**** BEGIN LOGGING AT Mon Oct 10 02:59:59 2016
Oct 10 05:37:21 <KotCzarny> oh, lol, i just got a notification about '#maemo-meeting' on 2016-10-09 at 14:00 UTC 
Oct 10 05:38:03 <KotCzarny> on another note, what's the point of additional channel? if they wanted general audience with irc users, this channel could do?
Oct 10 05:40:09 <KotCzarny> or not. *shrugs*
Oct 10 06:59:54 <Oksana> Synfig may be interesting to port...
Oct 10 07:12:38 <bencoh> to n900?!
Oct 10 07:13:02 <bencoh> oooh they released more morevna project "episodes"
Oct 10 07:14:27 <KotCzarny> well, before starting to port, open synfig, load sample project, then check memory usage in top etc
Oct 10 08:13:32 <Maxdamantus> Port as in optimising the UI for a small touchscreen/restricted keyboard?
Oct 10 08:13:38 <Maxdamantus> or port as in getting it to compile?
Oct 10 08:13:43 <KotCzarny> both
Oct 10 08:14:31 <Maxdamantus> The latter really shouldn't be a thing.
Oct 10 08:14:57 <Maxdamantus> Why hasn't someone just ported the maemo things to Debian?
Oct 10 08:14:58 <KotCzarny> might be if libs are too old
Oct 10 08:15:16 <KotCzarny> which might require adding glue funcs
Oct 10 09:41:12 <kerio> Maxdamantus: maemo is stuck at debian etch
Oct 10 09:41:20 <kerio> roughly
Oct 10 09:46:52 <Maxdamantus> Mainly because of glibc requiring 2.6.32?
Oct 10 09:48:20 <bencoh> no
Oct 10 09:48:55 <bencoh> there's a quick&dirty workaround for that one, and it's probably the least significant issue anyway
Oct 10 09:49:31 <KotCzarny> hmm
Oct 10 09:49:37 <KotCzarny> so why dont we have 2.6.32 yet?
Oct 10 09:49:49 <Maxdamantus> 2.6.32 isn't the easy bit.
Oct 10 09:50:06 <Maxdamantus> the easy bit with glibc would be just patching glibc itself.
Oct 10 09:50:30 <bencoh> it's more about closed-source core applications/services that depends on old ABIs
Oct 10 09:50:33 <KotCzarny> go for it then send a patch to fmg ?
Oct 10 09:50:40 <bencoh> like glib, gtk, telepathy, ...
Oct 10 09:50:54 <bencoh> -s
Oct 10 09:51:43 <Maxdamantus> Presumably those should be fixable gradually.
Oct 10 09:51:59 <Maxdamantus> by just using LD_LIBRARY_PATH/LD_PRELOAD hacks around programs that require the old libraries.
Oct 10 09:52:18 <bencoh> you cant afford that on a memory-starved embedded device
Oct 10 09:52:28 <KotCzarny> true that
Oct 10 09:52:36 <KotCzarny> otoh newer libs are usually fatter
Oct 10 09:52:48 <Maxdamantus> Mm, maybe.
Oct 10 09:52:50 <KotCzarny> so upgrading might not have positive total effect
Oct 10 09:52:56 <bencoh> and it'd be a real deps hell
Oct 10 09:53:05 * Maxdamantus is always running ntpd and mpd from a different root.
Oct 10 09:53:16 <KotCzarny> i wonder if maemo could be patched to some saner pkg system
Oct 10 09:53:24 <KotCzarny> ie. repack everything in the way arch has
Oct 10 09:53:37 <KotCzarny> (or slackware, he he)
Oct 10 09:53:44 <bencoh> the "real" solution would be to keep working on FOSS replacements for core components, but ....
Oct 10 09:53:47 <Maxdamantus> (though those probably involve libraries smaller than glib/gtk)
Oct 10 09:54:36 * Maxdamantus also uses Opera, so the browserd crap is mostly wasted.
Oct 10 09:54:49 <bencoh> once you can build from source every maemo core ports you're pretty good to go
Oct 10 09:54:57 <bencoh> but I don't see that happen anytime soon
Oct 10 09:55:01 <Maxdamantus> Dunno. I can't really see single duplicate gtk/glibc being a major issue.
Oct 10 09:55:06 <KotCzarny> maxd: sms and other components use browserd to render 
Oct 10 09:55:14 <Maxdamantus> KotCzarny: yes, that was my point.
Oct 10 09:55:22 <Maxdamantus> I basically have two browser engines in memory.
Oct 10 09:55:54 <KotCzarny> maxd: gtk and friends is ~10-15MB
Oct 10 09:56:09 <bencoh> and browserd is ~40M, but...
