| dsc_ | stuck for past few hours on this one https://plak.infrapuin.nl/selif/zdz4uezc.png :p | 00:00 |
|---|---|---|
| Wizzup | it should be the daedalus branch of image builder | 00:01 |
| Wizzup | with submodules | 00:01 |
| Wizzup | and then | 00:01 |
| Wizzup | cd vm-sdk | 00:02 |
| Wizzup | zsh -f -c 'source sdk && load devuan maemo && build_vm_dist' | 00:02 |
| dsc_ | yeah that's fine | 00:04 |
| dsc_ | toor | 00:04 |
| dsc_ | whoops | 00:04 |
| dsc_ | setting apparmor=0 to grub cmdline does not work (as per debian wiki), which is something I observed earlier | 00:16 |
| dsc_ | I am also purging apparmor in the `finalize_purge_packages` of our image-builder .config | 00:17 |
| dsc_ | very strange | 00:17 |
| Wizzup | and here we are getting funding to bring apparmor support as first class citizen to our devices ;) | 00:19 |
| dsc_ | thats fine, but not for my dev environment | 00:20 |
| Wizzup | sudo apt remove --assume-yes --purge apparmor | 00:20 |
| Wizzup | ? | 00:20 |
| dsc_ | i think perhaps apparmor is a dependecy of another pkg maybe | 00:20 |
| dsc_ | yeah im purging etc. | 00:21 |
| dsc_ | its strange | 00:21 |
| dsc_ | https://plak.infrapuin.nl/selif/k6ikrrhr.png | 00:21 |
| dsc_ | trying to remove+purge `libapparmor1` here | 00:21 |
| dsc_ | anyway, I am in maintenance mode because of the bad mountpoint thing | 00:23 |
| dsc_ | ok nice, /etc/fstab just had empty UUIDs | 01:01 |
| freemangordon1 | sicelo: there is OFONO_ATOM_TYPE_SMS which can be used to register watch in ofono on new sms arrival | 07:30 |
| freemangordon | I wonder if SF guys use that somehow | 07:38 |
| freemangordon | hmm, what about if we do ofono dbus activation? | 07:45 |
| freemangordon | that way we can make sure tp-ring(or whoever wants sms-es) is the first to start ofono | 07:47 |
| sicelo | cellulard would also activate it. maybe it's cellulard that should be dbus activated. | 07:53 |
| freemangordon | it does not expose any dbus interface | 08:07 |
| Wizzup | arno11: on swap, I think we just remove any swap entries from the image builder and fix it here yeah? https://git.maemo.org/leste/leste-config/src/branch/master/debian/leste-config-n900.postinst | 11:14 |
| Wizzup | I think the main issue is that we want the image builder to write a rootfs line | 11:14 |
| freemangordon | anyone with SFOS phone here? | 11:33 |
| sicelo | still thinking about the sms issue? | 11:41 |
| sicelo | maybe one of us can join their irc chat room and ask | 11:41 |
| sicelo | i guess piggz would know, and he generally is in there | 11:41 |
| sicelo | i'll ask on telegram ... i think he's also there, and maybe others who know | 11:47 |
| arno11 | Wizzup: yeah @swap | 12:29 |
| arno11 | the last stuff that overrides fstab is arm-sdk/extra/n900/etc/fstab iirc | 12:31 |
| arno11 | @rootfs, no clue about the proper way | 12:32 |
| freemangordon | sicelo: yes, still about sms issue | 12:37 |
| freemangordon | I really wonder how they solve it | 12:37 |
| Wizzup | arno11: yeah the rootfs is the mai issue | 12:45 |
| Wizzup | main | 12:51 |
| sicelo | Wizzup: btw where are we taking one cpu offline on the d4? i checked in leste-config-* but doesn't seem to be there | 14:49 |
| Smatkovi | has anyone attempted cross-compilation on a pc for armv7? | 14:49 |
| gnarface | i did it for armv8, the process shouldn't be much different | 14:51 |
| gnarface | it should basically work everywhere | 14:51 |
| Wizzup | sicelo: I think this is part of mce | 14:57 |
| Wizzup | and we do this to save power when the screen is off | 14:57 |
| sicelo | we may want to revisit it at some point :-) | 15:01 |
| sicelo | if compiling via ssh, then you may want the 2nd core ;-) | 15:01 |
| Smatkovi | gnarface. | 15:09 |
| Smatkovi | sry. @gnarface how did you do it? | 15:09 |
| Wizzup | Smatkovi: I am not sure what your goal if but usually it's enough to make the applications work on the amd64 VM | 15:13 |
| Wizzup | the CI/CD will cross compile your program | 15:13 |
| Wizzup | unless you're running into architecture specific bugs? | 15:13 |
| dsc_ | https://plak.infrapuin.nl/selif/qv8hy9aa.png <== top bar not quite right, perhaps missing a package, or anyone who knows how to get 'the extra things' | 15:18 |
| Wizzup | check that hildon-meta is installed | 15:20 |
| gnarface | Smatkovi: installed some packages, set one extra variable... it was easy in devuan | 15:20 |
