 08:56:37 bencoh: it was.... Jan 15 09:24:10 freemangordon: did you check archive.org then? Jan 15 10:58:39 "vibrator works, that's good" Jan 15 10:58:40 heh Jan 15 10:58:54 vroom Jan 15 10:59:17 ^ x2 Jan 15 11:39:17 hmm somehow ifconfig is lost from my device Jan 15 11:40:12 lol forgot to be root Jan 15 11:40:17 scratch that Jan 15 11:42:45 Tried apkenv 42.4.0 for Fremantle from TMO, doesn't seem to work. With Petals Redux.apk, just opens a black/blank window... And slows everything down. Jan 15 11:44:45 Using GLES version 2, it says. Any dependencies I should have installed? Jan 15 12:15:08 bencoh: by archive team, I think Jan 15 12:23:47 Wizzup: indeed Jan 15 12:24:01 (they are not archive.org fyi) Jan 15 12:24:11 right, they're not exactly archive.org Jan 15 12:24:31 afaiu either the archive team or one of the archive team member is supposed to maintain it Jan 15 13:02:18 Hi! can anyone please tell me which groups normal user is assign to? I miss -a option when adding User to sudo group and now I cant root (sudo: gainroot: command not found) Jan 15 13:05:03 GoNeL: your problem is not of group assignment Jan 15 13:05:09 is that gainroot isn't installed Jan 15 13:05:19 command not found Jan 15 13:05:30 ceene, maybe path issue? Jan 15 13:05:43 ie. gainroot modifying user profile or something? Jan 15 13:05:55 but yeah, checking if its installed is also good idea Jan 15 13:05:56 could be, but... Jan 15 13:05:57 it is because before doing thar i just have to write root and rootsh do the job for me Jan 15 13:06:01 i don't have my n900 here right now Jan 15 13:06:07 i use rootsh anyway Jan 15 13:06:12 or ssh -l root Jan 15 13:06:39 i'm sshing it right now with root account Jan 15 13:07:34 im having weird issue with oscp on n8x0 Jan 15 13:07:40 when running as root, audio is ok Jan 15 13:07:54 when running as user its garbled, as if byte order swapped Jan 15 13:08:23 cant even begin to think what could be the problem Jan 15 13:10:17 ahm. stupid me. Jan 15 13:11:22 Pali: are you sure the stuff in /etc/init.d is not executed on maemo? Jan 15 13:11:31 IIRC no Jan 15 13:12:10 upstart does not have "native" support for /etc/rc*.d/ Jan 15 13:12:24 Pali: because in /etc/inittab there is inittab:l2:2:wait:/etc/init.d/rc 2 Jan 15 13:12:49 it is "emulated" by upstart service which execute then it Jan 15 13:13:24 but still, the result is that /etc/init.d/* services are getting executed Jan 15 13:13:32 I'm not sure if upstart even use inittab Jan 15 13:13:46 maybe Maemo or Debian upstart doing something different Jan 15 13:18:07 Pali: but then, why there are so much things in /etc/init.d? Jan 15 13:18:31 probably legacy Jan 15 13:18:37 from debian or older maemo Jan 15 13:19:37 hmm. there is no clockd in /etc/event.d Jan 15 13:19:45 arent there some compatibility job that would run them? Jan 15 13:20:21 looks loke Jan 15 13:20:30 but then, thing are gitting weird Jan 15 13:22:00 for example sapwood-server is executed by bot /etc/X11/Xsession.post and /etc/init.d/af-services Jan 15 13:22:05 but tbh on all my debian boxes i remove systemd and install sysvinit Jan 15 13:24:07 I can't grok what's going on here :( Jan 15 13:24:59 grep upstart scrips for init string? Jan 15 13:25:28 maybe some binary executes them directly Jan 15 13:25:44 hmmm, isn't maemo init system a hybrid (upstart+sysv)? Jan 15 13:26:16 KotCzarny: doesn;t matter, the point is that sapwood-server is