Flypig's gecko dev diary

In case other people want to follow along @flypig’s journey on the way to ESR91:

Fascinating read with a ton of great information on sfos, browser, platform development…


Thanks for posting this @throwaway69. I apologise for the content being so console-heavy and poorly structured. It’s more core dump than finely crafted narrative! I aim to post every day.


I’m absolutely fine with the stream of thought writing, makes for a good following and won’t take away too much of your precious time with explaining every little detail I hope.


Way to go.
Keep it, because when I’m better I’ll buy a 10 III and so I wish to use ESR91 instead of that old grandma :smiley:

1 Like

Cool that you share experience!Very cool!

1 Like

Thanks for doing this most of all, and for documenting in whatever way you see fit.

A tiny niggle from a web dev perspective: I love it when there’s lots of links embedded in the text, but the radically different color disrupts the reading flow. (It’s interesting to see how various (news) sites solve the issue of distinguishing links while not disrupting readability.)


@flypig > What’s there is clearly clang version 14, whereas the gecko build script was complaining (before I hacked it) about needing clang greater than 5.

Probably kick in the dark but one possibility is single digit check.


It is really fun to read your experience and to see how everyone tries to participate :slight_smile:


Just a general question, regarding the very high number of patches. Would it make sense to bring them upstream?

1 Like

You can maybe eek out one or two that can be upstreamed (and some are already picked from upstream), and another handful that when reworked will make things easier. But the whole point of the series is to bring back something that was actively dropped. Add to that, that upstreaming (what little you could) will start paying off in maybe 3-4 uplift cycles.

It would of course be very cool if Mozilla wanted these embed bindings back more or less as a whole. Seems that they don’t, especially not without a more active community behind them.


Since firefox runs on pretty much anything (???) modern doesn’t it make sense for mozilla to drop them?
I mean if we had a modern compositor we could have firefox and only need to modify the UI to fit the SFOS paradigm.


I can certainly see both sides. What you say is true; but actual proper, somewhat generic, embed-bindings are also very nice to have. Webkit has gained a lot of reach from theirs, and afaik, Blink keeps them. Oh, and Mozilla maintains specific ones for at least Android because evidently it isn’t enough for everything.

Also; i did not know it was a no-go for compositor reasons… got a source for that?


I think its been mentioned many times in the forums.

Also → Xwayland packages for Sailfish OS - #7 by artur


Right; but i have only ever seen it from people that aren’t actually involved - and it has looked a lot like speculation.
I’m not sure i’d directly equate lack of X-compatibility with being old/primitive/…, even if probably technically true in this case. Whatever needs it (if that’s what you are saying?) has a share in the problems.

Edit: oh, that’s what you linked to. Big screen does not do a good job of focusing you on a post near the end. In that case - they are still unreasonably hipster.

1 Like

I linked arturs answer about an interface that Firefox needs and its not even present in 5.15. (or tried to in case i screwed something up).


Thanks everyone for all of your kind comments.

Thanks for the advice and no need to apologise, it’s always good to get feedback. Could you share some links where you consider it to be done well (the news sites you mentioned)?

That’s a really nice idea. It’s not quite a single digit check, but you’re along the right lines. Check the latest post for more on this.

Yes, that’s a nice summary @attah and fits my understanding of the situation too.


"Only" modify the UI

carefull there with putting your hands on the cutting board :wink:

I mean if we had a modern compositor we could have firefox and only need to modify the UI to fit the SFOS paradigm.

So firefox is not running on evertyhing / wayland? That’s surprising. Perhaps the wayland protocol on SFOS is old?

The questionmarks on the original comment denote that i am not 100% sure. Firefox wayland requires stuff our compositor doesn’t have. Thats a given. We need a new compositor (among many new things we need) AND most importantly newer QT. Or a new toolkit.

Our wayland version is recent enough i think but thats not the problem.

1 Like

So extensions to lipstick / functionality? Is there a valid reference baseline avilable somewhere? Mir?

So nice to read. Thank you for taking this herculean task upon your shoulders!

1 Like