| gnarface | so my limited understand of it is, you basically just install the right cross-compiler packages and then set the CROSS_COMPILE environment variable to the right value to use them | 15:28 |
| Smatkovi | i see, thanks. i want to compile organicmaps from the source repo, but with all dependencies it's about 11GB in size | 15:32 |
| Smatkovi | so i thought i would compile it on the pc in virtualbox until i found out those images are amd64 images | 15:33 |
| gnarface | i've only really done it with kernels and u-boot, but 11GB sounds completely normal | 15:34 |
| dsc_ | oh, topbar was fine, just no accounts active | 15:35 |
| gnarface | Smatkovi: alternately you could boot an armv7 VM in qemu and build it there, but it probably will still take up a lot of space | 15:35 |
| Smatkovi | that's a good idea gnarface | 15:36 |
| Wizzup | Smatkovi: yes, organicmaps is hard to compile, even just on amd64 | 15:39 |
| Wizzup | did you manage to build a recent version? I have one from the start of 2024 or so, working on amd64 | 15:39 |
| Smatkovi | okay maybe i'll try with amd64 first | 15:41 |
| dsc_ | could you enable the 'release' feature for this repo so I can upload a 1.4gb .zip: sanderfoobar/vm-image-gui-dev | 16:51 |
| dsc_ | Wizzup: ^ | 16:51 |
| dsc_ | Wizzup: disregard, forked repos cannot create releases | 17:03 |
| dsc_ | iirc | 17:03 |
| Wizzup | dsc_: I think it would be better to serve large images from maedevu | 17:04 |
| dsc_ | yeah, but this VM is kind of "my hacky thing".. not sure | 17:05 |
| dsc_ | people may have other expectations when its hosted there ^^ | 17:05 |
| sicelo | what's different in this vm? | 17:09 |
| dsc_ | sicelo: https://git.maemo.org/sanderfoobar/vm-image-gui-dev#features | 17:10 |
| dsc_ | it just bootstraps with the things I need (no apparmor / more packages / more swap, cpu/mem) | 17:10 |
| dsc_ | no password SSH login :P | 17:11 |
| sicelo | ah :-) | 17:11 |
| dsc_ | and run all Telepathy related things in debug by default + logs | 17:13 |
| Smatkovi | i used libstdc++-11 and removed the u flag from "set -eo pipefail" in tools/unix/version.sh and now it compiled in the vm without errors | 18:06 |
| Smatkovi | i mean organicmaps | 18:06 |
| Wizzup | smatkovi, that is great! | 18:12 |
| Wizzup | what version is this? | 18:12 |
| Smatkovi | i don't know which version i just cloned the repo from the site you sent a few days ago | 18:15 |
| Wizzup | Smatkovi: can you share which? | 18:22 |
| uvos | freemangordon: dbus activateing ofono wont help you | 19:13 |
| uvos | since it could be sphone rather than ring that wakes ofono or celluard or so | 19:14 |
| uvos | so it still could start before converstations/tp is ready | 19:14 |
| Smatkovi | i'm on 2g now, but is it enough that it is from the maemo-extras repo you shared? | 19:14 |
| Smatkovi | otherwise i can tell on friday the latest | 19:15 |
| uvos | also if you dbus activate ofono from ring who onlines the modem and how dose it know when | 19:15 |
| uvos | freemangordon: the only solution to use this i can se interface is how it worked before we had conversations and cellulard. In this case sphone did everything, it connected to ofono and then onlined it, ofc knowing that itself was ready to handle incomeing calls/sms on dbus | 19:17 |
| uvos | i would suspect that sfos works the same way | 19:17 |
| uvos | this was still ofc sligtly broken, as nothing handles missed events while sphone was restarting after update or had crashed or whatever | 19:18 |
| sicelo | sphone used to online the modem? | 19:19 |
| uvos | yes original pre maemo sphone onlined the modem | 19:19 |
| uvos | i removed that pretty early (while makeing ofono support a module) but it did | 19:19 |
| sicelo | right. i just never remember that happening on Leste, but as you say, it didn't really do so here | 19:20 |
| uvos | anyhow the only way i can see to use this interface semi correctly is to have one process handle everything | 19:21 |
| uvos | but then i have no idea why ofono bothers with the dbus interface intead of being a lib :P | 19:21 |
| sicelo | this issue would not be solved by being a lib though :-) | 19:24 |
| uvos | sure it would then you would force there to be only one client and ofono would always start with its client | 19:24 |
| uvos | since only one process can open the modem kernel interfaces | 19:25 |
| uvos | only one process gets to use the ofono lib to online the modem :P | 19:25 |
| Smatkovi | but i have to say i compiled the amd64 version, so i will have to try crosscompiling | 19:26 |
