[SailfishOS] [Announce] Rockpool - Pebble daemon for Sailfish - Page 23 - maemo.org

Active Topics

The N900 is now 15 years old! (9)
to Nokia N900 by Arno_11 - 7 hrs, 54 mins ago
Sailfish OS for the Motorola Moto G7 Power (XT1955-5) - (ocean) (4)
to SailfishOS by Maemish - 5 days, 20 hrs ago
more...

Page 23 of 72

Thread Tools

piggz	2016-04-07 , 21:11
Posts: 213 \| Thanked: 414 times \| Joined on Nov 2010	#221

Originally Posted by ruff

Can you confirm it works ok on android/ios? I.e. app could be migrated to time series with no backward compatibility.

im not sure i follow you...used to work ok on the timeline fw on my bb10

Quote & Reply |

The Following 2 Users Say Thank You to piggz For This Useful Post:
abranson, Feathers McGraw

abranson	2016-04-07 , 21:34
Posts: 287 \| Thanked: 862 times \| Joined on Dec 2015	#222

Originally Posted by ruff

Yes, however gears have separate limits, so even if it compiles, it doesn't mean runtime is fine. Just to be sure it's really rockpool-specific(maybe for the certain platform) problem.

Speaking about the platform. Can you estimate from the top of your head protocol-level changes btw pebbled(2.x) and rockworkd(3.x) implementation? I'm just thinking whether ios/android could be falling back to older proto when speaking to aplite for certain transfer types. Which may lead to nacks for pure 3.x comm.

Afaik the appmessage protocol is identical in v2 and v3. The code is pretty much identical between rockwork/rockpool and pebbled.

Quote & Reply |

	Bundyo	2016-04-08 , 05:09
	Posts: 4,708 \| Thanked: 4,649 times \| Joined on Oct 2007 @ Bulgaria	#223

All Pebbles communicate in UTF-8. However in older models there was no space for the font characters, maybe that changed with PTR. Anyway, simple language pack install should bring cyrillic support to Pebble. You can find several such LPs here:

https://forums.getpebble.com/discuss...Comment_153542

Not sure if this can be installed through Rockpool, but works a treat on an Android device. Since the language pack survives firmware upgrades, you need to do this only once.

__________________
Technically, there are three determinate states the cat could be in: Alive, Dead, and Bloody Furious.

Quote & Reply |

The Following 2 Users Say Thank You to Bundyo For This Useful Post:
Amboss, Feathers McGraw

abranson	2016-04-08 , 05:22
Posts: 287 \| Thanked: 862 times \| Joined on Dec 2015	#224

That's what was suspected, thanks for the confirmation. If anyone can sniff that process and tell me how it's uploaded I'd be very grateful. I'm afraid I have no Android device I can use.

Quote & Reply |

The Following 2 Users Say Thank You to abranson For This Useful Post:
Bundyo, Feathers McGraw

ruff	2016-04-08 , 06:34
Posts: 207 \| Thanked: 482 times \| Joined on Mar 2016	#225

Originally Posted by abranson

Afaik the appmessage protocol is identical in v2 and v3. The code is pretty much identical between rockwork/rockpool and pebbled.

Ok, then we need logs from watches to understand why it's nacking appmsg. The message is fairly simple - simple scalar value, nothing what could potentially be wrongly serialized.

Does anyone have annotated firmware disassembly accidentally? To look for possible reasons to nack.

Last edited by ruff; 2016-04-08 at 06:49.

Quote & Reply |

The Following User Says Thank You to ruff For This Useful Post:
Feathers McGraw

abranson	2016-04-08 , 06:57
Posts: 287 \| Thanked: 862 times \| Joined on Dec 2015	#226

Originally Posted by ruff

Ok, then we need logs from watches to understand why it's nacking appmsg. The message is fairly simple - simple scalar value, nothing what could potentially be wrongly serialized.

I tried to switch the type of that first request for the token list to different types, but that didn't help. It could still be bad serialization, but there's only so many ways you can send a '1'.

Originally Posted by ruff

Does anyone have annotated firmware disassembly accidentally? To look for possible reasons to nack.

Appmessages on the whole work fine, but I think it's the binary on the watch that decides whether to ack or nack in response. I found the source:

https://github.com/cpfair/pTOTP

I think the next step is to confirm that iOS and Android work with older watches on v3, then we can try to contact the author.

Last edited by abranson; 2016-04-08 at 07:22.

Quote & Reply |

The Following 2 Users Say Thank You to abranson For This Useful Post:
Amboss, Feathers McGraw

ruff	2016-04-08 , 07:57
Posts: 207 \| Thanked: 482 times \| Joined on Mar 2016	#227

Yes I've looked at the code already and didn't find anything there, that's why i suspect it's firmware which is refusing appmessage for certain reason. In that case it may not even reach the app.

Quote & Reply |

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

abranson	2016-04-08 , 08:06
Posts: 287 \| Thanked: 862 times \| Joined on Dec 2015	#228

I'd like some more confirmation of this behavior in other watches. My watch might need a reset or something. Are you ready with the gecko branch? It definitely needs that.

Quote & Reply |

The Following User Says Thank You to abranson For This Useful Post:
Feathers McGraw

ruff	2016-04-08 , 08:19
Posts: 207 \| Thanked: 482 times \| Joined on Mar 2016	#229

yes, nothing atm on my list for it.
Do you want to merge gecko-only or the WIP which also includes mce handler and changed notification filter? with german translation and alert dialog on top of that.