[throw10920]

Yes, because the APIs aren't perfectly fleshed-out. And they may never be, and yet that's still completely OK because the WebExtension model is obviously better along the performance, security, portability, and API stability axes.

[wtallis]

WebExtensions aren't clearly better from a performance standpoint. Performance won by removing useful features doesn't count.

[throw10920]

This is blatantly false, and one of the most dishonest and manipulative claims that I've seen on HN.

Performance is performance. If one technology is more performant by removing features, useful or not, it it factually faster, and that performance absolutely does count. Features are completely irrelevant to performance measurements of a system.

If you have two cars, car A with top speed 160 MPH and a 0-60 of 3s, and car be with top speech 120 MPH and a 0-60 of 5s, some people may still prefer car B because it has better mileage or nicer features or is cheaper (which is the overall value judgement that you seem to be extremely confused about), but precisely zero sane people will tell you that car A "isn't clearly better from a speed standpoint" because it has less features than car B.

[wtallis]

Crippling my ad blocker doesn't make my browser faster on average, even though dishonest benchmarks from advertising companies may claim otherwise. Removing XUL also didn't make TreeStyleTab faster; quite the opposite.

Aside from crippling ad blockers, are there any other theoretical performance improvements enabled by WebExtensions, or is it all about reducing opportunities for badly-written extensions to have an impact?

[throw10920]

> Crippling my ad blocker doesn't make my browser faster on average, even though dishonest benchmarks from advertising companies may claim otherwise. Removing XUL also didn't make TreeStyleTab faster; quite the opposite.

OK, so now you're moving the goalposts, continuing to dishonestly redefine words, and cherry-picking specific instances of addons that support your point, while ignoring the fact that I soundly refuted your utterly insane previous argument.

> Aside from crippling ad blockers

No? WebExtensions clearly did not "cripple" ad blockers by any stretch of the imagination. Maybe you're conflating WebExtensions and Manifest v3?

> Removing XUL also didn't make TreeStyleTab faster; quite the opposite.

Cherry-picking items to try to support your point only proves that you don't have robust evidence to support it in general. This is the hasty generalizations fallacy. As someone who lived through the WebExtensions transition, I didn't perceive any slowdown in any of my dozen or so extensions.

> are there any other theoretical performance improvements enabled by WebExtensions

Yes - if you had any knowledge at all of the old addon model, you'd know that the old XUL-based addons prevented Firefox's move to the multi-process Electrolysis architecture, which significantly improved performance.

> is it all about reducing opportunities for badly-written extensions to have an impact

Yes, that is (on top of everything else) a performance benefit. Humans are not robots - all humans write bad and buggy code, and the XUL model not only made it much easier to write buggy and slow code, but the lack of a well-defined interface resulted in ossification that massively inhibited Mozilla's ability to develop Firefox.Even if it didn't, making changes that help/force the lower 99% of programmers to write better code while mildly inhibiting the ability of the top 1% of of programmers is absolutely worth it, and in practice has massively improved performance.

If you tried to run old Firefox on a modern CPU with a bunch of extensions, you'd very clearly see the performance difference due to the ability to actually take full advantage of more than one core, and due to the improvements that Mozilla was able to make by deprecating the old API.

Perhaps stop commenting unless you can stop committing numerous fallacies, making utterly insane statements, pretending that human factors don't exist, and making statements about things that you have no understanding of.

[wtallis]

Firefox moved to multi-process before moving to WebExtensions. You've lost count of how many mass extinctions the Firefox extension has been through, but there were XUL extensions that were updated to be compatible with multi-process, and then later had to be totally rewritten with completely new UIs when XUL was killed. And the usability hit that extensions like AdBlock Plus and NoScript suffered was crippling, even if it wasn't quite as bad as MV3, and NoScript lost features that went beyond just UI.

[throw10920]

> Firefox moved to multi-process before moving to WebExtensions.

And? Removal of XUL addons were still a prerequisite for the multi-process architecture. Mozilla just realized that WebExtensions was a sane, performant extension API that worked well with e10s, and would useful for compatibility with Chrome.

> You've lost count of how many mass extinctions the Firefox extension has been through

Two? Hardly a lot.

The XUL model is inferior to the Web Extensions model. No amount of trying to cherry-pick specific instances of extensions that had local functionality or performance losses will detract from the facts that (1) the XUL addon system was inferior and (2) had to be removed in order to make Firefox (both the browser and ecosystem as a whole) more performant, secure, stable, and easier to maintain.