Oct 10 09:56:12 <KotCzarny> so you will have to double that regarding memory usage
Oct 10 09:56:58 <bencoh> anyway I wouldn't want to deal with a double glib/gtk setup
Oct 10 09:57:01 <KotCzarny> imo on memory starved system having anything but essentials is a waste
Oct 10 09:57:33 <KotCzarny> ie. is browserd required while on desktop?
Oct 10 09:58:05 <Maxdamantus> "on desktop"?
Oct 10 09:58:17 <Maxdamantus> You mean without SMS windows open? I think so.
Oct 10 09:58:21 <KotCzarny> if i open sms app, let it load, but when im outside any of those apps it shouldnt be in mem
Oct 10 09:58:30 <Maxdamantus> though I suspect not much of it will be in memory at that point.
Oct 10 09:58:51 <bencoh> browserd actually eats quite a lot even when doing nothing
Oct 10 09:58:55 <Maxdamantus> Well, if you've already loaded it it will be in memory.
Oct 10 09:59:03 <bencoh> (mem-wise)
Oct 10 09:59:10 <KotCzarny> maxd, make it a stub that will unload when nothing is using it?
Oct 10 09:59:15 <Maxdamantus> It's not very good to constantly move things in and out.
Oct 10 09:59:22 <bencoh> true
Oct 10 09:59:41 <KotCzarny> or at least make some browserd-lite
Oct 10 09:59:43 <Maxdamantus> far better to just have a tiny libx11-based program that can always be in memory.
Oct 10 10:00:16 <Maxdamantus> KotCzarny: "unload when nothing is using it" means "loads when something uses it"
Oct 10 10:00:41 <Maxdamantus> Linux will already discard it from memory when newer things try to use it.
Oct 10 10:00:58 <KotCzarny> that means swap grind
Oct 10 10:01:12 <Maxdamantus> The browserd binaries/libraries are mmapped.
Oct 10 10:01:28 <Maxdamantus> swap would obviously be relevant for the anonymous pages.
Oct 10 10:01:34 <KotCzarny> hum
Oct 10 10:02:57 <KotCzarny>  /usr/bin/osso-addressbook this thing shouldnt be constantly in memory
Oct 10 10:03:10 <KotCzarny>  and this /usr/bin/image-viewer
Oct 10 10:03:20 <Maxdamantus> It's usually not all going to be in memory.
Oct 10 10:03:20 <KotCzarny> that would free 15MB alone
Oct 10 10:03:25 <Maxdamantus> They're mmapped files.
Oct 10 10:03:42 <KotCzarny> how to check what is in mem then?
Oct 10 10:03:59 <Maxdamantus> As processes try to access the content, they'll cause page faults which can copy things from the filesystem into memory.
Oct 10 10:04:54 <Maxdamantus> Then as other things try to use memory Linux will discard those pages since it knows it can just read them again from the filesystem when required.
Oct 10 10:05:11 <KotCzarny> i've killed image-viewer temporarily
Oct 10 10:05:15 <KotCzarny> 15MB of mem got freed
Oct 10 10:05:35 <KotCzarny> i mean 9
Oct 10 10:05:42 <KotCzarny> 4MB of swap and 5MB of ram
Oct 10 10:05:49 <Maxdamantus> If you were judging by "free memory" in `free`, that would probably be anonymous maps, not /usr/bin/image-viewer
Oct 10 10:06:06 <Maxdamantus> Killing a process won't normally immediately remove files from memory.
Oct 10 10:06:23 <KotCzarny> maxd, i just wanted to see if it makes a difference
Oct 10 10:06:32 <KotCzarny> bastard got restarted anyway
Oct 10 10:06:58 <Maxdamantus> That happens lazily: as processes try to access pages that aren't currently in memory, Linux will discard pages that haven't been accessed recently (possibly moving them to swap in the case of anonymous maps).
Oct 10 10:07:40 <KotCzarny> maxd, then i repeat my question, how to check what is currently occupying real mem?
Oct 10 10:07:57 <Maxdamantus> There are some obscure interfaces in /proc
Oct 10 10:08:36 <Maxdamantus> maybe `vmallocinfo`
Oct 10 10:09:51 <Maxdamantus> Hm, no, not that.
Oct 10 10:09:54 <KotCzarny> smaps?
