Twitter | Search | |
Anne van Kesteren Dec 4
It’s greatly encouraging in that talking to people at the Mozilla All Hands, we are still committed to driving the Gecko and Servo implementations of the web platform.
Reply Retweet Like
Anne van Kesteren Dec 5
Replying to @annevk
This might be good time to reiterate why independent implementations, both at a high-level (e.g., browser) and low-level (e.g., text encodings) are valued and important to a healthy and thriving web platform. (And please do feel free to chime in and discuss.)
Reply Retweet Like
Anne van Kesteren
No single company should control the future of the web platform.
Reply Retweet Like More
Anne van Kesteren Dec 5
Replying to @annevk
With a single implementation, bugs will become features much more quickly, leading to ecosystem decay. (We might still have the IE6 three pixel gap.)
Reply Retweet Like
Anne van Kesteren Dec 5
Replying to @annevk
Independent implementations ensure standards can be implemented from first principles and are not just a bunch of words to appease a complicit standards group.
Reply Retweet Like
Anne van Kesteren Dec 5
Replying to @annevk
It allows companies to experiment with radical new strategies to solve old problems better (e.g., Servo). Such a refactoring would be impossible without standards and tests.
Reply Retweet Like
Anne van Kesteren Dec 5
Replying to @annevk
Standards with independent implementations are vastly easier to adopt in new devices and platforms.
Reply Retweet Like
Anne van Kesteren Dec 5
Replying to @annevk
Software monocultures are bad for security. Almost every security bug will affect all users.
Reply Retweet Like
Anne van Kesteren Dec 5
Replying to @annevk
It protects against stagnation of the web platform. E.g., if the company behind it stops investing in it. (Again, see IE6.)
Reply Retweet Like
Anne van Kesteren Dec 5
Replying to @annevk
Independent implementations, and therefore voices, can also block bad ideas much more successfully. (E.g., PNaCl, Dart, ActiveX.)
Reply Retweet Like
Anne van Kesteren Dec 5
Replying to @stpeter @__farre__
Again, I welcome your thoughts. Many thanks to and for helping out!
Reply Retweet Like
Nicholas Nethercote Dec 5
Replying to @annevk
If I had to write 25 words on this topic, I would say: IE6 IE6 IE6 IE6 IE6 IE6 IE6 IE6 IE6 IE6 IE6 IE6 IE6 IE6 IE6 IE6 IE6 IE6 IE6 IE6 IE6 IE6 IE6 IE6 IE6.
Reply Retweet Like
Rich Tibbett Dec 5
Good points here. I appreciate the current status quo of web platform development being split among truly independent projects has its merits but it is not without its problems too. Every web developer knows that pain.
Reply Retweet Like
Rich Tibbett Dec 5
Alternative take: One engine truly maintained by many companies. Governed by non-profit orgs. To iterate quickly. To block bad ideas. To develop competition around UI/UX. To stop competition around gate-keeping or preventing adoption of Web APIs for years. The Unix/Linux Model.
Reply Retweet Like
Rich Tibbett Dec 5
Don’t mind me of course. But if you squint hard enough at the future and allow enough latitude in its imagining surely we could see other approaches working equally well or better?
Reply Retweet Like
Rich Tibbett Dec 5
So my point comes down to could we make a fully collaborative effort around code work while negating all the downsides? That doesn’t exist today of course.
Reply Retweet Like
Fabrice Desré Dec 5
Replying to @annevk
Isn't it about browser market share instead? If a Firefox browser based on Blink ships without bad-feature-X, is that different from a Gecko based Firefox on that point?
Reply Retweet Like
Fabrice Desré Dec 5
Replying to @nnethercote @annevk
Do you really think the Chrome team is as stagnant as the IE6 one was? Looks like the problem is more that they are running so much faster than everyone else.
Reply Retweet Like
Valentin Goșu Dec 5
But is it in the right direction? For example, if we only had blink engines, would wasm exist, or would we be stuck with NaCl?
Reply Retweet Like
Valentin Goșu Dec 5
Replying to @fabricedesre @annevk
What is the incentive of shipping without a feature, even if crappy? We're still struggling to get rid of AppCache, which everyone agrees it sucks, but there's people still using it 😒
Reply Retweet Like