atk has quit [Quit: Well this is unexpected.]
atk has joined #maemo-leste
Pali has quit [Ping timeout: 240 seconds]
kvw_5_ has joined #maemo-leste
kvw_5 has quit [Ping timeout: 265 seconds]
diejuse1 has quit [Quit: Leaving.]
Oksana has joined #maemo-leste
belcher_ has joined #maemo-leste
belcher has quit [Ping timeout: 246 seconds]
tschakram_N900 has joined #maemo-leste
<tschakram_N900> Hello, great work with maemo leste.
<enyc> =)
tschakram_N900 has quit [Quit: Connection closed]
tschakram_N900 has joined #maemo-leste
<tschakram_N900> Can I asked some questions about respositorys? or did someone have some hits for me for usage of N900 ? :-)
tschakram_N900 has quit [Quit: Connection closed]
pere has quit [Ping timeout: 240 seconds]
tschakram_N900 has joined #maemo-leste
pere has joined #maemo-leste
tschakram_N900 has quit [Ping timeout: 240 seconds]
cr4y1 has joined #maemo-leste
tschakram_N900 has joined #maemo-leste
Pali has joined #maemo-leste
tschakram_N900 has quit [Client Quit]
tschakram_N900 has joined #maemo-leste
tschakram_N900 has quit [Quit: Connection closed]
<sicelo> Ask away:-)
diejuse1 has joined #maemo-leste
belcher_ is now known as belcher
diejuse1 has quit [Read error: Connection reset by peer]
diejuse1 has joined #maemo-leste
uvos has joined #maemo-leste
<Wizzup> is anyone else getting this on the d4?
<Wizzup> May 10 10:53:31 localhost sshd[19204]: User user not allowed because account is locked
<parazyd> Set a password
<Wizzup> > Both "!" and "!!" being present in the password field mean an account is locked.
<Wizzup> Do we want to user account to be locked?
<parazyd> Wizzup: It means that you have no password set
<parazyd> This is the way it works
<Wizzup> no
<Wizzup> that is *
<Wizzup> >By some research on internet and through this thread, I can understand that * means password never established, ! means locked.
<Wizzup> If I change the shadow entry to use * instead of !, it does alow me to log in with pubkey set
<parazyd> Yes
<parazyd> But I mean if you set a password it should get unlocked.
<Entitlement> parazyd - [ hildon-base/postinst at master · maemo-leste/hildon-base · GitHub ]
<Wizzup> What do we set it to by default?
<parazyd> See link
<Wizzup> I see the link and I cannot parse
<parazyd> echo "user:user" | chpasswd
<Wizzup> so the password is user?
<parazyd> Yes
<Wizzup> Weird that my image didn't have that set, then
<parazyd> Beucase you have an old image?
<Wizzup> maybe, not so sure it is that old
<parazyd> I did this four months ago
<parazyd> Jan 2021
<Entitlement> parazyd - [ Set a password for user "user". · maemo-leste/hildon-base@b596ffc · GitHub ]
<Wizzup> right, and it only runs if the user account doesn't exist yet, so the upgrade isn't rolled out
<Wizzup> (regardless of image age)
<parazyd> Yeah, because if the user exists already, people are supposed to set their own password.
<parazyd> Why would we forcefully set/change passwords for existing users?
<Wizzup> Where do we document that? I don't see a point to setting a password for the phone user really
<Wizzup> (just trying to understand)
<Wizzup> could we not grep for 'user:!:' or something?
<Entitlement> parazyd - [ Getting Started - Maemo Leste Wiki ]
<parazyd> No I don't agree with changing this.
<parazyd> It's insane to change existing user passwords
<uvos> im with parazyd here
<Wizzup> If a user is locked, I don't think they have a password set?
<uvos> what we need to do is just have a app that runs the first time and forces the user to set a pw
<Wizzup> I mean I changed it from user:!: to user:*: because I don't want a password
<Wizzup> sudo is already passwordless and I use pubkeys for ssh, so what's the point of a password
<parazyd> uvos: In images since Jan 17 the password is set.
<uvos> parazyd: sure but its mutch user frednlier to have a pw the user selects
<parazyd> Wizzup: That sudo argument is bad, because why do we then maintain all the .sudoers files
<uvos> Wizzup: sudo should not be passwordless
<uvos> Wizzup: its just for convieniance right now
<parazyd> Wizzup: Yes, and if you don't want a password that's exactly what you do.
<uvos> Wizzup: eventually we need to ship askpass + normal sudo
<Wizzup> well, I disagree, but I won't argue since it looks like it's decided already
<Wizzup> I think ending up in a situation where the main account is locked is not sensible
<Wizzup> but it's fine, I can work on what I actually wanted to work on now
<Wizzup> :P
<parazyd> But you're in a situation where root login is enabled as well
<parazyd> Nobody will hit a complete lock
diejuse1 has quit [Quit: Leaving.]
