Hacker News Reader

Comment by WilliamLP | Hacker News Reader

Doing anything in JS and Canvas seems exactly like programming in the early 90s. I think Commander Keen on a 386 may have had more performance headroom than this! This is what is supposed to save us from proprietary tools like iPhone and Flash? I really don't get it.

View on HN

Programming in the early 90s felt like the Wild-West - anyone could strike gold, anyone could make pickaxes, and anyone could build a town around a set of rail-tracks. Almost all of the large game companies today started as small programming teams in the 90s. I don't know what could be more exciting for startup.

The fact that feature-rich tools, IDEs and libraries don't exist for JS/Canvas games is EXACTLY why I love it. There is so much opportunity here for those who love to build and innovate. If you don't believe that JS/Canvas has a future, I can't change your mind. But if you think that there is a future for browser-based games running JS, then this a very exciting field to jump into.

Before prototype, mootools, and jQuery etc., JavaScript frankly sucked. It was alerts, statusbar texts, and title bar changes. I see Canvas being in that category right now. There is a lot that needs to be done before you will see a thousand "Top 10 tricks to speed up your Canvas game" posts but that's exactly where the most exciting part lies.

mootools made javascript a better language? Javascript is a wonderful language regardless of the availability of good tools and libraries.

He didn't say the language sucked, he said 'javascript' sucked, which includes the language and what people do with it.

"Before prototype, mootools, and jQuery etc., JavaScript frankly sucked."

Proto, mootools and jQuery are written in JavaScript. So your statement compiles to "JavaScript sucked until people figured out what it could do."

(Which is essentially your larger point, but still.)

> "JavaScript sucked until people figured out what it could do."

Actually yes. Do you remember any amazing JS apps from 2002? IE6 has been around since 2001 and even on slow machines, it can still run most of the JS apps (provided you make the fruitless effort to support it). Where were the JS apps in 2002? Only things people used JS for were form validation and annoyances.

I know these frameworks are written in JS. What I meant to say was that very few people had actually used JS to its full potential and capabilities till just 5 years ago when protoype/scriptaculous/RoR and apps like Google Maps started to show up. I wrote an Ajax-enabled app for IE6 in 2003 and it was so much more work than $.ajax({url:foo,callback:bar});

There was the nice "asteroids" microsoft DHTML demo. But Microsoft was afraid of DHTML...

> Do you remember any amazing JS apps from 2002?

Yes, I remember Outlook Web Access providing a desktop-caliber mail client in the browser using ajax and javascript.

for rather low values of desktop-caliber

I think all the hubbub is because the browser has dominated computing in the past decade. Being able to play a game on both your web browser and your smartphone with only minor changes (input, screen scaling, and perhaps locked fps) is a big deal because everyone knows how to use a browser and the barrier to accessing your game is as trivial as currently possible. Having to actually install a game, even if it's just downloading and running an installer, is apparently a lot more annoying to users than you'd think.

Unless some kind of native backend is written to make HTML5 apps just as responsive as plugin-based ones, I don't really see how this is going to be successful for games. The performance hit is just too gargantuan.

I think the performance won't be a big problem in the foreseeable future. There's some great competition going on between Safari, Chrome, Firefox and Opera. With every new version their Javascript engines get faster and with the coming WebGL we'll have hardware accelerated drawing in <canvas>.

This competition is precisely what was missing for Flash.

I think that Javascript, the language, itself has barriers to performance. I think few programmer know, for instance, that a dictionary lookup is 100 times slower than a switch statement. (If you don't believe me test it! It's what I found under Firefox anyway.) The language absolutely begs for a real map type... Objects just don't cut it because of performance (it has to be concerned that someone might have extended the Object prototype) and because of their limitations, e.g. confusion if someone defines a key string called "hasOwnProperty".

If you actually try to use some of it's nice functional features, as well, you can forget about performance. Maybe you think engines can solve these things, but I doubt it will ever compare to any language built from the ground up to have reasonable performance.

Even so, browser games as-is are bottlenecked by render performance, not language performance; both platforms are capable of running Quake(1 or 2), but JS holds the advantage as far as running it at reasonable resolutions and framerates, while the Flash implementations are hobbled by software rendering.

Even that bottleneck is going away; see: http://en.wikipedia.org/wiki/WebGL

Not quite powerful enough to run, e.g., Doom 3 or something.. but headway is being made.

Reply on news.ycombinator.com