started from 2 places Jan 15 13:26:26 bencoh: no idea Jan 15 13:26:33 but looks like Jan 15 13:26:35 iirc both upstart scripts and regular init scripts are interpreted Jan 15 13:27:32 add 'echo $0 >>/tmp/rclog;date >>/tmp/rclog' everywhere? Jan 15 13:27:56 KotCzarny: what for? Jan 15 13:28:12 to pinpoint at which point those sapwood things run? Jan 15 13:28:14 I anm not interseted in who starts it, but why it is started twice Jan 15 13:28:44 I know when is it started Jan 15 13:28:49 what happens if you run it third time? graceful exit? Jan 15 13:29:19 the problem is that it seems to be firs started from a script in /etc/init.d, before there is X running Jan 15 13:29:32 and of it fails Jan 15 13:29:35 *ofc Jan 15 13:29:45 disable it and check if system still works? Jan 15 13:29:51 :( Jan 15 13:30:03 before knowing why it is like that? Jan 15 13:30:12 maybe if x/wm dies its just for restarting? Jan 15 13:30:24 I doubt Jan 15 13:30:36 if X dies on maemo, you got a reboot Jan 15 13:30:58 what is sapwood for anyway? Jan 15 13:31:23 KotCzarny: https://github.com/fremantle-gtk2/sapwood Jan 15 13:34:18 so, pixmap server? Jan 15 13:36:07 GoNeL: watch out with sudo! it doesn't work the way you might think Jan 15 13:37:18 actually messing with etc/sudoers is one of the most frequent reasons for bootloop Jan 15 13:37:40 Pali: what if dsmetool is asked to start the same binary twice? Jan 15 13:37:47 i'm not understanding what is going on here... maybe it was the problem.. Jan 15 13:37:57 do not know :-) need to test Jan 15 13:38:33 you know how can i revert it? I remove User from sudo group (27) but nothing changed Jan 15 13:38:45 maybe this is it - all those mutiple starts are through dsmetool and it just ignores the request when the binary is already started Jan 15 13:38:49 cannot edit PATH, permission dennied Jan 15 13:39:22 fmg, disable from sysvinit and/or upstart and check what stops working Jan 15 13:39:58 hmm, I am sudoers (with password) on my n900, and it works Jan 15 13:41:14 KotCzarny: no, I'd rather check when is stuff in /etc/init.d started on maemo Jan 15 13:41:31 see? told you to add rclog Jan 15 13:41:32 :) Jan 15 13:43:31 check out update-sudoers script! Froups are not the problem. Our problem is very old sudo which is pimped to act like newer one, with scripts Jan 15 13:43:54 IroN900:~# which update-sudoers Jan 15 13:43:55 /usr/sbin/update-sudoers Jan 15 13:44:02 ~jrtools Jan 15 13:44:02 extra, extra, read all about it, jrtools is http://wiki.maemo.org/User:Joerg_rw/tools Jan 15 13:44:15 right, I checked update-sudoers beforehand and added it properly Jan 15 13:44:21 don't remember where though Jan 15 13:45:11 ah, in sudoers.d Jan 15 13:47:42 Nokia-N900:~# update-sudoers Jan 15 13:47:42 Nokia-N900:~# which update-sudoers Jan 15 13:48:30 it's /usr/sbin/update-sudoers Jan 15 13:51:44 read it, see /etc/sudoers.d/ Jan 15 13:52:05 tl;dr: never edit /etc/sudoers ! Jan 15 13:53:18 if you want to edit sth about sudo, do it in /etc/sudoers.d/* (ideally your own file you add there) - then **run update-sudoers** Jan 15 13:54:22 then check if sudo still works as supposed, **while keeping a root shell open, to revert what you messed up** Jan 15 13:56:43 about groups, see `man sudoers` - and keep in mind what you just learned from reading update-sudo script Jan 15 13:56:55 you can find some very terse hints in Jan 15 13:56:58 ~jrtools Jan 15 13:56:59 it