<Wizzup> I don't tend to use root login
<parazyd> Because either you have a user password, or you have root login enabled
<parazyd> This is the switch that's been done in January
<parazyd> (Where we also stopped maintaining our own sshd_config)
<parazyd> 11:11 <Wizzup> I don't tend to use root login
<parazyd> How did you discover this only now then? :p
<Wizzup> all of that is fine, I just think that it's weird to have the main account be registered as locked because my image predates that commit
<Wizzup> parazyd: for devel I don't, since I end up running things as root and that breaks
<uvos> Wizzup: witch is another problem
<Wizzup> uvos: what exactly?
<uvos> Wizzup: we need to eventually make our stuff work with mulitple users
<Wizzup> well I don't mean that
<parazyd> ++
<uvos> lots off stuff assumes one global user
<Wizzup> I mean that if half the system is running as one user, running it as another is not a good idea
<uvos> ok, i dont know why that would happen if your just using the root account to change your password
Dg-2 has quit [Ping timeout: 240 seconds]
<Wizzup> well, I didn't say that
<Wizzup> I didn't set a password
<Wizzup> I edited the shadow file and replaced '!' with '*'
<Wizzup> and now everything works as expected
<uvos> sure same thing more or less
<Wizzup> because '*' means no password set, whereas '!' means account locked
<uvos> yeah i know
<uvos> so i think we should ship user account locked and have a small app that runs under x as root before any of the maemo stuff starts
<uvos> and asks the user what do do about the normal account (name and pw)
<parazyd> uvos: Eventually we'll want the Fremantle welcome dialog
<parazyd> uvos: That's where we can implement this if necessary
<uvos> parazyd: right something akin to that
<uvos> untill then i dont see why we cant just document that people need to set a up the "user" user on first boot by hand
<parazyd> But it's set up
<uvos> but not with the users password
<uvos> *prefered
<parazyd> It is
<parazyd> ah right
<parazyd> It's just user:user, which is documented
<uvos> we get lots of questions on here with what is the default pw all the time
<uvos> right
Entitlement has quit [Ping timeout: 265 seconds]
<uvos> to bad we dont have any default browser or something like that
diejuse1 has joined #maemo-leste
<uvos> i think we could improve the expierance greatly be simply showing a webpage with some info on first boot
<parazyd> Really not sure what to pick though
<uvos> main problem is imo that what is sensible for n900 is not sensible for any of the other devices really
<parazyd> I think it's more the other way around
<uvos> hmm?
<parazyd> N900 is the low-power, slow device. So we can't expect something like Firefox which works relatively fine on mapphone to work there.
<uvos> right
<uvos> thats what i ment
<uvos> also firefox works fine on d4 only really
<uvos> since we need real him integration elsewhere
Entitlement has joined #maemo-leste
<Entitlement> parazyd - [ Getting Started - Maemo Leste Wiki ]
M75ch4kr4aa[m] has joined #maemo-leste
<M75ch4kr4aa[m]> Hello
<Wizzup> hi
diejuse1 has quit [Quit: Leaving.]
<Entitlement> uvos - [ png (960 x 540) ]
<uvos> ^^^ uses qt to render a html document on first boot
<parazyd> hah cute
<parazyd> There's a ticket about this too, made by Pavel IIRC
<Wizzup> uvos: nice, fun :p
<Entitlement> parazyd - [ Create some kind of "release notes" / "welcome application"? · Issue #229 · maem... ]
<Entitlement> uvos - [ GitHub - IMbackK/greeter: greeter app for mameo leste ]
<uvos> for the first boot part i just created a self deleting script in xsession
<uvos> ill package it some other time
<parazyd> We have some firstboot initscript
<parazyd> I don't recall from which package it comes atm
<uvos> i dont think gui applications are apropriate in an initscript
<uvos> it needs to be run by the session manager not openrc
<parazyd> I meant it could be set up using the initscript
<parazyd> Like, copy the xsession in place
<uvos> sure that would be fine
<Wizzup> maybe it can an icon on the desktop, or a widget, like in fremantle
diejuse1 has joined #maemo-leste
<Wizzup> uvos: parazyd: here's a thought I had: what if we allow the location status applet to keep location on if there are no clients, in some special mode
<Wizzup> that way we could have libgps clients work if the user has location-status do that
<Wizzup> (ofc that is assuming we stop gpsd once the last liblocation client is gone)
<parazyd> We aren't stopping gpsd once it's started.
<diejuse1> Hello, I am checking start commands to find out why tapping over "Internet Connection" is not responding. I have seen that I wasn't starting "hildon-sv-notification-daemon". How do I launch it? I get "segmentation fault".
<Wizzup> which I think makes sense - what if someone turns off location services and gpsd is still running?
<uvos> wo also dont have to
<Wizzup> parazyd: I know we're not atm ^
<uvos> Wizzup: why would we want to stop gpsd
<uvos> Wizzup: execpt if the user turns off gps in settings?
<Wizzup> cpa->location->gps->disable
<parazyd> That button can have a callback to stop it
<Wizzup> that feels incredibly hacky to me
<Wizzup> but ymmv
<parazyd> Why?
