The Web 2.0 Expo is in full swing in San Francisco, CA. One of the more interesting panels to take place earlier today was called What to Expect from Browsers in the Next Five Years. The panel’s roster included some of the biggest names in the browser industry, with representatives from Palm, Yahoo, Mozilla, Opera, Google, and Microsoft. Below are my notes from the talk (quotes are paraphrased).
Douglas Crockford (Yahoo) – Architect at Yahoo, discovered JSON
Charles McCathieNevile (Opera) -Chief Standards Officer at Opera. Co-chair of the W3C WebApps working group
Alex Russell (Google) – Develops Chrome Frame plugin, Helped create Dojo Toolkit
Giorgio Sardo (Microsoft) – Web Technical Evangelist
Dion Almaer (Palm) — Director of developer relations at Palm. Co-founded Ajaxian.com with (fellow moderator) Ben Galbraith.
Ben Galbraith (Palm) — Director of developer relations at Palm, co-founded Ajaxian.com.
Q: How many people would like to see IE9 implement canvas?
*Everyone raises hand, including Sardo, who works for MS*
Sardo (Microsoft): We’re not done yet, we are investing a lot in HTML5. We believe in HTML5. We believe we need to have a professional grade HTMl5 implementation. We look at developer feedback, we look at the spec, we look to make sure it will be consistent, we look at performance. Everything will be hardware accelerated.
Almaer (Palm): In my opinion it’s tiny to do Canvas and SVG is huge..
Eich? — It’s true SVG is huge. Canvas is pretty small. We have implemented it for five years now, something like that. It’s easy.
Almaer (Palm)- How do we stop what happened with IE6 from happening again where we get stuck in time?
Russell (Google) – I’m excited for IE9. I can’t wait — it has hardware accelerated SVG, hardware accelerated rendering. Competition is great. That’s the baseline — that’s where we want the competition to be. We want browsers to be looking to the future to see what users want and need. When the system is working, and browser vendors are shipping features quickly. In 2001 IE6 was fantastic. The problem is that things stopped getting better. We can get stuck in a generation. Plugins are one way to get out of the hole, I don’t think they are a long term solution (Russell works on Chrome Frame, a plugin).
Q: As a browser vendor facing all these specs. How do you prioritize?
McCathieNevile (Opera) - You talk to developers. Look at what people are using. A little bit of it is opportunistic — you’ve got a guy who wants to do it, he just does it.
Eich (on how Mozilla approaches this) - Mozilla has been open source for over 10 years. We have devs come to us and actually patch support for things. We’re seeing that now for things like Web Forms. We have a lot of HTML5 already implemented. What I most value are web devs who maybe can’t contribute to C++ code, but can tell us what’s missing.
Q: Doug, you’ve talked about things needing to change…
Crockford (Yahoo): The web was left for dead in 2000. Microsoft believed as many other did that the web was finished, just like Hypercard was finished and that we were going to go on to something else. There were several alternatives offered (Flash among them). Then the web in 2005 took off again with AJAX. The browser is the world’s most important app delivery system. Unfortunately because it was left for dead and WC3 abandoned its role as steward of web standards. We have the same web standards we had in 1999, which were not even state of the art then. Devs are trying to move forward with something that is clearly inadequate. We need to first solve the IE6 problem. A sig. portion of the world is on IE6 and it’s not changing. Five years ago I said it would, and it hasn’t. It hasn’t happened because web devs are doing such a good job supporting users on IE6! It’s worse internationally. In some markets , 40-60 percent on IE6.
The IE6 problem has to be solved by major website devs. One day all of us redirect to a page that says, hey, try one of these browsers. We all have to do it on the same day, otherwise we get worried that we’re sending them to our competitor. I propose that day is 30 days after all modern browsers fully implement ES5.
? – A lot of this is Active X customizations, if you can’t replace those, you can’t replace IE6…
Crockford – Another big problem in China is a lot of people aren’t using licensed OS’s [they are using pirate copies of Windows]…
Sardo (Microsoft) – Windows update is our agent to update users. Most of what is updated is checked for proper licenses. IE is an exception — even if it’s pirated users can download the new update… IE9 will not be supported by Windows XP. We are building all of HTML5 with hardware acceleration. We need a modern OS to do this.
Russell (Google) – I recognize Opera and Mozilla and on Chrome we’re all doing hardware acceleration. And, all of us are doing it on XP. What you’re describing is a situation where people are less behind. The way is to not leave users behind. The question is, do we have a plan so that we can give devs a choice to use HTML5 across the board?