Oct 10 10:10:01 <KotCzarny> ie /proc/<pid>/smaps
Oct 10 10:11:16 <KotCzarny> its heap is ~2.5MB
Oct 10 10:11:23 <Maxdamantus> I think pagemap
Oct 10 10:11:24 <KotCzarny> which means allocated buffers
Oct 10 10:11:33 <Maxdamantus>     * Bit  62    page swapped
Oct 10 10:11:33 <Maxdamantus>     * Bit  63    page present
Oct 10 10:12:24 <Maxdamantus> ah, and kpagecount was the global thing I was thinking of.
Oct 10 10:15:01 <KotCzarny> i think private_dirty are things that are counted into real mem usage
Oct 10 10:15:18 <bencoh> KotCzarny: there is a .desktop property to autostart process in maemo, btw
Oct 10 10:15:20 <KotCzarny> (and shared_dirty)
Oct 10 10:15:20 <Maxdamantus> "real mem usage"?
Oct 10 10:15:27 <bencoh> something-prestart
Oct 10 10:15:27 <Maxdamantus> real mem usage should normally be 100%
Oct 10 10:15:33 <Maxdamantus> unless you've just booted the system.
Oct 10 10:15:35 <bencoh> (x-maemo-prestart or something like that)
Oct 10 10:15:38 <Maxdamantus> unused RAM is basically a waste.
Oct 10 10:16:49 <Maxdamantus> See also: http://www.linuxatemyram.com/
Oct 10 10:17:01 <KotCzarny> maxd: im talking about memory that would have to be swapped out if new app requires more mem etc
Oct 10 10:17:48 <Maxdamantus> So resident anonymous maps?
Oct 10 10:18:08 <KotCzarny> apparently shared_dirty+private_dirty
Oct 10 10:18:46 <KotCzarny> bencoh: thx, i've just disabled it (for testing)
Oct 10 10:20:12 <KotCzarny> i guess i will go kamikaze and disable prestarting of browserd too
Oct 10 10:20:49 <KotCzarny> and messaging ui
Oct 10 10:21:36 <Maxdamantus> So then you just have to wait for it to load from the filesystem when someone sends you the first SMS rather than from swap (assuming other processes have pushed it out of memory).
Oct 10 10:21:50 <KotCzarny> hrm, it doesnt exit after clicking 'x'
Oct 10 10:22:04 <KotCzarny> maxd: just tried it, and startup time isnt much worse
Oct 10 10:22:25 <KotCzarny> ~1s when it loads msg db
Oct 10 10:22:33 <Maxdamantus> and yeah, it doesn't use a separate browserd process for each window.
Oct 10 10:22:50 <KotCzarny> well, im talking about /usr/bin/rtcom-messaging-ui
Oct 10 10:23:59 <KotCzarny> pity. prestarted trick would only help initial device boot time
Oct 10 10:24:08 <KotCzarny> apps dont quit on clicking 'x'
Oct 10 10:24:54 <Maxdamantus> Typical applications should. rtcom-messaging-ui probably considers itself special.
Oct 10 10:25:00 <Maxdamantus> like IE in some versions of Windows.
Oct 10 10:25:20 <KotCzarny> is it opensourced?
Oct 10 10:25:40 <Maxdamantus> Dunno.
Oct 10 10:25:51 <KotCzarny> ~closed
Oct 10 10:25:51 <infobot> methinks closed is http://wiki.maemo.org/Why_the_closed_packages or https://wiki.maemo.org/Fremantle_closed_packages, or http://elinux.org/N900
Oct 10 10:25:52 <Maxdamantus> Some of it is made up of accessible HTML/JavaScript/CSS, at least.
Oct 10 10:26:20 * Maxdamantus changed the way messaged are laid out by modifying that.
Oct 10 10:27:44 <KotCzarny> almost none of rtcom- packages are opensourced
Oct 10 10:28:44 <KotCzarny> https://wiki.maemo.org/images/9/94/Top_level_architecture.png
Oct 10 10:45:16 <KotCzarny> oh, it unloaded, yay
Oct 10 10:45:26 <KotCzarny> (after some inactivity)
Oct 10 10:45:51 <KotCzarny> so, right now i have ~41M free and 84M cached according to top
Oct 10 10:46:08 <KotCzarny> without any adverse effects (yet)
Oct 10 10:46:23 <Maxdamantus> It was probably due to activity, not inactivity.
Oct 10 10:46:47 <Maxdamantus> if you're talking about file pages being discarded from memory.
Oct 10 10:47:06 <KotCzarny> maxd, ive launched 'conversations' and closed it by 'x', then went doing some work, now i dont see it in the process list
Oct 10 10:47:21 <KotCzarny> that's what i mean by 'unloaded'
Oct 10 10:47:52 <KotCzarny> lets go leroy jenkins and unload call ui too
Oct 10 10:47:52 <Maxdamantus> Right, so its RSS went down.