<Wizzup> diejuse1: well, are all the required services running?
<parazyd> To me it's a button that says enable/disable GPS
<parazyd> And logic says it should
<Wizzup> parazyd: so we'd do what, pkill gpsd as sudo from a button in cpa?
<uvos> this seams reasonable to me to
<uvos> Wizzup: it would ask openrc to stop it
<parazyd> OpenRC yeah
<diejuse1> Wizzup: I don't know what are the required services
<Wizzup> uvos: how about setting the gconf key to disabled and having location-daemon figure that it should no longer relay info and stop gpsd?
<Wizzup> since it's all the one to start gpsd
<Wizzup> It could just monitor for the gconf key change
<uvos> Wizzup: i would avoid using gconf as ipc when possible
<uvos> btw
<parazyd> What if location-daemon isn't running?
<parazyd> And gpsd is
<Wizzup> parazyd: then gpsd won't be running, per my logic
<Wizzup> but not in the current situation
<parazyd> It's a bit complicated
diejuse1 has quit [Quit: Leaving.]
<Wizzup> I think it's weird that location-daemon starts on demand, but gpsd does not
<Entitlement> parazyd - [ pkg-gpsd/gpsd.init at master · maemo-leste-upstream-forks/pkg-gpsd · GitHub ]
<Wizzup> I would expect both or neither
<parazyd> gpsd starts on demand too
<parazyd> By location-daemon
<Wizzup> only the first time, then it stays on
<parazyd> Yes
<uvos> Wizzup: we could start gpsd as soon as something uses its soccet
<parazyd> And this is fine as long as you have GPS-enabled in gconf
<Wizzup> well, that's not what I mean by on demand
<uvos> Wizzup: like location-deamon or anything else
<uvos> Wizzup: idk how to do this with openrc but systemd can do this
<uvos> maybe you cant with orc
<parazyd> uvos: xinetd probably
<parazyd> Wizzup: This problem is why I initally wanted to have the control panel control this behaviour
<Wizzup> uvos: I don't want gpsd to start unless location-daemon wants it to start, essentially
<uvos> Wizzup: why
<uvos> Wizzup: i want navit to just work...
<parazyd> gpsd is disabled in runlevel
<Wizzup> becaus them we have to patch gpsd to read the gconf val
<uvos> Wizzup: no
<parazyd> And the initscript checks gconf
<uvos> Wizzup: so if this where systemd i would do this
<parazyd> If you want to start gpsd with a root shell, you're on your own
<parazyd> Covering these edge-cases is counterproductive
<uvos> Wizzup: wirte a service.soccet that runs gpsd if something uses its soccet but only if gconf dosent have gps disabled
<Wizzup> parazyd: I am responding to uvos saying that we should use socket activation on the gpsd socket
<Wizzup> that is not 'root shell'
<parazyd> I don't like that either
<parazyd> You could simply add your gpsd to a runlevel if you want it autostarted
<Wizzup> I think the edge cases are important fwiw
<parazyd> It has zero cost to be running with no clients
<Wizzup> but I think I'll go for a walk or something, give it some more runs
<uvos> parazyd: some ram
<Wizzup> I've already seen the gpsd starting fail in various a few cases
<Wizzup> openrc forgot gpsd was running
<Wizzup> but it was running
<Wizzup> and location-daemon failed to start (I fixed that in master)
<Wizzup> but still it won't help if we want to stop gpsd and openrc forgets about it, however it happens
<parazyd> Then let's port the initscript to supervise-daemon rather than start-stop-daemon
<parazyd> And we're good
<parazyd> This way openrc can't "forget" about it
<Wizzup> I really don'r see why we need to keep gpsd running once we no longer need it, but I'll go for a walk and try to see your side of it
<uvos> i also dont see the point of having it run after location-deamon exits and not starting it on boot
pagurus has quit [Ping timeout: 268 seconds]
<uvos> either it should allways run
<uvos> or only with lcocation-deamon
<uvos> or soccet activation
<Wizzup> yes, that's what I meant
<uvos> but really i want gpsd only clients to work
<uvos> so allways running gpsd or soccet activation is the only thing that would work for me
<parazyd> gpsd caches useful data, like skyview
<parazyd> If you stop it, you often slow down getting a future fix
<uvos> parazyd: ok thats fine
<uvos> parazyd: but why not start it at boot then?
<parazyd> Because the disclaimer has to be accepted.
<uvos> the init scrpt checks the disclaimer
<parazyd> So you only raise this with a liblocation client
<parazyd> Correct
<uvos> so the current behavior could be before the first gpsd client is run and sets it
<uvos> and then on every other boot gpsd just runs
<Wizzup> did you benchmark if keeping gpsd on helps? just wondering
diejuse1 has joined #maemo-leste
<Wizzup> (I know this is all pretty hard to benchmark)
<uvos> (or runs with the first client via soccet activation)
<parazyd> Wizzup: Not specifically, but I get a (maybe placebo) faster fix with subsequent cgps runs
<parazyd> uvos: Well, the initscript will simply fail if the disclaimer is not accepted or gps is disabled.
