[SailfishOS] [Announce] Rockpool - Pebble daemon for Sailfish - Page 6 - maemo.org

Active Topics

Porting apps to Leste (29)
to Maemo 7 / Leste by Arno_11 - 2 days, 18 hrs ago
Install Fonts in N900 (68)
to Maemo 5 / Fremantle by teroyk - 3 days, 21 hrs ago
more...

Page 6 of 72

Thread Tools

abranson	2016-03-19 , 18:01
Posts: 287 \| Thanked: 862 times \| Joined on Dec 2015	#51

Originally Posted by Astaoth

Ok, so here is the logs when I start rockpoold from the command line : pastebin.com/p4brZPED .
If we take a look at the end, we can read that the new version is detected, but the internal status about available update is set to false. And with the calls from qdbus, I have the same results than before.
Maybe because my version number has an unexpected format rockpool can't define if my version is the latest or not ?

Thanks for that. I think Michael only anticipated that vX.Y.X format, probably because everyone was stopped on 3.9.1. I'll fix that tonight for you.

Edit: version 0.5-2 should be able to cope with you now.

Last edited by abranson; 2016-03-19 at 18:25.

Quote & Reply |

The Following 4 Users Say Thank You to abranson For This Useful Post:
Amboss, Bundyo, Feathers McGraw, tangent

	Feathers McGraw	2016-03-19 , 21:01
	Posts: 654 \| Thanked: 2,368 times \| Joined on Jul 2014 @ UK	#52

Code:

nemo ~ $ qdbus org.rockwork /org/rockwork/B0_B4_48_C6_10_B1 org.rockwork.Pebble.PerformFirmwareUpgrade

Now running firmware version 3.10.1 on my Pebble Time Steel. Thank you

Quote & Reply |

The Following 3 Users Say Thank You to Feathers McGraw For This Useful Post:
abranson, Amboss, tangent

ruff	2016-03-20 , 10:53
Posts: 207 \| Thanked: 482 times \| Joined on Mar 2016	#53

Originally Posted by Fuzzillogic

It's elegant, but will it be enough?

Daemons are controlled by systemd or by user. Even when it is dbus unit it's still called and spawned via systemd. So either you start it in cli manually or via systemd.
Conflict statement allows isolation of the services to avoid "strange unpredictable" behaviour.
Pebbled app allows start/stopping the service and see its status. rockpoold is autostarted by app, moreover app locks down when daemon is not running (falls to waiting for service page).
This basically grants certain predictability, not providing ultimate solution. Bcz ultimate solution should be a user's choice, nor programmer's.

To be completely on the safe side we might implement checks for the stray pebbled processes (eg service is down but process is running) but that's too much overhead and unless reported with certain persistence by users - should be avoided.

Quote & Reply |

The Following 3 Users Say Thank You to ruff For This Useful Post:
Amboss, Feathers McGraw, tangent

abranson	2016-03-20 , 11:02
Posts: 287 \| Thanked: 862 times \| Joined on Dec 2015	#54

Originally Posted by ruff

Daemons are controlled by systemd or by user. Even when it is dbus unit it's still called and spawned via systemd. So either you start it in cli manually or via systemd.
Conflict statement allows isolation of the services to avoid "strange unpredictable" behaviour.
Pebbled app allows start/stopping the service and see its status. rockpoold is autostarted by app, moreover app locks down when daemon is not running (falls to waiting for service page).
This basically grants certain predictability, not providing ultimate solution. Bcz ultimate solution should be a user's choice, nor programmer's.

To be completely on the safe side we might implement checks for the stray pebbled processes (eg service is down but process is running) but that's too much overhead and unless reported with certain persistence by users - should be avoided.

I'd rather not have explicit references to other applications in the code, but we're not fixed to the RockWork design. A lot of that comes from workarounds for the Ubuntu platform. What would you think about having the same service control in Rockpool as Pebbled? I'd already partially migrated the ServiceControl over. It would be easy to add methods to enable or disable the service.

Tbh I'm coming round to the idea of declaring a conflict with pebbled in the RPM. I don't think smokku would mind really.

Btw I added a gitter chatroom. I thought it might be a good place to discuss finer implementation details in realer time.

https://gitter.im/abranson/rockpool

Quote & Reply |

The Following 4 Users Say Thank You to abranson For This Useful Post:
Amboss, Feathers McGraw, ruff, tangent

ruff	2016-03-20 , 11:19
Posts: 207 \| Thanked: 482 times \| Joined on Mar 2016	#55

Originally Posted by abranson

What would you think about having the same service control in Rockpool as Pebbled? I'd already partially migrated the ServiceControl over. It would be easy to add methods to enable or disable the service.

