GNOME Might Need To Crack Down On Their JavaScript Extensions

[duby229]

The blog post Michael linked is really concise, logical and clear. I recommend you read it if you haven't already:
https://eischmann.wordpress.com/2018...ll-extensions/

This is anecdotal at best, but it's very telling... That dudes blog is on wordpress... arghhh....

[cygn]

Come to Plasma.

[tessio]

I'm switching to XFCE.. So long and thanks for the fish!

[cybertraveler]

Yeah "looked like a good idea" to someone, very concise, logical and clear. Sorry, but that problems had been known for a long time, you need to coalesce drawing in one thread for performance (single rendering context, no locking) and you only do thing you can verify to not do random things (ie dont run client code, just client data).

java-script applets would not be the problem, letting them directly draw is.

His blog post is indeed concise, logical and clear. I made no claim about the decisions of Gnome devs when they started work on Gnome 3. Also to re-apply my claim to a single fragment of one sentence of the blog post is unreasonable and absurd. I was clearly referring to the entire post.

If you were to compare his clear & logical blog post with your barely comprehensible comment above, then by comparison the blog post stands out as being deserving of even more positive praise.

"discordian": name checks outs.

[kpedersen]

What do you mean by "cannot support multiple users on the same box"?.

Honestly it is just easier for you to try it to understand how Gnome 3 / Wayland fails.

Set up a computer in your house with VNC and see if your Mom, kid, wife, dog can all use their own Gnome 3 desktop at the same time.

Make sure that they all try running xprop and making sure that fails, otherwise you are obviously running Gnome via X11 instead of Wayland by mistake.

Good luck. Let me know how it goes

[ssokolow]

This is anecdotal at best, but it's very telling... That dudes blog is on wordpress... arghhh....

There are a wide variety of reasons someone's blog may be on WordPress. For example, I'm still on WordPress because, though I want to migrate to a static site generator, I haven't yet had time to prepare for a data migration that's up to my own standards.

The main things I need to find time for being:

1. Writing the tools necessary to verify that I've broken no links in the migration process (I haven't broken a link that I had the power to preserve since I switched away from hosting-provider-specific URLs a little over a decade ago)
2. Writing tools and doing manual auditing necessary to verify that everything that's not strictly HTML (eg. WordPress shortcodes) gets translated properly
3. Writing an alternative to migrating all of my comments into Disqus. (I put a high value on supporting in-page comments, even if I only get them infrequently.)

[Tomin]

Honestly it is just easier for you to try it to understand how Gnome 3 / Wayland fails.

Set up a computer in your house with VNC and see if your Mom, kid, wife, dog can all use their own Gnome 3 desktop at the same time.

Make sure that they all try running xprop and making sure that fails, otherwise you are obviously running Gnome via X11 instead of Wayland by mistake.

Good luck. Let me know how it goes

Okay, I see. That's quite a bit more than what I thought supporting multiple users on the same box would mean. Sort of thin or fat client system then. There probably isn't yet such a system using Wayland because there is quite little demand for it.

[wizard69]

JavaScript is not so much of an awful idea on its own. I don't know what implementation they're using in GNOME but most of them are decently fast and memory usage is not too awful, especially not when you have to deal with something as simple as a shell extension. I don't like the language and I hate anything related to web tech, but it's hardly really going to matter here.
The big bonus with JS here is that many people know how to write in JS and that it isn't hard to learn either because it's so widly used.

This is actually the whole problem iwht Java Script in this sort of situation. You end up with programmers that would be better off milking the cow than to engage in producing shell extensions. I really hate to say this but Java Script attracts the type of people that will likely never develop into a competent programmer. By the way that doesn't mean there aren't professional Java Script programmers out there that are highly skilled.

If shell extensions manage to crash GNOME, it's not really the language at fault,

If the language attracts the wrong type of developer it is part of the problem. A better API might help but we should also look back on the early days of web browsers that suffered greatly from crap Java Script.

but the way bindings are done. From what I'm seeing from this dev's blog, they're absolute insanity if your focus is on stability. I bet it's trivial to take the whole shell down with a few lines of JS, just because you have such a huge amount of control over the shell.
If the design is similar to what I think it is, it's surely because it was at the time the easiest solution to let extensions be really extensive while not being a nightmare to implement.

Reliability to me is far more important than any shell extension I've seen to date. I'd rather see the GNOME team remove all dynamically loadable extensions myself. I just see it as completely silly if you expect to have a reliable and integral pice of the desktop working all the time.

Seeing the comments on that blog post, it looks like many of those crashes could certainly be prevented by hardening the code to things that could as well be warnings.

From my perspective extensions are the opposite of what one expects from a UNIX like platform. If people need additional functionality write an app!!!!!!

[wizard69]

Or C/C++/Rust/<insert compiled language here>?

As a whole I really wish that open source developers would move away from the world of C and onto a modern language. My feelings about C++ are mixed, it has been greatly improved in recent times but it has plenty of dark areas. Apples Swift is really grabbing my attention and frankly I'd like to see it explored more fully to implement open source software in the future. I like that Swift in many ways has the feel of an interpreted language, yet delivers the speed of many compiled languages.

So yeah I have to agree if the GNOME team wants to impress me, start using languages like C++/Rust or Swift. Build your software so that it has a long future and is maintainable and understandable by many. By the way I know that Rust and Swift are not really ready for prime time. In an ideal world both would be standardized much in the way C or C++ have been.

It is my understanding that the people on the QT side of the world are hard at work adapting some of their older code to the latest C++ standards. If that goes well they could move that tool kit way ahead of GNOME. Note I've never been a KDE /QT fan, preferring the simpler environments. That could change though if they can deliver superior products that are reliable and maybe a little less gaudy.

[wizard69]

A window manager shouldn't be able to load third party code,

Exactly! If someone wants a feature or special behaviour they should generate their own copy and be responsible for the bugs.

if it crashing brings down the entire session. That's just begging for issues. GNOME itself should not be the Window Manager, they need to split. What an awful design decision that seems to be just to "rush to get it onto Wayland"

It is a massive productivity killer.

By the way, on Fedora 28 I've had the WM crash and I have not load a single extension. So yeah a desktop crash is pretty bad.

Both Windows and OS X can do session recovery, so either recover the session or split GNOME and Mutter seems the most future-proof solution.

Or just don't crash in the first place.

FYI I Use Gnome as my main DE, but I too am annoyed at its issues in performance.

That is actually odd, I find Fedroa 28 performing better than Windows and nearly as good as OS/X. I'm going to work with it until the end of the year to see if it can become my daily driver.