<parazyd> Which can also be fine, and keep it in the default runlevel.
<parazyd> Anyway, to sum up my thoughts:
<parazyd> I think gpsd should always be running as long as GPS is enabled in control panel, and should be disallowed to run if it's disabled in control panel
<parazyd> disallowed by initscript
<parazyd> If someone wants to start it as root in a shell, this is a case we can't cover.
<parazyd> And we shouldn't IMHO, it would be feature creep and platform lockdown
<uvos> trying to controll what people do in root shell is silly anyhow
<Wizzup> (jfyi I wasn't talking about root shell anywhere)
<parazyd> Wizzup: I'm using it as a metaphor for user starting gpsd outside of the environment/ecosystem
<Wizzup> sure, agreed
<parazyd> Further, personally I see nothing wrong with the control panel applet button + Save would stop gpsd via initscript when pressed/saved.
<parazyd> stop when GPS is chosen to be disabled I mean. It shouldn't be starting anything.
<parazyd> This is basically the only interface the user has to "control" GPS, and I think one would expect it to stop and disable after the user chooses to.,
<uvos> i would be fine with this, i have no strong optinon except that plain gpsd apps must work with no additional work.
<parazyd> Maybe ask some of your friends, what they would expect when this button is pressed.
<uvos> *opinion
<parazyd> The three of us just thinking about our cases isn't often the correct solution :p
<diejuse1> I am trying to start hildon-sv-notification-daemon like "/usr/sbin/dsmetool -t /usr/bin/hildon-sv-notification-daemon" but I get "dsmesock_connect: connection refused". Any orientation?
<parazyd> You should start dsme first
<parazyd> There's two initscripts
<parazyd> /etc/init.d/dsme-dbus and /etc/init.d/dsme
<parazyd> Also see their dependencies
<diejuse1> they are both already started: "WARNING: dsme... is already starting"
<diejuse1> maybe you could give me an ordered list of starting commands to boot normally? Because I think that is the problem.
<parazyd> Run the commands inside manually
<Wizzup> diejuse1: try to figure out how to use openrc and the x session scripts from within the chroot
<Wizzup> that's likely the only way to really do thi
<Wizzup> s
<Wizzup> iirc openrc can run in a chroot
<Wizzup> parazyd: if you keep it always on, how is it started on boot?
<diejuse1> mmmm
<Entitlement> Wizzup - [ OpenRC - Gentoo Wiki ]
<diejuse1> any command for openrc?
<diejuse1> ok
<Wizzup> something like that - with the config like that
<Wizzup> not sure if it's exactly like that, we're not gentoo ofc, but I think a path like that is the proper solution for you
<Wizzup> because managing starting processes manually is likely very painful
l_bratch has joined #maemo-leste
<Entitlement> parazyd - [ pkg-gpsd/gpsd.postinst at master · maemo-leste-upstream-forks/pkg-gpsd · GitHub ]
<parazyd> Wizzup: Until disclaimer is accepted, it would just not start.
<parazyd> Same goes for having GPS disabled.
diejuse1 has quit [Quit: Leaving.]
<Wizzup> parazyd: so you want to also toggle if it's in the runlevel based on the button press / gconf change?
<parazyd> No, the only thing the button press + save (important!) should do is rc-service gpsd stop
<parazyd> Let's leave gconf change listening/hooks out of this.
<parazyd> Actually it's not button press, but rather toggling it and saving. If GPS is the disabled, stop gpsd via initscript.
<Entitlement> parazyd - [ location-control/liblocation_applet.c at master · maemo-leste/location-control ·... ]
<Entitlement> parazyd - [ location-control/liblocation_applet.c at master · maemo-leste/location-control ·... ]
<parazyd> So in one of these two places it fits.
pere has quit [Ping timeout: 240 seconds]
<sicelo> What is this "button" being referred to?
<parazyd> sicelo: The checkbutton "GPS Enabled" in ControlPanel->Location
<sicelo> Ok
diejuse1 has joined #maemo-leste
<Wizzup> parazyd: so if I check it again, is gpsd re-added to the runlevel? idgi
<Wizzup> or do you want to attempt to start it and fail on boot?