has been said that jrtools is http://wiki.maemo.org/User:Joerg_rw/tools Jan 15 13:58:08 on a broader note: I don't know if I ever heard of a really 100% successful add of new user (except 'root' and 'user') in maemo Jan 15 13:58:56 a full interactibe user account != user|root might not work like expected, occasionally Jan 15 13:59:04 interactive* Jan 15 13:59:20 ok thnks, i'ĺl it! Jan 15 13:59:48 there's a lot of tricky stuff about users, in maemo Jan 15 14:00:25 e.g. you should be *very* careful of redefining *user* shell, while root's shell is mostly harmless to redefine Jan 15 14:01:13 ~liskeys sudo Jan 15 14:01:26 ~listkeys sudo Jan 15 14:01:27 Factoid search of 'sudo' by key (6): cmd: sudo (.*?) ;; sudo get me a sandwich ;; sudo path ;; sudo ;; sudo get me a sandwitch #DEL# ;; sudo make me a sandwich. Jan 15 14:01:46 ~sudo Jan 15 14:01:46 [~sudo] Better than su, according to talon. It allows a permitted user to execute a command as the superuser or another user, as specified in the sudoers file. Or can allow you to do silly things like run X11 apps with root privileges; also good in scripts with "username ALL = NOPASSWD: /some/program", or http://www.aplawrence.com/Basics/sudo.html, or good for ordering sandwiches, or not pseudo Jan 15 14:03:03 ~sudo make me a sandwich Jan 15 14:03:03 make me a sandwich: sudo make docscrutinizer05 a sandwich. Jan 15 14:03:43 lol Jan 15 14:04:02 I recently learned that there's a risk I become a sandwich whe doing this Jan 15 14:04:24 * DocScrutinizer05 doesn't want to be a sandwich Jan 15 14:04:42 sudo turn DocScrutinizer05 into a sandwich Jan 15 14:05:51 wasn't there a `make cow` or sth? Jan 15 14:06:48 bencoh: weird: I thought xkcd was native speaker. https://xkcd.com/149/ Jan 15 14:07:49 so why did he use this very ambiguous form? Jan 15 14:09:24 GoNeL: re sudo - what is it exactly you want to achieve? Jan 15 14:10:07 infobot slap sudo Jan 15 14:10:07 * infobot slaps sudo, keep your grubby fingers to yourself! Jan 15 14:11:07 * DocScrutinizer05 also idly wonders if we might get decent contemporary sudo in CSSU, instead of this obsolete bitrotten shit and update-sudoaers could get lost in a place nobody ever again remembers Jan 15 14:12:19 a new decent sudo would allow smart stuff like wrapping up HAM into sudo Jan 15 14:12:41 **WITH PASSWORD** Jan 15 14:13:29 Pali: bencoh: KotCzarny: mystery revealed - maemo supports both upstart and init.d scripts, the same goes for devuan. the difference is that in devuan there must be LSB header for the correct startup order Jan 15 14:15:46 -A, --askpass Jan 15 14:15:48 Normally, if sudo requires a password, it will read it from the user's terminal. If the -A (askpass) option is specified, a (possibly graphical) helper program is exe- Jan 15 14:15:49 cuted to read the user's password and output the password to the standard output. Jan 15 14:17:23 freemangordon: :) Jan 15 14:19:08 IroN900:~# sudo -A /bin/true Jan 15 14:19:10 sudo: illegal option `-A' Jan 15 14:19:38 guys I might have to become religious and thank god or something, i dont know why but now it just start working! Jan 15 14:21:04 IroN900:~# l `which sudo` Jan 15 14:21:05 -rwsr-xr-x 2 root root 91584 2009-04-17 10:50 /usr/bin/sudo Jan 15 14:21:07 IroN900:~# sudo -V Jan 15 14:21:08 Sudo version 1.6.8p12 Jan 15 14:21:27 GoNeL: you forgot to run update-suoers and you rebooted which does that for you Jan 15 14:22:34 