Hacker News Reader

Comment by potch | Hacker News Reader

So when a user types '-webkit-', and doesn't type '-ms-,-o-,-moz', when they all work, that's laziness on the vendor's part? Or better yet, when the syntax 'linear-gradient' works and you don't need a prefix at all, and you don't bother because you can't be bothered, that's laziness on the browser vendor's part? The vendors have bent over backwards to be fully compatible on gradients, animations, transition, border radii, box shadows, keyframes, and boat loads more every day. Every vendor makes compromises, the prefixes get removed and the whole web moves forward as one unified. But you'd never know it because hack developers can't be arsed to at least open one other browser to see whether it even worked. But yes, the vendors are lazy.

View on HN

While the w3c likes to think the univrse revolves around it, html5 has several competitors and given that the syntax for a particular css property doesn't matter much, they should have just standalized it the first time a browser added that feature.

Beside, why use a non-webkit browser?

What are html5's competitors? Are Opera and Firefox not even notable anymore?

Flash, native apps, Silverlight.

It's not CSS, but the <canvas> tag was invented by Apple for Safari, and was originally just a thin wrapper around the OS X drawing API, which was (thanks to OS X's heritage) basically just the PostScript drawing API, including all the various Porter-Duff software compositing methods (because they were all naturally supported by the API, so why not).

It turns out, a decade later, that most platforms don't want to render PostScript-compatible graphics in software, they want to render graphics with hardware designed for APIs like OpenGL and Direct3D, which don't support all those compositing modes. In particular, I think IE deliberately does not implement most of the <canvas> compositing modes because they can't be hardware accelerated by Direct2D. So "just standardizing the first browser to add a feature" can lead directly to brittle, incompatible implementations, and that hurts the Web for everyone. Sure, you don't want people to wait decades for some feature or other to become widely implemented, but putting a little thought in before standardizing things helps a lot.

> Beside, why use a non-webkit browser?

Man, when you put it like that, why did we ever leave IE6? You could write your HTML just once and be sure it would work everywhere!

>Man, when you put it like that, why did we ever leave IE6? You could write your HTML just once and be sure it would work everywhere!

We left it because a) it stagnated and b) it didn't work on all platforms.

Surely not because of some zeal to support multiple independent implementations.

Since Webkit has neither problems, seeing that:

a) it's open source portable to all platforms with ports existing for all major platforms and tons of minor ones.

b) it's an engine, not a browser, and as such supports several widely varying browser GUIs, from Safari to Chrome to mobile browsers to command line tools.

c) It moves at a fast pace with lots of companies contributing to it, from Apple, to Google, to Nokia, to Adobe, to Samsung, to RIM to ensure that.

It doesn't make much sense to use anything else.

You could achieve all the dreams Mozilla has AND use Webkit at the same time. Just add your other stuff on top of it, instead of in a completely different implementation.

I'm not even against experimenting with new engine ideas. If you need to create a completely different engine do it, but only because there is an important difference and reason to do so.

Today we just have different engines not because of some special differentiating factor or novel way, but just because a few companies hold to their old engines from the early days of the Web (namely Mozilla, Microsoft and Opera). Almost everybody else that came after Webkit has adopted it.

Some things just make sense to have one single and common base. The same way most of us use, say, OpenSSH, and don't think much about it.

No software is infinitely malleable; no matter how much flexibility or portability you design into a code-base, it'll still only really work in situations that at least resemble the ones in which it was originally designed. Currently, the situations Webkit was designed for are a really good match for the situations we have (desktops, laptops, tablets, phones), and it's very difficult to find a situation where Webkit isn't a great choice. However, if there's one thing to learn from history, it's that situations change and usually in unpredictable ways. I have no idea what possible up-set to the computing landscape could arrive and make Webkit unfeasible, but I'm pretty sure something will change eventually, and wouldn't it be nice if we had a backup plan?

Also, consider what might happen if Webkit were the only browser engine. Sure, Webkit moves at a fast pace now, when it has competition from both other browsers and native apps, but consider that the two biggest supporters of Webkit both benefit directly from people using native apps over web-apps. Sure, they benefit from web-apps too, but Apple in particular gets actual revenue directly from native apps, so if they didn't have other browsers to compete with, maybe Webkit wouldn't be as shiny and cutting-edge as it is. I mean, it already renders existing websites, and anything that needs abilities the current web-platform doesn't support can just be implemented as a native app, right?

> If you need to create a completely different engine do it, but only because there is an important difference and reason to do so.