| sicelo | mmm, 3G data works for anyone (droid4)? | 19:28 |
| uvos | not since it was shut of in germany | 19:28 |
| uvos | but it did work before that | 19:28 |
| sicelo | at least when i select it from the dialog, there's no activity on ofono's (d)bus | 19:29 |
| uvos | i dont think i ever set the tech preferance | 19:29 |
| uvos | it was just on auto by default and it did choose 3g while that was still available | 19:29 |
| uvos | so could be that the ui for that never worked | 19:30 |
| sicelo | s/3g data/mobile data/ | 19:30 |
| uvos | oh | 19:30 |
| uvos | yeah that worked too f | 19:30 |
| uvos | for sure | 19:31 |
| uvos | but i think while there still was 3g in ger we where still at the state of seting that up manually | 19:31 |
| uvos | so could be that that never worked via icd | 19:31 |
| Smatkovi | https>//git.maemo.org/leste-extras/organicmaps was the repo i used | 19:32 |
| sicelo | it has worked. let me try with raw dbus calls | 19:32 |
| sicelo | mmm, not working with direct calls to ofono either. freemangordon or Wizzup , does it work for you? | 19:38 |
| Wizzup | @smatkovi, yes, that repo does work for amd64, but it's 1-2 years out of date I think | 19:52 |
| Wizzup | 19:13 < uvos> freemangordon: dbus activateing ofono wont help you | 19:53 |
| Wizzup | 19:13 < uvos> since it could be sphone rather than ring that wakes ofono or celluard or so | 19:53 |
| Wizzup | 19:14 < uvos> so it still could start before converstations/tp is ready | 19:53 |
| Wizzup | it's not about activating ofono, it's about setting the modem to online | 19:53 |
| Wizzup | it's pretty simple, don't online the modem in cellulard until we are ready | 19:53 |
| uvos | right, onlineing the modem after sphone and converstaions singal ready (somehow) would mostly hide the issue | 19:54 |
| Wizzup | we get to decide what signal that is: tp-ring being on dbus, h-d ready signal, whatever | 19:54 |
| Wizzup | it wouldn't hide the issue, it would make things work as intended :p | 19:54 |
| Wizzup | yeah, right, ok, I didn't think about conversations and sphone as well | 19:54 |
| Wizzup | but we control when we start those (in xsession) | 19:54 |
| Wizzup | so that's easier than tp-ring autostart | 19:54 |
| Wizzup | but if we assume that conversations causes tp-ring to start, we're mostly there I guess | 19:55 |
| uvos | i say hide as i mean the fact that a call or sms can get permanently lost is not great | 19:55 |
| uvos | the issue here is also that ofono acks the incoemeing signal from the operator | 19:56 |
| uvos | with ensureing that something handled the event | 19:56 |
| uvos | if it dident the operator would try again later | 19:56 |
| uvos | that includes even incomeing emergency messages | 19:56 |
| uvos | so a crash of ofonos client can cause lost messages | 19:57 |
| uvos | while in a well implemented stack that dosent need to be possible | 19:57 |
| uvos | *with out ensureing that something handled the event | 19:57 |
| sicelo | yes, i think we're all in agreement that ofono is at fault for not implementing a cache (or similar) for messages until it is sure they have been processed | 20:00 |
| Wizzup | I mean it seems like a design decision but yeah | 21:10 |
| Wizzup | ofono is kind of real time I suppose: they want to react quickly | 21:10 |
| arno11 | indeed, conversations seems to cause tp-ring to start. at least that's what i see on my n900 | 22:00 |
| arno11 | and it happens very late, like at the end of h-d loading | 22:01 |
| sicelo | arno11: i believe 2G/3G is still working on your N900? | 22:07 |
| arno11 | yes working fine | 22:07 |
| sicelo | sorry, meant gprs | 22:07 |
| arno11 | gprs is ok as well | 22:08 |
| arno11 | at least the last time i tried | 22:08 |
| sicelo | ah, another SIM works ... so something's up with my first sim. will investigate on N900 | 22:13 |
| arno11 | (will check gprs now) | 22:15 |
| arno11 | sicelo: yeah gprs ok, currently using it :P | 22:19 |
| sicelo | yeah, it was on the Droid 4 where i wasn't sure, but have found that a different SIM does work | 22:19 |
| arno11 | ah ok | 22:20 |
| arno11 | currently testing telegram in 2g | 22:21 |
| arno11 | *gprs | 22:22 |
| arno11 | works fine | 22:24 |
| arno11 | funny that a so old network is usable with modern messaging | 22:34 |
| Wizzup | I did a call test on the pinephone daedalus and it works | 23:22 |
| Wizzup | cool | 23:22 |
| arno11 | nice | 23:42 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!