<Wizzup> btw, I think I can live with gpsd running independently of liblocation/location-daemon, but I think it provides a very suboptimal ux, since you cannot see when gps is active if it is through libgps
<Wizzup> effectively nuking all control and visibility
<Wizzup> my solution with then status applet to control gpsd when there are libgps but no liblocation clients I think is more clear
<Wizzup> and it would provide insights through location-daemon as normal as well (wrt fix, blink or not)
<Wizzup> so if the location status button is always visible and has a UI to say 'keep gps on' and 'open control panel' I think would be a fair middleground
sunshavi has quit [Read error: Connection reset by peer]
<Wizzup> so the 'keep gps on' button would effectively turn the status applet into a normal location-daemon client, keeping gpsd alive
Dgby714 has joined #maemo-leste
<Wizzup> it should work for uvos his usecase
<Wizzup> if it is just about the work required, I am happy and motivated to do it
<Wizzup> btw, gpsd and modem are happy together for a bit now, so the reset seems random
<uvos> Wizzup: mutch better would be for the location status icon to monitor the gpsd soccect and pop up if its in use
<uvos> then the user dosent need to care about if an app uses liblocation or gpsd
<uvos> and it all just works
<Wizzup> I do not think gpsd can do that
<Wizzup> unless you wamt to ptrace gpsd
<uvos> Wizzup: you can check with the kernel if the soccet has connected users
<Wizzup> feels like a hack tbh
<uvos> this is true
pere has joined #maemo-leste
<uvos> but gpsd flok wernt very open to the idea of adding something better
<Wizzup> liblocation and the location-daemon dbus signals provide useful info to maemo components, usiong that path seems better
<Wizzup> yeah, gpsd folk are not helpful
<uvos> sure we should use that path aswell
<uvos> but the logic could be something like: check liblocation status and apply unless gpsd has a client thats not location-deamon, then display the icon
<Wizzup> like I said, I think a simple way to keep loc-daemon and gpsd on for libgps clients makes sense, and then if gpsd *stops* with loc-daemon we also have possibility of no backgrou nd libgps clients
<Wizzup> so the UI is leading and always source of truth
<Wizzup> but still allows for easy libgps clients
<Wizzup> click status, click location applet, click 'stay on', done
<uvos> not really you have to think about if a client you want to open is libgps or liblocation
<uvos> thats terrible ux
<sicelo> Not really
<Wizzup> mhm
<uvos> the user should not need to know about implmentation details
<Wizzup> I agree about that, that is not ideal
diejuse1 has quit [Ping timeout: 240 seconds]
random_yanek has quit [Ping timeout: 260 seconds]
<sicelo> There's a reason why stuff like geoclue was invented and is in use
<Wizzup> but I think it is not too bad compared to the alternatives
<sicelo> Basically it's what libloc is
<Wizzup> sicelo: can you elaborate?
<uvos> geoclue is gnome tech
<uvos> for location
<Wizzup> ok
<uvos> dose the same stuff as location-deamon
<uvos> just more and better
<uvos> (no offense)
<Wizzup> how does it manage gpsd<
<Wizzup> ?
<uvos> good question
<parazyd> 14:55 <Wizzup> parazyd: so if I check it again, is gpsd re-added to the runlevel? idgi
<parazyd> Forget the runlevels
<parazyd> Just consider gpsd initscript is enabled and in the default runlevel
<parazyd> The gpsd binary wouldn't start until the disclaimer is accepted + GPS is enabled.
<Wizzup> what state will openrc think gpsd is in? start requested but failed to start?
<parazyd> "Stopped"
<parazyd> Obviously
<Wizzup> I think the states are more complicated than that
<uvos> Wizzup: is like a deamon that exits sucessufly
<Wizzup> parazyd: so how do you think we should handle libgps clients wanting to use gps, and how do we show it in the UI?
<parazyd> That's the next step.
<parazyd> Let's solve this.
random_yanek has joined #maemo-leste
<Wizzup> ok
<Wizzup> please also look at my proposal above wrt that
<parazyd> I don't like it because I don't think the status applet should be any kind of control for this.
<parazyd> We have the control panel.
<parazyd> It's called "control" for a reason.
<parazyd> Easy enough to expand on that.
<parazyd> e.g. suboptions
<Wizzup> well, it's only control is that it asks for gps data
<Wizzup> just like 'maep' controls it
<Wizzup> no more, no less
<Wizzup> status applets also switch wifi and profiles
<Wizzup> those are not cpa
<parazyd> r3ok
<parazyd> ok
<parazyd> I can't win this argument really
<parazyd> So I'll move on to other stuff, I'm sorry
<parazyd> Got too much to do
<parazyd> Nothing personal, it just starts becoming a bikeshed
<uvos> Wizzup: parazyd: how about we just delgate it to parazyd to do what he thinks is best?
<uvos> since he implmented most of this
<Wizzup> parazyd: ok, I'll chdeck with fmg, see what he thinks?
<parazyd> And I can't really discuss UX with just the three of us
<parazyd> Please ask physical people around you
<parazyd> Get insights
<parazyd> It's not just the three of us
<parazyd> Show the UI and ask "what do you expect this and that does"
<parazyd> Let's then come to a conclusion
<uvos> ill just abstain from here on out.
<Wizzup> I will try to phrase some questions
<Wizzup> it would be good to have alternatives to handling libgps and ux though
<parazyd> What does "alternatives" mean?
<Wizzup> How we would handle a libgps client from Maemo perspective: (1) gpsd just runs and user in unaware (2) location-status does hacks to determine if gpsd has clients and acts appropriately (using libgps backend) (3) clients require gps+gpsd to be enabled specifically, ui implicit (4) ?