Yes, I'm also leaning to this idea, hardcoded autostart is also a bit impolite. RPM conflict is probably way too much, F.i. I'd prefer having both apps on my jolla. Just in case

Service conflict should fine. We may probably also add simple org.pebbled introspection to give a warning when target is published. A fairly simple call to implement in servicecontrol. May even do it right in QML - nemomobile dbus plugin by default is on board.
Chat is a good idea, will most probably need it while merging.

Quote & Reply |

Astaoth	2016-03-20 , 14:13
Posts: 134 \| Thanked: 370 times \| Joined on Oct 2012 @ France	#56

Originally Posted by abranson

Thanks for that. I think Michael only anticipated that vX.Y.X format, probably because everyone was stopped on 3.9.1. I'll fix that tonight for you.

Edit: version 0.5-2 should be able to cope with you now.

Thank you, it's working now.

I have a feature to suggest, if you have enough time. Do you think it will be possible to use the PT microphone during a phone call, and to have the call sound from the phone speaker ? It could be usefull when driving for example.

Quote & Reply |

abranson	2016-03-20 , 14:42
Posts: 287 \| Thanked: 862 times \| Joined on Dec 2015	#57

Originally Posted by Astaoth

Thank you, it's working now.

I have a feature to suggest, if you have enough time. Do you think it will be possible to use the PT microphone during a phone call, and to have the call sound from the phone speaker ? It could be usefull when driving for example.

Are you apps working too?

I know there's an microphone elements to the API, but I thought it was for voice commands instead of Blake's 7 style communications. I don't have a Pebble Time though, so I have no idea. I should think the Voice stuff will be on our shared roadmap with RockWork, but that's a way off yet I should think.

Quote & Reply |

ruff	2016-03-20 , 20:06
Posts: 207 \| Thanked: 482 times \| Joined on Mar 2016	#58

Ok, so with latest commit interface is somewhat usable.
I didn't touch yet DeveloperPage part (includes screenshot page), firmware update and side-load (import page). The rest is more or less usable. (Finally I've set my fav. apps from jolla

)

I've noticed that rockpoold segfaults occasionally on reconnection (need to enable coredumps to debug)

Code:

[D] Pebble::onPebbleDisconnected:494 - Pebble disconnected: "Pebble Time C229" 
[D] WatchConnection::scheduleReconnect:38 - Attempting to reconnect in 10 seconds 
Segmentation fault

Another observation is that app_icon is mostly generic. I.e. I'm receiving mail notifications however icon is still generic. probably because the source is still system event hub, will need to monitor dbus attributes and probably perform certain remapping.

Quote & Reply |

The Following 6 Users Say Thank You to ruff For This Useful Post:
abranson, Amboss, Feathers McGraw, MikeHG, mrsellout, tangent

abranson	2016-03-20 , 21:01
Posts: 287 \| Thanked: 862 times \| Joined on Dec 2015	#59

Originally Posted by ruff

Ok, so with latest commit interface is somewhat usable.
I didn't touch yet DeveloperPage part (includes screenshot page), firmware update and side-load (import page). The rest is more or less usable. (Finally I've set my fav. apps from jolla )

Brilliant, that's a great start! Let's get it out there!

Originally Posted by ruff

I've noticed that rockpoold segfaults occasionally on reconnection (need to enable coredumps to debug)
Code:
[D] Pebble::onPebbleDisconnected:494 - Pebble disconnected: "Pebble Time C229" 
[D] WatchConnection::scheduleReconnect:38 - Attempting to reconnect in 10 seconds 
Segmentation fault

Thanks, I'll keep an eye out. Stability is next priority I think.

Originally Posted by ruff

Another observation is that app_icon is mostly generic. I.e. I'm receiving mail notifications however icon is still generic. probably because the source is still system event hub, will need to monitor dbus attributes and probably perform certain remapping.

I'm not sure what you mean here, do you mean the icons on the notifications on watch? The current mapping of notifications to pebble icons is in SailfishPlatform:: onNotification. That chooses an icon based on notification owner, category, android package etc.

Quote & Reply |

The Following 3 Users Say Thank You to abranson For This Useful Post:
Amboss, Feathers McGraw, tangent

ruff	2016-03-20 , 22:03
Posts: 207 \| Thanked: 482 times \| Joined on Mar 2016	#60

Originally Posted by abranson

I'm not sure what you mean here, do you mean the icons on the notifications on watch? The current mapping of notifications to pebble icons is in SailfishPlatform:: onNotification. That chooses an icon based on notification owner, category, android package etc.

No, I mean icons which are used for notifications filter, presumably they should represent local apps which are emitting notifications. So far I see 3 in the filter, gmail, email and twitter, but all of them are having dialog-question-symbolic icon.