News for hackers. Stuff that matters.
Re:Why Did Mozilla Remove XUL Add-Ons? by ekianjo on Friday August 21, 2020 @03:30AM (#24232862)
> * oh, and by the way, since add-ons could do everything, they could very easily do anything to the operating system, from stealing passwords to pretending to be your bank.

So now there's way less things that add-ons can do, and it's supposed to be all peachy for Firefox users? Talk about competitiveness.

Re:Why Did Mozilla Remove XUL Add-Ons? by KMag on Friday August 21, 2020 @05:16AM (#24233354)
They're much better off starting with a highly restricted sandbox and extending that sandbox in safe ways when demand is demonstrated.

The alternative is letting their largely tech-naive user base install all kinds of add-ons under the illusion that it's no more dangerous than installing phone apps from an audited walled-garden app store.

Mozilla doesn't have the resources to audit all of the add-ons out there, and generalized automated auditing with a wide-open API would require solving the halting problem. (Accurately enumerating all reached states of a Turing-complete program would allow you to check if it reaches the halted state, thus solving the halting problem.)

Re:Why Did Mozilla Remove XUL Add-Ons? by cultureulterior on Friday August 21, 2020 @06:00AM (#24233550)
But they don't extend the sandbox (Look at tab tree). Also, it's impossible to demonstrate demand for an extension that doesn't exist.
Re:Why Did Mozilla Remove XUL Add-Ons? by takluyver on Friday August 21, 2020 @07:31AM (#24233960)
I'm not sure what you mean about tab tree. There are tab tree WebExtensions.
Re:Why Did Mozilla Remove XUL Add-Ons? by eitland on Friday August 21, 2020 @08:50AM (#24234437)
yep, but you either have to let it coexist with the main tab bar or go further and further to manually disable that.
Re:Why Did Mozilla Remove XUL Add-Ons? by chaorace on Friday August 21, 2020 @08:53AM (#24234459)
Yes, but they are pitifully weak compared to the same addons back when they were implemented in XUL. It's not just because XUL was more free. WebExtensions as an API just lacks the scope for such enhancements. Mozilla essentially copies Google's API and they're very hesitant to make customizations, so major enhancement requests will just sit around for years without traction. I'll list out a few major issues with the current API for the sake of example:

* No native hotkeys.

Ok, fine, there are technically native extension hotkeys, but they break so often that nobody uses them. Even if they were fixed, the bindings are arbitrarily limited to two-modifier combinations.

Instead, you're stuck hacking around the limitation by injecting hotkeys into each individual page via JS. If the page you're on isn't a normal webpage (new tab, 404, PDF...), you're out of luck. It's mouse time.

Imagine the pain of writing and rolling your own high level key binding API, now imagine the additional memory/compute from running that in every single tab instance. This is already a huge mess that makes any tab extension a second-class experience.

* UI extension is hopelessly neutered

WebExtensions basically have 3 main places where you can render UI: solo pages, a toolbar popout, and the sidebar. Of these three, only the sidebar is a permanent, persistent place that works for the purposes of a tab extension. Unfortunately, the sidebar is hopelessly neutered. The sidebar is shared between all extensions AND the native bookmark/history managers, so bad news if you need it for something else! Since it's shared, you can't control the size/position programatically (Oh yeah... and the manual resize threshold is limited to about 1/6 horizontal screen width).

To add insult to injury: the sidebar steals page focus, so I hope you enjoy clicking back into the web page every time you interact with the sidebar UI via mouse! This dovetails nicely with point #1. There's no way to bind a key to your extension's sidebar. That means you can't toggle your tab bar (a la F11). It also means you have to fish for your mouse, fiddle with the toolbar, then switch sidebar panes any time you lose the sidebar for any reason.

Imagine the pain of supporting an extension that literally does nothing unless you get your users to dig into the toolbar and enable a little-known UI element. Now... imagine the pain of asking them to do that, repeatedly, whenever it goes away.

* Horizontal tab bar is a pipedream

Your control over actual tabs, in the actual tab bar, is incredibly limited. If you're not outright hiding the tab bar and replacing it with a sidebar hack, it's basically guaranteed to be a mess. You cannot style nor group tabs programatically. All you can do is hide or reorder them, which is a pretty bizarre decision, imo...

The best that anyone can do for non-overhaul extensions, like "tab grouping", is stashing the UI away into the toolbar popout menu. I hope it's fairly obvious why this isn't an ideal way to present that information, especially when you consider that it's a focus-stealing modal.