<sicelo>
i see a few regressions in daedalus, at least on D4
<sicelo>
(1) the keyboard layout seems to be missing some important customizations, e.g. OK+P no longer produces the pipe (|) symbol
<sicelo>
(2) rotation doesn't work anymore
<sicelo>
i think both are due to upstream packages ...
<sicelo>
for (1), most likely xkb-data. for (2), i have indeed seen that iio-sensor-proxy does not start automatically since there is only a systemd init script
<sicelo>
wonder how Devuan intends for that to get started
<sicelo>
i'm thinking that we could ship these things (iio-sp openrc script, xkb layout, etc.) using leste-config or similar. because yes, there are some nice benefits to using upstream packages as much as possible
<sicelo>
of course for iio-sp, i do request a fork just so we can get the latest and greatest features
<gnarface>
sicelo: see if the missing init script is in the orphan-sysvinit-scripts package, and if it's not report it to fsmithred
<gnarface>
(debian has been killing them off one by one, they just add them all to that package)
<gnarface>
temporarily you can just probably copy the one from chimaera for a quick fix
<gnarface>
it might have already been noticed and put in the orphan-sysvinit-scripts package though
<gnarface>
unless it's something that is intended to get started by the session manager or something weird like that now - there's precedent for that, like with pulseaudio
<gnarface>
yea, just report it to them
<gnarface>
it's a known ongoing issue
<sicelo>
anyway, we will fork iio-sp for now, for the reason i mentioned above, so we can ship the init script without issues (we used to in the past as well). maybe by the time we switch versions, Leste will be using systemd ;-)
<gnarface>
fork the whole distro for all i care, but i think you're being foolish and spiteful
<sicelo>
uh ?
<Wizzup>
sicelo: pipe works for me on d4
<sicelo>
OK+p ?
<sicelo>
weird. doesn't work here. maybe you upgraded to daedalus? i used the image
<Wizzup>
yeah, but iirc I did rebase
<Wizzup>
will look
<sicelo>
i think image got mostly upstream packages (which is a good thing in most cases)
<sicelo>
so i guess on your D4, autorotation is still working, right?
<Wizzup>
yes
<sicelo>
yes. so the previous configs were still there
<sicelo>
on new image it isn't working. that said, if you start iio-sp manually, then it does work
<sicelo>
so anyway, my main question/aim for discussion was - how to fix these? is shipping the customized stuff via leste-config-* fine, or would that have gotchas?
<sicelo>
e.g. the D4 layout ... instead of forking xkb-data like we did in the past, just add it in correct place using leste-config-mapphone or whatever it's called?
<sicelo>
gnarface: apologies. i want an iio-sp fork because i have worked with upstream iio-sp to add some features with the intention of using them on Leste (supporting N900's proximity sensor; reporting face-down and face-up orientations, which Leste can use to silence alarms, etc.)
<gnarface>
sicelo: can you confirm the last time the actual init script was seen? it's apparently not showing up in chimaera either. if it ever actually had one they'll want put it in that package regardless of whether you'll use it
<sicelo>
as mentioned, we had it in the forked iio-sp. as for Devuan/Debian, i don't know if they ever had it
<gnarface>
hmmm... well which debian version did you fork that from?
<sicelo>
i think the script was manually added by the dev who packaged it for Leste, i.e. it was not from debian
<gnarface>
maybe so, it wasn't in bewoulf either apparently
<sicelo>
Wizzup: i must finish up the upower work this weekend. seems to be working perfectly fine for me so far on N900 & D4. plus the upower maintainer seems to be accepting of my patch .. just needs tests
<freemangordon>
sicelo: is it possible that iio-sensor-proxy starts by dbus activation?
pere has joined #maemo-leste
<Wizzup>
sicelo: sounds great @ upower
<Wizzup>
sicelo: I prefer forking even for missing init scripts, but if devuan has a solution that contains it, I am ok with it too
<Wizzup>
prefer either to having it in leste-config
<Wizzup>
we could put it in hildon-initscripts but meh
<arno11>
indeed iio-sp is deactivated on my daedalus too
<gnarface>
the word from devuan is this package never seems to have had an init script, at least not since jessie... and i can't get debian's package search page to talk to me about anything older than beowulf
<Wizzup>
yes, we added it in our fork
<Wizzup>
we're just trying to not have to fork it
<Wizzup>
so we can get it added to this devuan pkg that gnarface mentioned, or just fork iio and get the init script upstreamed
<gnarface>
i assume we can get older versions from the archive, but there's no evidence anything earlier would have it anyway, right?