Oct 10 10:48:31 <KotCzarny> though this might make me miss calls and i dont have second phone to check atm
Oct 10 10:49:00 <Maxdamantus> because other thing caused page faults trying to use memory, so those browserd pages became the least recently used at some point.
Oct 10 10:49:23 <KotCzarny> maxd, nope, phone was idle and there is ~40M free atm
Oct 10 10:49:49 <Maxdamantus> it's probably never truly idle.
Oct 10 10:50:07 <KotCzarny> idle as in screen blanked
Oct 10 10:50:17 <Maxdamantus> so it's still on.
Oct 10 10:50:25 <KotCzarny> yup
Oct 10 10:50:30 <Maxdamantus> things are still running, accessing memory.
Oct 10 10:50:35 <KotCzarny> yup
Oct 10 10:50:57 <KotCzarny> i hunt things that arent used by system/other things when user does nothign
Oct 10 10:51:01 <KotCzarny> *nothing
Oct 10 11:34:15 <bencoh> Maxdamantus: he just means that process eventually stopped running
Oct 10 11:35:22 <Maxdamantus> I don't think that's what he means.
Oct 10 11:35:36 <Maxdamantus> afaik browserd/rtcom-messaging-ui never just stops.
Oct 10 11:35:54 <KotCzarny> [root@Main-N900:/usr/share/applications/hildon]# ps axww|grep messaging
Oct 10 11:35:54 <KotCzarny>  2563 root      1596 S    grep messaging
Oct 10 11:35:54 <KotCzarny> [root@Main-N900:/usr/share/applications/hildon]#
Oct 10 11:36:09 <Maxdamantus> Hm. Weird.
Oct 10 11:36:21 <KotCzarny> if you set it not to preload always it will exit eventually
Oct 10 11:38:20 <KotCzarny> which is actually nice if you dont use sms much
Oct 10 12:07:20 <Sicelo> i have a silly bluetooth problem - N900 does not generate  a pin to pair for one particualr device (which it was able to pair fine before). it can still generate pin & pair others. any ideas how to hunt this one down?
Oct 10 12:07:47 <KotCzarny> maybe it's paired/authorized?
Oct 10 12:08:02 <Sicelo> i deleted all pairings on N900
Oct 10 12:08:40 <KotCzarny> <obligatory joke> reflash? </obligatory joke>
Oct 10 12:08:49 <Sicelo> haha
Oct 10 12:09:12 <KotCzarny> reset/take out batteries out of the bt device?
Oct 10 12:09:29 <KotCzarny> though it might have things written to some eeprom
Oct 10 12:09:57 <Sicelo> did that. no help :(
Oct 10 12:10:38 <KotCzarny> clear /var/lib/bluetooth ?
Oct 10 12:11:36 <KotCzarny> (maybe backup it first)
Oct 10 12:12:23 <Sicelo> didn't try that one .. will try shortly
Oct 10 12:12:46 <Sicelo> what i did was remove all entried in gconf
Oct 10 12:16:52 <Sicelo> and wanted to try reinstalling all bt* stuff
Oct 10 12:28:50 <Sicelo> or do we have alternative way to pair, e.g. via cmdline?
Oct 10 13:51:39 <Sicelo> yay! did it with simple-agent :)
Oct 10 14:08:21 <_maniac_> heh. I tried to update to CSSU-testing and phone became unresponsive. nokia logo -> blank screen and nothing. not even five dots.
Oct 10 14:08:57 <_maniac_> any way to find out what's broken?
Oct 10 14:09:26 <KotCzarny> low battery?
Oct 10 14:32:49 <_maniac_> no, battery was about 80% full when I did the update.
Oct 10 14:33:08 <KotCzarny> what happens when you try to turn it on?
Oct 10 14:34:07 <_maniac_> nokia logo and then black screen.
Oct 10 14:34:32 <_maniac_> it is in PC-Suite mode when I plug it in, will try ssh'ing now.
Oct 10 14:34:33 <KotCzarny> maybe bad kernel (ie. stock one)
Oct 10 14:34:56 <KotCzarny> were you updating from stock or from cssu?
Oct 10 14:35:02 <_maniac_> well, no replies from 192.168.2.15
Oct 10 14:35:06 <_maniac_> from cssu.
Oct 10 14:35:17 <_maniac_> I had cssu stable and power-kernel
Oct 10 14:35:25 <KotCzarny> got any weird packages? (ie. multiboot)
Oct 10 14:37:30 <_maniac_> bootmenu
Oct 10 14:37:40 <KotCzarny> ~bootmenu
Oct 10 14:37:48 <KotCzarny> ~ping
Oct 10 14:37:48 <infobot> ~pong
Oct 10 14:38:18 <KotCzarny> does it work still?
