> Interesting that fixing "how to center a div" is considered harmful, but WebSerialPort is actually very good?
It is certainly "interesting", but "true" nonetheless: one determined person--think Fabrice Ballard if you want an example--is in a great position to throw together a web browser and even implement ALL of the crazy API wrapper specs, but when if they aren't you simply don't need most of them to browse any given website.
But, as it stands, my only a-few-year-old copy of Safari can barely even browse the web anymore as it is missing some new corner case of CSS or web components or whatever and I just get blank screens a lot; the result: people have burned years of large teams into trying to maintain implementations of HTML/CSS and have given up.
The web should really just be a handful of really core specs for getting platform access--which of course have innovated over the years so you'd have all of canvas, WebGL 1/2, and WebGPU, which would take SOME effort but isn't like, INSANE--and then all of the layout should be done end-to-end in libraries.
The world NEEDED to be like this to prevent us from ending up with only a handful of web browsers that can only be maintained by giant companies: it needs to be sufficiently easy to build a web browser that we would end up with a ton of small implementations that would be difficult to move as a unit, forcing progressive enhancement as a permanent norm.
How are you measuring this? Like, I would expect someone to want to ship e.g. WPF or something into the browser as their UI toolkit. Why would this fit in 5 MB?
So... you prefer the end result we got, with there being only ~1.75 browsers in existence--and only 1 that truly matters to developers--where ~1.66 of them are owned by companies that would prefer to implement this specification? :(
I think, given your history, is that what you’re looking for is apps but distributed on the web. However I believe there is also a market for app clips of sorts that are meant to be more lightweight and have some default APIs available to them, for cases where people don’t actually want the overhead of apps.
It is certainly "interesting", but "true" nonetheless: one determined person--think Fabrice Ballard if you want an example--is in a great position to throw together a web browser and even implement ALL of the crazy API wrapper specs, but when if they aren't you simply don't need most of them to browse any given website.
But, as it stands, my only a-few-year-old copy of Safari can barely even browse the web anymore as it is missing some new corner case of CSS or web components or whatever and I just get blank screens a lot; the result: people have burned years of large teams into trying to maintain implementations of HTML/CSS and have given up.
The web should really just be a handful of really core specs for getting platform access--which of course have innovated over the years so you'd have all of canvas, WebGL 1/2, and WebGPU, which would take SOME effort but isn't like, INSANE--and then all of the layout should be done end-to-end in libraries.
The world NEEDED to be like this to prevent us from ending up with only a handful of web browsers that can only be maintained by giant companies: it needs to be sufficiently easy to build a web browser that we would end up with a ton of small implementations that would be difficult to move as a unit, forcing progressive enhancement as a permanent norm.