fmg: that's good, because i bet there are still people using sysvinit (/me being one of them) Jan 15 14:22:49 odds however are that after reboot it doesn't 'magically work' but rather any fsckup you introduced causes a bootloop Jan 15 14:23:13 that's why you *ALWAYS* run update-sudoers after messing with sudo Jan 15 14:23:24 as I explained above Jan 15 14:24:48 `less /usr/sbin/update-sudoers` -- really _highly_ recommended Jan 15 14:25:08 yeah i think that was my problem Jan 15 14:25:18 thank yo DocScrutinizer05 Jan 15 14:28:55 Pali: freemangordon: could we get sudo >1.8 in CSSU? pretty please! Jan 15 14:29:51 and nuke that update-suoers shite then, since it's not needed anymore for >=1.8 Jan 15 14:32:22 it would also speed up boottime by prolly 0.73 seconds, _and: add "sudo -A" option and the equivalent askpass function to 5 sudoers Jan 15 14:33:22 if sudo.conf(5) contains a line specifying the askpass program, that value will be used. For example: Jan 15 14:33:23 # Path to askpass helper program Jan 15 14:33:25 Path askpass /usr/X11R6/bin/ssh-askpass Jan 15 14:36:29 freemangordon: Pali: we could change stuff like "%users ALL = NOPASSWD: /usr/libexec/apt-worker" to do proper asking for (root!) password then Jan 15 14:37:50 would most likely work like a charm :-) Jan 15 14:42:37 GoNeL: note that visudo is pointless on maemo (for interactive use), since you're _not_ supposed to ever edit the file /etc/sudoers Jan 15 14:46:55 GoNeL: you still could use `visudo -f /etc/sudoers.d/mysudo` but since you're supposed to run update-sudoers after any such edit anyway, it's quite pointless Jan 15 15:03:12 and honestly what is this (last line in my /etc/sudoers.d/01sudo): "user ALL = (install) NOPASSWD: /usr/bin/app-installer-tool" Jan 15 15:03:26 o.O Jan 15 15:03:41 hmm Jan 15 15:03:42 ? Jan 15 15:03:55 hmm, how one takes screenshot on n8x0 ? Jan 15 15:04:02 I wonder if *I* added that nonsense eventually Jan 15 15:04:23 there's no /usr/bin/app-installer-tool Jan 15 15:06:07 of course also no user "install" Jan 15 15:07:40 * DocScrutinizer05 idly wonders if some crap like fapman or whatever might have done `echo "user ALL = (install) NOPASSWD: /usr/bin/app-installer-tool" >>/etc/sudoers.d/01sudo` in postinst Jan 15 15:11:14 `grep "01sudo" /var/lib/dpkg/info/*.postinst` -->zilch Jan 15 15:11:53 bencoh: could you please check last line in your /etc/sudoers.d/01sudo Jan 15 15:13:04 same thing Jan 15 15:13:09 duh! Jan 15 15:13:14 wtf is this? Jan 15 15:13:15 legacy stuff maybe Jan 15 15:13:22 prolly, yes Jan 15 15:22:27 maemo4 actually? Jan 15 15:22:59 * DocScrutinizer05 finally needs to fix his N810 rootfs Jan 15 15:24:06 it's embarrassing it eventually started bootlooping and I never got around to fixing that Jan 15 15:39:24 grrr http://maemo.org/packages/view/sudo/ Jan 15 15:47:50 https://github.com/search?q=org%3Acommunity-ssu+sudo :-/ Jan 15 15:58:26 freemangordon: >>The include directive<< https://wiki.debian.org/sudo Jan 15 15:58:57 hmm? Jan 15 15:58:57 freemangordon: could you compile a recent version of sudo for fremantle? (and provide it) Jan 15 15:59:13 there should be, in the repos Jan 15 15:59:22 where? can't find it Jan 15 15:59:31 oh, wait Jan 15 15:59:42 I was thinking about busybox Jan 15 15:59:47 :) Jan 15 15:59:50 meh!! :-) Jan 15 15:59:56 DocScrutinizer05: can;t do it now Jan 15 16:01:54 it's most likely a bad idea to add it to extras-devel