Oct 10 14:38:50 <KotCzarny> i would try flashing kernel alone (from cssu-testing)
Oct 10 14:40:32 <_maniac_> mhm, will try
Oct 10 14:52:38 <_maniac_> can't find kernel package in cssu-testing repo
Oct 10 14:54:01 <bencoh> I'd rather try kernel-power, but ...
Oct 10 14:54:14 <bencoh> aaah, right
Oct 10 14:54:21 <bencoh> which kernel did you have?
Oct 10 14:54:50 <bencoh> and the problem with reflashing kernel only is that you might end up with missing compatible modules
Oct 10 14:55:20 <KotCzarny> he had powerkernel before update
Oct 10 14:55:55 <KotCzarny> but also had bootmenu, which could be anything
Oct 10 15:20:35 <Sicelo> if he had bootmenu, then the vanilla kernel should boot, and he can then figure out which PK he had
Oct 10 15:42:35 <_maniac_> welp, my bootmenu don't have any kernel options, just 'internal nand (maemo 5)' and bootmenu utilities.
Oct 10 16:16:54 <Sicelo> that's vanilla kernel then ..
Oct 10 16:19:22 <_maniac_> oO
Oct 10 16:19:40 <_maniac_> I'm pretty sure I had power-something in uname -a before
Oct 10 16:22:10 <Sicelo> you may have .. but what i am saying is the vanilla kernel most likely still had the correct modules. that is what the default bootmenu setting uses. how you were booting your pk i don't know (assuming you didn't create any *.item files for it). just load the vanilla kernel without flashing and see if that gets you in. it should
Oct 10 16:56:35 <KotCzarny> http://i.imgur.com/eWqKRIK.png
Oct 10 16:56:57 <KotCzarny> winamp skin mode done
Oct 10 16:57:16 <KotCzarny> now onto fixing few bugs and packaging
Oct 10 17:37:34 <_maniac_> "if he had bootmenu, then the vanilla kernel should boot" - nope, my bootmenu boots to 2.6.28.10-power53
Oct 10 17:39:14 <_maniac_> I mounted rootfs but can't find anything in /boot there. according to /var/lib/dpkg/info/kernel{,-power}.list kernel images should be there.
Oct 10 17:45:56 <Sicelo009N> mine are in MyDocs/bootmenu.img.d/
Oct 10 17:47:07 <_maniac_> I don't have such directory
Oct 10 17:47:29 <_maniac_> but I extracted them from downloaded deb and going to try flashing now.
Oct 10 17:47:32 <Sicelo009N> then how are you using bootmenu? or for what? :-)
Oct 10 17:49:03 <_maniac_> I installed boot menu to be able to do full fs backup in case of fuck ups like this.
Oct 10 17:49:08 <_maniac_> but I didn't do one.
Oct 10 17:49:44 <KotCzarny> well, you can always run rescueos for that
Oct 10 17:49:50 <KotCzarny> no need to flash anything to device
Oct 10 18:18:56 <KotCzarny> http://i.imgur.com/zI6XzIE.png
Oct 10 18:18:58 <KotCzarny> :)
Oct 10 18:20:18 <_maniac_> almost like the original!
Oct 10 19:45:39 <_maniac_> ah, I give up and going full reflash.
Oct 10 19:45:49 <KotCzarny> :)
Oct 10 19:46:13 <KotCzarny> new start, new possibilities
Oct 10 19:46:18 <KotCzarny> go for css-thumb
Oct 10 19:46:21 <KotCzarny> cssu-thumb
Oct 10 19:48:27 <_maniac_> well, latest reflash  (which I was running before) was broken in various small ways
Oct 10 19:48:54 <_maniac_> aptitude segfaulting, extra decoders package doing NOTHING, missing url input in UML-browser
Oct 10 19:49:38 <KotCzarny> sounds like you were having fun with it
Oct 10 19:50:49 <_maniac_> !!FUN!!, yes.
Oct 10 21:11:49 * Maxdamantus just makes squashfs backups occasionally .. and used one once.
Oct 10 22:29:31 <_maniac_> nice. default HAM giving me "temporary failure resolving" anything, while I am able to nslookup from commandline properly.
Oct 10 22:33:15 <_maniac_> but cssu-testing enabler installs fine
Oct 10 22:33:32 <_maniac_> I hate computers.
**** ENDING LOGGING AT Tue Oct 11 02:59:58 2016