GNOME Shell Picks Up Performance Improvements For Extensions

[phoronix]

Phoronix: GNOME Shell Picks Up Performance Improvements For Extensions

While days too late for squeezing into GNOME 3.34.0, the GNOME Shell has landed a one year old merge request providing various fixes and performance improvements to its extension system...

http://www.phoronix.com/scan.php?pag...xtensions-Perf

[uid313]

Last time I tried to write a GNOME Shell extension, I used the gnome-shell-extension-tool, and it generated some ugly JavaScript code, it couldn't use modern JavaScript ES6 classes.

[treba]

AFAIK, 3.35 has not been branched yet, so this is all still going into 3.34

[Britoid]

AFAIK, 3.35 has not been branched yet, so this is all still going into 3.34

Possibly 3.34.1

[treba]

3.34.1 might end up eslint clean \o/ The extensions really need to up their CI game. How can they live with poor hosting on github? No wonder some extensions break when they miss out on Gitlab CI.

https://gitlab.gnome.org/GNOME/gnome...e_requests/716

While Gitlab is surely great, you can have all sorts of external CI hooked up to github as well. Talking about travis-ci etc.

[Britoid]

3.34.1 might end up eslint clean \o/ The extensions really need to up their CI game. How can they live with poor hosting on github? No wonder some extensions break when they miss out on Gitlab CI.

https://gitlab.gnome.org/GNOME/gnome...e_requests/716

The extension system imho should just be removed or restricted to only a few areas (e.g. adding things on the top status menu).

[tildearrow]

The extension system imho should just be removed or restricted to only a few areas (e.g. adding things on the top status menu).

That'd make the desktop even less flexible, and after that I'm pretty sure many would move away from it...

[jacob]

Is the problem that all extensions run within the main UI thread being addressed, or is that too disruptive a change that will have to wait until GNOME 4?

[jo-erlend]

Is the problem that all extensions run within the main UI thread being addressed, or is that too disruptive a change that will have to wait until GNOME 4?

Good question. Sounds difficult to me. JS is still single-threaded, isn't it?

[cynical]

Good question. Sounds difficult to me. JS is still single-threaded, isn't it?

Mutter is written in C, not JavaScript. And Gnome-Shell itself is also partly implemented in C, so the language is not the barrier.

Is the problem that all extensions run within the main UI thread being addressed, or is that too disruptive a change that will have to wait until GNOME 4?

Way too disruptive. You can read a related proposal for "GNOME Shell 4" here, outlining the changes that need to be made.