autobuilder, I guess? Jan 15 16:03:10 (not like I'd be savvy how ro accomplish that, I'd rather actually install a SV VM and try locally, but really I'd rather avoid that) Jan 15 16:03:30 SB* Jan 15 16:04:15 I'd ruthlessly gab HARM binary and run it on fremantle, if only I could find it Jan 15 16:05:01 but then: GARM (aka Aegis) sudo?? BWAHAHAHA Jan 15 16:09:44 http://www.sudo.ws/dist/sudo-1.8.19p2.tar.gz would be nice Jan 15 18:46:23 Pali: do you have a clue how 'root' user on maemo is able to access 'user' user dbus session bus? Jan 15 18:47:03 yes :-) Jan 15 18:47:09 run stanalone .sh Jan 15 18:47:18 or how it is called that script Jan 15 18:47:34 run-standalone.sh Jan 15 18:47:35 Pali: sure, but how dbus allows it? Jan 15 18:47:54 communication with dbus daemon is done via filesystem socket Jan 15 18:48:07 yes,yes Jan 15 18:48:08 DBUS_SESSION_BUS_ADDRESS Jan 15 18:48:11 and socket file is passed to dbus app via env Jan 15 18:48:26 yes, but that soesn't work on debian Jan 15 18:48:31 *doesn't Jan 15 18:48:33 and address for user session dbus is in some /tmp file Jan 15 18:48:58 DBUS_SESSION_BUS_ADDRESS=unix:abstract=/tmp/dbus-NHndQJLZj1,guid=52eee25672b0202e4b74c9cc587b9421 Jan 15 18:49:05 BUT... Jan 15 18:49:05 also dbus-daemon has some policy file which specify who and how can access it Jan 15 18:49:23 so you can filter out some users for some dbus interface Jan 15 18:49:36 hmm, do you know which file is that Jan 15 18:49:49 in /etc/ is for system dbus Jan 15 18:50:09 yes, but what about session bus? Jan 15 18:50:17 do not know Jan 15 18:50:34 according to google, root cannot access other user's session bus Jan 15 18:50:50 even if it have the socket address Jan 15 18:51:10 it must use seteuid in order to do that Jan 15 18:54:11 hmm, there is some session.conf file Jan 15 18:58:20 are there any download/install stats for packages on extras/-devel ? Jan 15 19:00:31 crzyp3ck: use: screen /dev/ttyUSB1 115200 maybe? Jan 15 19:00:46 ECHAN. eh Jan 15 19:19:41 ~repos Jan 15 19:19:41 well, repos is "deb http://maemo.muarf.org/apt-mirror/mirror/downloads.maemo.nokia.com/fremantle/ssu/apps/ ./ ;; deb http://maemo.muarf.org/apt-mirror/mirror/downloads.maemo.nokia.com/fremantle/ssu/mr0/ ./", or see http://wiki.maemo.org/Repository#List_of_Maemo_repositories, or see ~jrrepos Jan 15 19:19:52 ~maemo-repos Jan 15 19:19:53 rumour has it, maemo-repos is http://wiki.maemo.org/Repository#List_of_Maemo_repositories Jan 15 19:34:17 ~jrrepos Jan 15 19:34:17 somebody said jrrepos was http://maemo.cloud-7.de/maemo5/et_al/HAM-catalogs/ Jan 15 19:35:28 * DocScrutinizer05 thinks ~repos is maximum non-helpful Jan 15 19:41:12 whoah! installed python-matplotlib .. expected huge amounts of rootfs to get used up .. seems to have had negligible effect. interesting Jan 15 19:41:39 most likely because python optified already Jan 15 19:41:50 it should install to the bindomount python stuff I'd guess Jan 15 19:44:46 ssh root@iron900 "mount|grep pymaemo" Jan 15 19:45:00 cmd: /opt/pymaemo/usr/lib/python2.5 on /usr/lib/python2.5 type bind (bind) Jan 15 19:45:02 cmd: /opt/pymaemo/usr/share/pyshared on /usr/share/pyshared type bind (bind) Jan 15 19:45:03 cmd: /opt/pymaemo/usr/lib/pyshared on /usr/lib/pyshared type bind (bind) Jan 15 19:45:05 cmd: /opt/pymaemo/usr/share/python-support on /usr/share/python-support type bind (bind) Jan 15 19:45:06 cmd: /opt/pymaemo/usr/lib/python-support