<uvos> Wizzup: your mixing 2 problems anyhwo
<Wizzup> I think some of questions boil down to how we want to handle that
<Wizzup> uvos: care to split it out further?
<parazyd> Wizzup: We started this discussion in whether gpsd should autostart from init or not, and who should stop it.
<parazyd> Wizzup: Having userspace/GUI reporting of gpsd clients is a different problem
<uvos> Wizzup: problem 1. when to start gpsd (we must enforce "gps enabled" and "disclamer accepted" in all cases, even malicious clients) 2. how to show gpsd is in use form a source other than location-deamon
<uvos> problem 1 we can deal with by only starting gpsd when those are accepted by the user and setting permissions on gnss0 adequatly.
<uvos> those are the problems
<Wizzup> parazyd: ok, the only solutions I could come up with combine them, but if you have other solutions to the UI?
<uvos> id just lett parazyd find solutions he likes
<Wizzup> uvos: problem 1 is more complicated
<Wizzup> the disclaimer does not relate to location being on whe nit should not
<uvos> pardon?
<uvos> when it should not
<uvos> ok
<Entitlement> parazyd - [ pkg-gpsd/gpsd.init at master · maemo-leste-upstream-forks/pkg-gpsd · GitHub ]
<uvos> um yes it dose, the disclamer not being accepted should block gps
<uvos> form all sources
<Wizzup> uvos: other way around
<uvos> libgps liblocation and open() on /dev/gnss0
<Wizzup> the disclaimer being accepted doesn't imply location should be on
<Wizzup> I will go for a bike ride and type from my laptop a more verbose summary
<Entitlement> parazyd - [ pkg-gpsd/gpsd.init at master · maemo-leste-upstream-forks/pkg-gpsd · GitHub ]
<Wizzup> I am on my d4
<uvos> right the initscript works perfectly fine as is
<uvos> question is just when to run it
<uvos> and ill defer to parazyd on that
<parazyd> My proposal is to have it always running as long as privacy is accepted, and gps is enabled in settings
<uvos> which is fine by me
<parazyd> This same place (where you change the GPS enabled/disabled setting) should then STOP GPSD when the user chooses to disabled GPS and save the setting.
<uvos> and how its implmened that ill also defer to parazyd
<parazyd> Via initscript
<parazyd> That's all.
<parazyd> There is no more to problem 1.
<parazyd> If anyone thinks it's more complicated than this, then something's not right.
<uvos> parazyd: /dev/gnss0 is world readable
<uvos> parazyd: atm
<uvos> parazyd: we need to enforce that gpsd is the only thing that can read it
<uvos> parazyd: but thats details
<parazyd> Easy, initscripts often fix permissions.
<uvos> we can run gpsd in some group
<uvos> and assing gnss0 via udev
<parazyd> gpsd really needs root, as far as I've been told by the devs. It then drops privileges asap.
<uvos> ok
<uvos> can we have it drop privlages to a gpsd user?
<uvos> and just have /dev/gnss0 owend by that?
<parazyd> "Otherwise various permission bugs happen"
<uvos> or dose it open() as root?
<uvos> ok
<parazyd> It does root and group dialout
<uvos> ok
<parazyd> So we can group own gnss0 as dialout
<uvos> right
<uvos> and remove o+r
<parazyd> Yes
<uvos> sounds good
<Wizzup> hi
<Wizzup> I think the one of the questions we should answer to move forward is: how do we think we should handle libgps clients when liblocation is not active? I think we seem to have different opinions on that specifically
<Wizzup> Regardless of init script, what runs when, I think being able to answer the above questions help with figuring out how to continue
<parazyd> They should be able to use gpsd if GPS is enabled in gconf and gpsd is running.
<Wizzup> So the user would not be able to tell from the UI, corret?
<parazyd> Until we implement something in location-status, yes.
<parazyd> My proposal currently is this: https://parazyd.org/pub/dev/random/gps-complication.txt
<Wizzup> I'm trying to not let the technical parts (what is running when) mix into the answer
<parazyd> And this is not related to location-status in any way
<parazyd> I don't think anything else makes sense when it comes to user's control over gpsd
<Wizzup> I think that personally, I would prefer that libgps does *not* work, unless the user knows that he/she started a client that uses gps. And if that means the applications don't work unless gps is enabled (through one way or another, not just a disclaimer), then I think that is entirely acceptable, what I think is not acceptable is clients using gpsd without the user knowing
<parazyd> As for location-status telling you when there are libgps clients, that's an entirely different problem.
<uvos> Wizzup: from a birds eye view yes libgps should work and yes the user should be informed when it is used
<uvos> Wizzup: but for now we can only do 1 and we will do 2 later
<Wizzup> I want to discuss both, since I don't think we can sensibly do (2) if we do a specific (1) now
<parazyd> Yes we can
<uvos> here im in the just let parazyd do what he wants camp
<Wizzup> uvos: I think that's a non-argument and also parazyd asked me to do some work on this
<parazyd> (2) is a completely different thing
<uvos> Wizzup: yes it is explicetly a non-argument :)
<uvos> i lack the detailed knowlage that parazyd has on this subject to even think about trying to tell him what is better
<parazyd> For me 3 hours of this talk today is enough.
