Notices

Thx for this nice work! It even helped me to develop my first small Pebble App to get fuel prices for Germany. And all APIs and PebbleJS things I needed and used just worked like a charm .

Dax spoke to me on Telegram today about the webkit problems. He's had a lot of experience with it doing WebPirate, and thinks it would be easier to fill the gaps rather than move to mozilla, which is says carries a very large overhead and other peculiarities like a global context. He said he'd join the gitter chat later on.

Yes, the context is global but it is architectural decision, and we need just one honestly speaking.
In general I agree, embedding wrapper around gecko is too fragile, and it works (starts) slower (due to the overhead). It renders fast of course, but settings pages aren't supposed to be that complex to requires super-fast browsers.
The only concern I have around the webkit is that it is abandoned, all major players switched to other engines, even QT themselves are moving elsewhere cutting ties little by little with it.
So on jolla we have out-of-the-box only WebKit (part of QT 5.2) and Gecko-embedlite (part of sailfish-browser). Everything else would require static linking or external dependency.
Gecko is modern, it supports HTML5 and all the bells and whistles, while webkit is old plain html4.1

That's an old issue, you can find lots of topics here in TMO and TJC about QtWebKit which is old/buggy/deprecated/abandoned/limited and so on

WebKit shipped with Qt 5.2 was taken from a 2010-11 dev tree.
There is some HTML5 support, but most of them was disabled during packaging in "order to save memory" (?)

I'm eagerly waiting when (and if) jolla updates sailfish to Qt 5.6 which is LTS in order to use Blink through Qt WebEngine.

I tried myself to embed Gecko in WebPirate: it "works", but imho it makes the code complex and difficult to read and maintain and afaik the only "documentation" available is the jolla-browser, I have also tried CEF (Chromium Embedded Framework) but it's even harder to maintain: compiling chromium Content APIs and maintain a QT/Qml binding for it in order to display webpages is too much work for one or two people.

In conclusion, in those two years I have tried lots of ways, but everyone finished with an giant epic fail and they forced me to come back to QtWebKit and wait for QtWebEngine

I think that I have explained all my reasons about QtWebKit and it's status and why me and some other devs are still using it.
It's just to make sure that you don't fall in the same errors and waste time with that

Hi Dax, I clearly understand your reasoning - because your target is full-blown browser which competes with built-in browser, providing features it lacks.
In our case we merely need WebView. WebKit gives us easy and usable WebView - but it's... what it is.
Gecko, on the first approximation - requires just several lines of code to embed.
Yes, documentation is a nigtmare, whatever documentation is available is either incorrect or misleading, but mostly it just doesn't exist.
That's why I spent almost a week to embed it (reverse-engineering gecko-dev snapshot and sailfish-browser), but in the end - it's just about those several lines of code.

Can I add a feature request?

Would be good to be able to react on alarms from the watch - snooze and dismiss.

I'd also like this, though I have no idea if it's possible to interact with reminders and/or alarms. I have a vague recollection of a discussion on here that said that it wasn't yet possible.

Ah sorry, didn't knew the binary could be used as debugging.
Anyway, this is what I get when I try to access the settings page:

(And it keeps on repeating the same thing over and over again)

In the Android app, you have to log in to the store to actually access the store. Now, it's quite convenient (and good for privacy) that rockpool can access the store without an account, but afaik the store account (or hash) can be used by apps to identify the watch. Perhaps this makes a difference.

Did it work with pebbled before?

Since I couldn't access the settings on pebbled, I couldn't try it out
But what you said might be true.
Maybe we can use some "universal" hashes to fake the ID?
Or an option to enable for the ones who wants to use their own hash?