on /usr/lib/python-support type bind (bind) Jan 15 19:46:55 special case of Jan 15 19:47:01 ~optification Jan 15 19:47:02 optification is a inventive duct tape workaround to reclaim space in fs root, done due to the fact the systeminit *and* partitioning is FUBAR, http://wiki.maemo.org/Documentation/Maemo_5_Developer_Guide/Packaging,_Deploying_and_Distributing/Installing_under_opt_and_MyDocs, or ""OMG - I wish they looked into FHS and moved /usr to eMMC"", http://www.pathname.com/fhs/pub/fhs-2.3.html#PURPOSE2 bullet1,2 and fhs-2.3.html#PURPOSE16 dot3" Jan 15 19:47:47 sicelo-: ^^^ Jan 15 19:48:04 sicel o-: Jan 15 19:48:13 at least this solves one of my problems/concerns .. have something that needs matplotlib, and i had been holding back unnecessarily Jan 15 19:57:49 ssh root@iron900 "du -sh \$(mount|grep pymaemo|sed 's/^$[^ ]*$ on .*/\1/')" Jan 15 19:58:01 cmd: 63.4M /opt/pymaemo/usr/lib/python2.5 Jan 15 19:58:03 cmd: 7.1M /opt/pymaemo/usr/share/pyshared Jan 15 19:58:04 cmd: 2.8M /opt/pymaemo/usr/lib/pyshared Jan 15 19:58:06 cmd: 280.0k /opt/pymaemo/usr/share/python-support Jan 15 19:58:07 cmd: 148.0k /opt/pymaemo/usr/lib/python-support Jan 15 20:00:20 i don't like optification the way it was done Jan 15 20:00:42 it should've been done at system level, without modification of .deb Jan 15 20:00:44 yeah. symlinks would be mych better Jan 15 20:00:57 i was thinking of unionfs Jan 15 20:01:21 huh? Jan 15 20:01:49 unionfs allows to show a filesystem as an overlay of two different mount points Jan 15 20:02:04 you might not be completely aware of the way optification got implemented Jan 15 20:02:17 packages need to be optified Jan 15 20:02:24 I know what's unionfs Jan 15 20:02:41 i don't want packages to be custom designed to allow or not allow for optification Jan 15 20:02:50 that should be a system setting Jan 15 20:03:46 maemo-optify-auto-opt.sh maemo-optify-firstboot.sh maemo-optify-lib.sh maemo-optify-make-mountbind.sh Jan 15 20:04:29 all of that is unnecesary if you just unionfs mount /opt with / Jan 15 20:04:41 bindmount sounds terribly inticated and fragile Jan 15 20:04:47 err sorry, unionfs Jan 15 20:05:06 again, see: Jan 15 20:05:12 ~optification Jan 15 20:05:13 optification is a inventive duct tape workaround to reclaim space in fs root, done due to the fact the systeminit *and* partitioning is FUBAR, http://wiki.maemo.org/Documentation/Maemo_5_Developer_Guide/Packaging,_Deploying_and_Distributing/Installing_under_opt_and_MyDocs, or ""OMG - I wish they looked into FHS and moved /usr to eMMC"", http://www.pathname.com/fhs/pub/fhs-2.3.html#PURPOSE2 bullet1,2 and fhs-2.3.html#PURPOSE16 dot3" Jan 15 20:05:58 Nokia failed to sanitize that stuff with PR1.2 Jan 15 20:06:58 so, is there any sensible way to fix this all // rebuild maemo system atop armhf jessie devuan or whatever ? Jan 15 20:07:11 I thought they would do optification as a transitional botch/hotfix and fix it in PR1.2 (or was it 1.3?). Alas they kept legacy crap Jan 15 20:07:46 i don't even have separate partitions on my PCs, where i have 1Tb+ hard disks Jan 15 20:07:59 so? Jan 15 20:08:01 sounds ridiculous having so many separate partitions on phones with limited storage Jan 15 20:08:21 that's also a problem on android phones Jan 15 20:08:40 how would you make one partition of two different *types* of storage? And don't say LVM now! Jan 15 20:08:56 sometimes you can't upgrade an