<Wizzup> parazyd: ok, for (2) how will you let users know libgps clients exist?
<uvos> Wizzup: for this there are several solutions that need investiagtion
<uvos> Wizzup: we can check what gnome/kde do
<uvos> Wizzup: we can try to submitt some patch to gpsd
<sicelo> Geoclue
<uvos> sicelo: i know
<Wizzup> uvos: ok, so we should do more research before continuing?
<uvos> Wizzup: we can patch libgps
<Wizzup> uvos: one of solutions being 'gpsd does not run'
<parazyd> Wizzup: First thought is to watch gpsd's fds in /proc for $DEVICE. Second, look @ geoclue, Third, convince gpsd devs
<uvos> Wizzup: we can ask the kernel
<Wizzup> uvos: yes, that also came to mind, to patch libgps
<parazyd> There's many solutions
<parazyd> All unrelated to (1)
<Wizzup> parazyd: watching implies polling
<uvos> Wizzup: we can just have the status bar check with the kernel if gps is in use directly
<Wizzup> parazyd: gpsd *not* running is related to (1)
<parazyd> I don't care, I'm just giving examples of problem relations
<uvos> Wizzup: via the gnss interface
<Wizzup> uvos: that implies polling and wakeups, does it not?
<uvos> Wizzup: no
<parazyd> Wizzup: There shouldn't be gpsd clients if GPS is disabled in control panel.
<parazyd> Because gpsd wouldn't be running
<uvos> Wizzup: all of the solutions i offerd above are totaly indipendant of problem (1)
<Wizzup> uvos: yes, and I think most of them are not great solutions
<Wizzup> maybe I just need to research them some to get a better sense
<uvos> ok Wizzup
<uvos> its clear there is no consensus here
<uvos> please wirte up how you would mangae the whole location stack
<uvos> and link to it here
<parazyd> ++
<Wizzup> ok
<uvos> if you want something different than what parazyd wants
<parazyd> Backlog became too messy
sunshavi has joined #maemo-leste
<parazyd> Ah just want to also add
<parazyd> If we patch gpsd, I'd prefer if we then drop location-daemon altogether
<parazyd> Because we can just modify the dbus export to be what we want.
diejuse1 has joined #maemo-leste
<Wizzup> fixed one typo
<bencoh> /59/59
<bencoh> woops
<Wizzup> feel free to add other downsides/upsides that I missed
<parazyd> Wizzup: So how do I use foxtrotgps with this stack implementation?
<parazyd> Wizzup: Or whatever other libgps client
<uvos> parazyd: Wizzup wants you activate "active liblocation" mode
<uvos> so that the location-status becomes a liblocation client
<uvos> that just keaps the interface open
<uvos> and then open foxtrotgps
<uvos> basicly: "start gpsd by hand" (via gu_
<uvos> basicly: "start gpsd by hand" (via gui)
<parazyd> So I need to open the status menu, then click the button, and then open the control pannel applet and enable permanent gpsd?
<parazyd> lol
<uvos> yeah
<Wizzup> No
<uvos> no?
<Wizzup> uvos: what you said is almost correct: I suggest that (1) location-status is always visible in status menu (2) you can use it to 'Activate GPS' (not enable, that is different)
<uvos> right
<Wizzup> (2) would then allow libgps clients to work directly, as it has the effect of starting location-daemon via dbus
<Wizzup> which starts gpsd
<uvos> "start gpsd by hand, via gui"
<uvos> effectivly
<Wizzup> And then with the same action in reverse, you kill all the libgps clients
<Wizzup> uvos: If you phrase it like that, yes, but it does not require CPA interaction
<parazyd> ok, do it :)
<parazyd> lgtm
<parazyd> Assuming I can forcefully put gpsd into the default runlevel and I can always have it run
<uvos> right
<uvos> thats what i would also do if this is implmented
<parazyd> And we can keep the current behaviour of it not being in a runlevel by default upon installation.
<Wizzup> yeah, that should not be a problem
sunshavi has quit [Read error: Connection reset by peer]
sunshavi has joined #maemo-leste
<Wizzup> we could make that a configurable option, too, to have it start on boot
<Wizzup> depending on how confusing it gets for the user
jmlich has quit [Remote host closed the connection]
<uvos> how about we shim libgps to location-deamon
<uvos> libgps is really trivial
<uvos> then maybe we can also just ditch gpsd entirely
<uvos> and rely und linux gnss subsystem
<diejuse1> I think I have located the problem: "[FAIL] eudev requires devtmps support, udevd did not start ... it failed!"