<gnarface>
it does beg the question: how was it supposed to start?
<gnarface>
but i don't think it would be the first debian package i've encountered that just never had one
<gnarface>
though occasionally you find an example one in /usr/share/doc
<gnarface>
that doesn't seem to be the case here
<freemangordon>
gnarface: as it is a dbus service, maybe the idea was that dbus starts it
<freemangordon>
unfortunately I have to run now, so maybe somebody else can confirm
<freemangordon>
if not, I can check tomorrow
<sicelo>
freemangordon: no, iio-s-p doesn't start via dbus-activation. if it did, then rotation would work. fwiw, upower has same init script 'problem' but it does support dbus-activation, hence it works
<sicelo>
Wizzup: I understand @forking
<sicelo>
gnarface: upstream ships a systemd service. hope that answers your 'how was it supposed to start?' question.
<gnarface>
hmm, so maybe debian wheezy had it?
<gnarface>
systemd hadn't been in the distro the whole time, had it?
<gnarface>
i thought they only added it in squeeze
<gnarface>
or was that just when they made it the default?
<gnarface>
to the best of my recollection, most packages didn't just immediately delete the init script way back in jessie, i don't recall noticing init scripts getting assassinated until much more recently
<Wizzup>
gnarface: we have the init script in our fork, so surfacing it is not the problem I think
<Wizzup>
sicelo: we never shipped an init script for upower btw
<sicelo>
gnarface: no idea. Anyway, we're definitely forking iio-s-p for the daedalus cycle at least (primarily for the reasons I mentioned). actually I guess adding it in the orphans package wouldn't help us for daedalus anyway ... maybe for excalibur
<sicelo>
yes @upower ... that one starts via dbus activation
<sicelo>
with Trixie already in early stages of its freeze cycle, probably too late for iio-s-p orphan script too
<gnarface>
hmm, package doesn't appear in wheezy or squeeze, so i guess it really never had an init script
<gnarface>
i really wonder what that means but i guess it's more academic curiosity, sorry to have added unnecessary drama to this
vectis_ has quit [Read error: Connection reset by peer]
<Wizzup>
these rules only activate a systemd service
<sicelo>
no, the udev rules don't start it. they just add tas on the sensor, so when iio-sp eventually starts it just looks for the tags to find the sensor
<freemangordon>
but, the way those are written (and judging on the git history), it was never meant to activate anything but a systemd service
<sicelo>
making it dbus activatable is actually trivial.
<sicelo>
and i think dbus activation is upstreamable vs. an openrc service
<freemangordon>
mhm
apac has joined #maemo-leste
<sicelo>
h-d gets orientation information from mce?
<sicelo>
asking because with iio-sp being dbus-activated, hd doesn't autorotate. dbus shows the service, with "activatable" status. so whatever that h-d is using does not seem to try hard enough to poke at the dbus service
<sicelo>
if i do something like `busctl tree net.hadess.SensorProxy` it then gets activated and autorotation works. need to find how to fix whoever informs h-d about orientation
<Wizzup>
I think we just listen for signals
<Wizzup>
that might not activate it
<sicelo>
right, yes
_fab has quit [Ping timeout: 265 seconds]
arno11 has left #maemo-leste [#maemo-leste]
apac has quit [Ping timeout: 244 seconds]
akossh has quit [Ping timeout: 265 seconds]
lyubov has quit [Read error: Connection reset by peer]