application that would fit in available space because the partition it is to be downloaded prior to installation is full Jan 15 20:09:01 * Maxdamantus has Maemo/home all on a single 27.4 GiB partition. Jan 15 20:09:22 or you can download the installer because partition for temp files is empty but you can't install the app because app partition is full Jan 15 20:09:42 whereas it is all internal storage that doesn't make sense spliting Jan 15 20:09:55 then you have wrongly partitioned system Jan 15 20:09:55 so? this is of pretty minor relevance to maemo and N900, right? Jan 15 20:09:57 DocScrutinizer05: if you want to have MyDocs under vfat that could've been implemented as a filesystem on a file Jan 15 20:10:10 imo we should simply use nand only for rescueos/bootloader/boot and move whole os to emmc Jan 15 20:10:17 no, that's exactly the same that happens on n900, where you have plenty of space but it's in the wrong partition Jan 15 20:10:30 KotCzarny: is there a performance penalty? Jan 15 20:10:42 most of the os is on emmc anyway Jan 15 20:10:44 ceene: again, what got that to do with optification and different types of memory? Jan 15 20:10:51 once I used full Maemo on eMMC it was slower Jan 15 20:11:00 that optification shouldn't have needed to even exist Jan 15 20:11:05 having Maemo on both nand and eMMC is fater Jan 15 20:11:13 *faster Jan 15 20:11:17 pali, how much slower? Jan 15 20:11:23 ceene: please first read and understand what optification actually is Jan 15 20:11:26 Maxdamantus: .. how is your user experience with your setup? Jan 15 20:11:28 I remember slower bootup Jan 15 20:11:46 and whole system had more lags Jan 15 20:11:53 KotCzarny: way slower Jan 15 20:11:57 hum. Jan 15 20:12:05 from sd card at least Jan 15 20:12:09 maybe we need to optimize os too, to see what writes more Jan 15 20:12:09 DocScrutinizer05: from your first link: The N900 has about 100 MB of free space in the root file system partition. This is not very much and would fill up quite quickly when installing additional applications. As a workaround, the /opt directory has been linked to a different partition with more space (about 1 GB) and /home/user/MyDocs is also available in certain cases, with even more space Jan 15 20:12:15 didn;t try from emmc Jan 15 20:12:15 ain't that what i'm saying? Jan 15 20:12:38 sicelo-: it's still usable. My bootup is a lot slower, but I think that's mostly due to the fact that I'm using ext4 and it fscks before mounting. Jan 15 20:12:50 there is nothing wrong in having / on nand Jan 15 20:12:52 sd cards differ, try again with samsung evo+ Jan 15 20:12:54 (and usually when the device boots, it's because it crashed/froze) Jan 15 20:13:03 (so it has to do a long check) Jan 15 20:13:05 it would have been even better if it was read-only Jan 15 20:13:09 no, you say optification is not needed with different partitioning, but you ignore there's physically no possibility to make one partition out of two storage chips with different type of storage Jan 15 20:13:18 optification is a workaround needed because of a not opt-imal structure of memory Jan 15 20:13:28 but there is Jan 15 20:13:36 you can fake that with unionfs, as i said Jan 15 20:13:36 ceene: no, it is because n900 has only 256 MB of nand Jan 15 20:13:39 I also have swap on an SD card, which makes a decent difference iirc Jan 15 20:13:45 and you yourself suggested lvm Jan 15 20:13:54 and nand