<diejuse1> I am trying to mount /dev with that "devtmps support" and looking for other solutions
<diejuse1> any idea is appreciated :)
cato`[m] has quit [Quit: Idle for 30+ days]
ravelo has joined #maemo-leste
<Wizzup> uvos: liblocation uses libgps
<uvos> Wizzup: sure
<Wizzup> diejuse1: if you document how we can reproduce what you have, it might be easier to help / figure out what you have / what is wrng
<Wizzup> to make reproduction easier
<uvos> Wizzup: but not directly
<uvos> we could staticly compile a libgps with liblocation
<uvos> *location-deamon
<Wizzup> yeah, we could
<uvos> and then replace system libgps
<Wizzup> I also thought of this, but I feel like this is a potential next step
<Wizzup> one other thing I forgot to mention
<Wizzup> I think android has a 'enable location' button, and iirc it just keeps gps *on* all the time
<Wizzup> so the status button could be seen as analogue to that
<uvos> Wizzup: no
<uvos> it dosent
<Wizzup> are you sure?
<uvos> yes
<Wizzup> my brother has an android and his phone always knows his location (google too, etc)
<Wizzup> maybe google has some android app that does that in the background, then?
<uvos> yeah that keeps _network_ location on
<Wizzup> aha
<uvos> and real android dosent even have that
<uvos> (ASOP)
<Wizzup> something I also saw is that geoclue uses a-gps by just looking up the approx location (in delta km) by looking up the lac etc
<uvos> this is a google services addition
<Wizzup> it doesn't seem to feed it back
<Wizzup> uvos: ok
<uvos> Wizzup: yeah thats a good way to do agps
<Wizzup> well, it doesn't help the modem get a more accurate lock, though?
<Wizzup> (was mostly my point)
<Wizzup> but I suppose it's nice if there's a use for granularty of kilometers
<uvos> right
<uvos> the point is to have the location interface
<uvos> provide the best possible location at all times
<uvos> sometimes that means accuracy of killometers
<uvos> ofc it needs to report this uncertenty
<Wizzup> right
<Wizzup> location-daemon seems like the place that could eventually support that
<Wizzup> (or geoclue if we end up using it.)
<uvos> bonus points to using geoclue
<uvos> its the companion to iio-sensor-proxy and lets it provide true north via the magnetometer
<diejuse1> Can I launch the "profile" UI with a command?
<uvos> btw Wizzup
<uvos> i have had it multiple times now that dbus-scripts will just stop working for no reason
<uvos> it just stops updating the ts rotation untill i restart it
<uvos> no errors or anything and the process is alive
<Wizzup> uvos: hm, that is worth investigating, I just ported it from fremantle mostly
<Wizzup> can you install the dbg symbols and gdb attach?
<Wizzup> diejuse1: I think you should launch it from controlpanel
<uvos> Wizzup: sure when it happens next time
<Wizzup> ty
<diejuse1> Wizzup. I would know if I could launch from terminal
<Wizzup> diejuse1: ok, let me take a look for you
<Wizzup> try 'mameo-invoker controlpanel.launch'
<Wizzup> or just 'controlpanel'
<diejuse1> But controlpanel launch "settings"
<Wizzup> is there not a profiles applet there?
<diejuse1> I would want to launch "select profile"
<diejuse1> directly
<Wizzup> I don't know how to do it directly, that's code in status applet
<diejuse1> From panelcontrol I don't see any access to "select profile". I see "themes", "date and time", "language & region", "display", "notification light", "text input", "internet connections" and "device lock"
<Wizzup> right, there should be one in the application manager
<Wizzup> I forgot do we don't have the original one
<diejuse1> ok
<diejuse1> I saw that is associated to "profiles_status_menu_item.so". Then is not possible to launch directly a .so file? excuse my ignorance
sunshavi has quit [Remote host closed the connection]
<uvos> no you have to dlopen it
<Wizzup> diejuse1: do you not have the status bar on your hildon desktop?
diejuse1 has quit [Ping timeout: 260 seconds]
sunshavi has joined #maemo-leste
cato`[m] has joined #maemo-leste
pagurus has joined #maemo-leste
* enyc meows
<Wizzup> tmlind: gps has been solid for 7 hours now
<Wizzup> no modem resets
<Wizzup> weird
diejuse1 has joined #maemo-leste
<diejuse1> Wizzup: Yes, true, it is there.
<Wizzup> diejuse1: ok, and you can do it from there, but want to do it directly? (launch)
cockroach has joined #maemo-leste
RedW has quit [Remote host closed the connection]
RedW has joined #maemo-leste
<Wizzup> looks like it broke about 10 mins after my msg (gps)
<Wizzup> [Mon May 10 23:25:27 2021] phy-mapphone-mdm6600 usb-phy@1: modem status: 0 off
<Wizzup> [Mon May 10 23:25:35 2021] phy-mapphone-mdm6600 usb-phy@1: modem status: 4 awake
uvos has quit [Ping timeout: 240 seconds]
Entitlement has quit [Ping timeout: 240 seconds]
Entitlement has joined #maemo-leste
cr4y1 has quit [Ping timeout: 265 seconds]