The thing is, if Webkit were the only browser-engine, even if you did come up with some completely different engine, your idea would be useless. You wouldn't be able to achieve compatibility with existing sites without mimicking all the weird bugs and corner-cases of Webkit, which would not be feasible, and if your idea is so completely different you can't just fork the Webkit code and add your idea on top.

> You could achieve all the dreams Mozilla has AND use Webkit at the same time.

"all the dreams Mozilla has" is pretty much "make sure there are always multiple independent, interoperable implementations of the Web technology stack", so no, that wouldn't really be possible. :)

>However, if there's one thing to learn from history, it's that situations change and usually in unpredictable ways. I have no idea what possible up-set to the computing landscape could arrive and make Webkit unfeasible, but I'm pretty sure something will change eventually, and wouldn't it be nice if we had a backup plan?

Sure, but that "backup plan" wouldn't be another similar engine codebase, like Trident or Gecko. I can't fathom any future technical case where those would do but Webkit would not.

But I think I covered that in my comment concerning alternative engines. Make one if you have a real need to implement something different to cover a new situation or provide something not there today. Not merely to keep your own mostly same implementation that introduces slight incompatibilities and standards lag. Current alternative engines do the latter.

>Also, consider what might happen if Webkit were the only browser engine. Sure, Webkit moves at a fast pace now, when it has competition from both other browsers and native apps, but consider that the two biggest supporters of Webkit both benefit directly from people using native apps over web-apps. Sure, they benefit from web-apps too, but Apple in particular gets actual revenue directly from native apps, so if they didn't have other browsers to compete with, maybe Webkit wouldn't be as shiny and cutting-edge as it is.

I don't think this is true. For Google is isn't true at all: it makes more money from people using the web than from people using native apps. As for Apple, well, they started the whole webkit project, and they don't make much from native apps anyway. They make their money from devices (which are used to browse the web also, so need a top notch engine) and laughably less from selling native apps (it's just a side business for them). Plus, all this time with native apps and the like, they have enhanced Webkit and mobile Safari the same.

But, let's say that what you say is true: if some companies indeed favored native apps instead of web apps, wouldn't that make even more important that all other companies that care about the web flock to maintain ONE and only engine, instead of working each in it's own engine?

That way they could make it as good as it can be, remove compatibility issues and generally make the web attractive and achieve progress and new features faster.

>The thing is, if Webkit were the only browser-engine, even if you did come up with some completely different engine, your idea would be useless. You wouldn't be able to achieve compatibility with existing sites without mimicking all the weird bugs and corner-cases of Webkit, which would not be feasible, and if your idea is so completely different you can't just fork the Webkit code and add your idea on top.

The thing you describe as bad is actually BETTER than what you have to do today.

Because today, if you come up with some "completely different engine", in order to "achieve compatibility with existing sites" you would not just have to "mimic all the weird bugs and corner-cases of Webkit" but also those of the other major engines.

Better to have just ONE engine to mimic, no?

> Sure, but that "backup plan" wouldn't be another similar engine codebase, like Trident or Gecko. I can't fathom any future technical case where those would do but Webkit would not.

Certainly, any situation that made Webkit unviable would probably affect Trident or Gecko or Presto the same way. It's not that we specifically need to preserve some other code-base so that we can bust it out if something happens to Webkit, it's that we need to make sure that it's possible for new rendering engines to spring up and keep applying the pressure of competition.

> Because today, if you come up with some "completely different engine", in order to "achieve compatibility with existing sites" you would not just have to "mimic all the weird bugs and corner-cases of Webkit" but also those of the other major engines.

If the majority of sites each only worked in one major browser, or if they served completely independent markup to each browser, then yes, you'd have to reverse-engineer them all to obtain decent compatibility. However, that's not the case these days - the vast majority of sites stick to the parts of the Web platform that are reliably, consistently, interoperably implemented across the major browsers.

If something has been implemented correctly four times, even if it took a while to reach that milestone, it's much, much, much more likely that it can be implemented a fifth time, than some feature that has exactly one implementation and it's not clear which behaviours are intentional, which behaviours are accidents, and whether the accidents must be preserved for compatibility reasons.

I understand and appreciate your concern with regards to the canvas, but that is properly also the single most complex new element.

That is why I was commenting on css properties -- because it really doesn't matter whether you write it as background: linear-gradient(...) or background: gradient(linear, ...) really doesn't matter; the most important thing is to make it work in as many browsers as possible.

Hmmm, yeah, we tried that 10 years back. Didn't work.

Reply on news.ycombinator.com