is times faster than emmc Jan 15 20:14:06 freemangordon: well, of course all storage problems can be solved via more storage Jan 15 20:14:30 ceene: please read ~optification *again*, fully, until last char Jan 15 20:14:30 ceene: and how would you do that on already assembled/shipped device? Jan 15 20:14:36 i just think optification could've been done differently Jan 15 20:15:01 moving core libs and binaries into /bin /sbin /lib and all non-core to /usr Jan 15 20:15:02 DocScrutinizer05: it was implemented in some of the PRs, right? 1.2? Jan 15 20:15:10 and then moving whole /usr into eMMC Jan 15 20:15:21 that is more unix way Jan 15 20:15:21 freemangordon: optification? iirc 1.1 Jan 15 20:15:26 ok Jan 15 20:15:40 and probably better to performance Jan 15 20:15:40 for some reason you keep unacknowledging my idea of using unionfs to implement optification Jan 15 20:15:46 Pali: we have maemo booting with devuan upstart :) Jan 15 20:15:46 Pali has it the right way Jan 15 20:15:46 instead hell of symlinks Jan 15 20:15:57 ou, nice! Jan 15 20:16:10 I'd rather explicitly have the "core" things stored separately. Jan 15 20:16:29 ceene: because pf ~optification explains that it's not needed at all, and you do NOZ need any unionfs to avoid that Jan 15 20:16:35 but now in systemd world /usr and / is one thing... Jan 15 20:16:42 Pali: some stuff like ttyusb0 and couple of others still need love, but those are not critical Jan 15 20:16:52 last time i brought similar idea, you folks have said nokia did bad things and essential tools for boot are on /usr too Jan 15 20:16:55 udev does not boot when /usr is in separate part as / Jan 15 20:17:25 DocScrutinizer05: having a set of six .sh scripts do what a simple entry on /etc/fstab would do seems more complex than necessary Jan 15 20:17:34 KotCzarny: yes... but this is now common for all linux distros Jan 15 20:17:37 that's one of the reasons why systemd is a pile of shit for embedded Jan 15 20:17:38 because of udev.. Jan 15 20:17:56 ceene: so what? Jan 15 20:18:19 so that's all i'm saying Jan 15 20:18:32 you may not agree, that's okay Jan 15 20:18:40 so read ~optification a 5th time, it says exactly same Jan 15 20:19:03 i don't think we're even discussing the same thing Jan 15 20:19:09 Well, does upstart replace udev? Can upstart handle /usr and / being in separate partitions? Could optification be changed in devuan-maemo? Jan 15 20:19:13 ~optification Jan 15 20:19:13 optification is a inventive duct tape workaround to reclaim space in fs root, done due to the fact the systeminit *and* partitioning is FUBAR, http://wiki.maemo.org/Documentation/Maemo_5_Developer_Guide/Packaging,_Deploying_and_Distributing/Installing_under_opt_and_MyDocs, or ""OMG - I wish they looked into FHS and moved /usr to eMMC"", http://www.pathname.com/fhs/pub/fhs-2.3.html#PURPOSE2 bullet1,2 and fhs-2.3.html#PURPOSE16 dot3" Jan 15 20:19:23 is what we're discussing Jan 15 20:19:49 yeah, but you're not giving any arguments in favour or against anything in particular i'm saying Jan 15 20:19:51 Oksanaa: afaik upstart and udev are different things Jan 15 20:19:56 what in ""OMG - I wish they looked into FHS and moved /usr to eMMC"" is unclear? Jan 15 20:20:00 so i'm even doubting this is a discussion :) Jan 15 20:20:42 Oksanaa: upstart is init daemon like sysvinit or sytemd Jan 15 20:21:05 it does not implement dbus or other parts... Jan 15 20:21:07