Crockford (Yahoo) – I recommend all users of XP migrate to something that isn’t IE.
McCathieNevile (Opera) - There’s a lot of hardware that isn’t modern hardware.. that’s why people aren’t moving away from XP. A lot of people don’t want to have to keep up with technology.
Q: I want to talk about browser competition. For a long time Mozilla used Firefox were positioned as the alternative to IE… now that other browsers innovating, is Mozilla’s goal to still drive FF’s share up?
Eich (Mozilla): We have a mission to preserve choice and innovation, and to make user king or queen over their experience. There’s more competition now. But… the problem is that those companies (Google, Apple, etc) have agendas. Apple makes great products but they want to control what SDK or API you can use. Google is more aligned with the open web, but they have to have an agenda because of search. With Mozilla we don’t care. We blind ourselves, we don’t see the data, we will never do behavior marketing. Look at what FB has done recently.. users need to have control over their data.
Q: What’s the state of JS?
Eich (Mozilla) – The committee is operating in a mode called harmony which means we are not fighting. We’re prototyping proposals for Harmony. We’re working pretty aggressively on things like the module system. Doing things to get rid of host objects. The DOM kinda sucks..
Russell (Google) – Interpreters are getting faster. It’ all getting faster. The bits that aren’t getting as fast as we would prefer is network behavior. Google has introduced a new protocol called SPDY. Fundamental latency isn’t getting as fast as local device capabilities are. Hopefully in process of making a better browser we also upgrade the DOM itself.
Sardo – All browsers have pretty good JS performance (it has room for improvements).
Q: I think we should look at some comments made by Joe Hewitt this week.. He recently said over Twitter, “I love the web, but it sucks”. We see proprietary platforms and how fast they are evolving. Joe said this loosely coupled system the web uses to establish standards isn’t working.
Russell (Google)- Respectfully (I’m an enormous fan of Joe’s work). I feel his pain. I worked on a JS toolkit, that’s as bad as it gets. He’s done the same thing. He built iUI. I feel like a lot of his comments rung true about two years ago. Things are getting much better in webkit land. A lot of the things iUI did are solved. CSS animations. Those things are coming. Rate of improvement is increasing so much I think it’s easy to understand that things were stuck. It’s not all better yet, but it’s getting better at a faster pace..
Q: Seems real action is in mobile browsers/touch interfaces. Where do you see that going?
Eich (Mozilla) – We’re investing heavily in Firefox on mobile. I agree it’s going to make desktop worry. I have to say, multi-touch out of Apple is great, I think it should be rapidly standardized. A lot of the stuff that happened on the iPhone was because (even though I believe Apple wants Cocoa to be first) they’ve done work to support the web. That’s been helpful. What Joe is talking about — there’s a problem where you get a lot of people in a room, it’s not a good place to design new things, to innovate. So you need to just do it, innovate…
Q: What was logic behind Opera’s mobile browser on iPhone (using Opera Mini vs the full fledged Opera mobile browser)
McCathieNevile (Opera) - We’ve shipped hundreds of millions of browsers on mobile. Opera Mini, which is what we put on the iPhone, was designed in part for markets where people haven’t got an iPhone, but where they have a crappy handset. They want to get on the Internet. It wasn’t an iPhone strategy. It was, this is browser more people use than any other. It’s a strategy for the developing world.
Q: In the 90′s we only had to deal with 3 major browsers to make our sites compatible with. Now have to deal with half a dozen.
McCathieNevile (Opera): An important part of standardization is to write better standards. And handle backwards compatibility. In HTML5 we didn’t want to turn over web and start again, we wanted the stuff to keep working.
Crockford (Yahoo) – The thing I’m most worried about as we move into mobile, is that we have a strong risk of losing openness. Either proprietary app platforms win because the web can’t innovate fast enough. Or the web gets captured in a proprietary platform and someone else decides what you can run on devices you’ve paid for.
[Sardo (Microsoft) talks more about standards and the possibility of switching between rendering engines to maximize compatibility. He sort of ignored Crockford's statement, which was frustrating because it's a major issue]
Russell (Google) – There are economic costs to keeping old content working with new browsers. And orgs that act as mediators for old tech for users have costs, deciding what has to be left behind.. Switchable renderers one potential way where new stuff can be added to ecosystem faster.
Q: No one is talking about CSS feature detection.
McCathieNevile (Opera) - What you’re supposed to do is use CSS to take existing content that runs on anything and make it pretty. It’s nice. But you can do without it so maybe that’s why you can’t detect if this works, or at least that’s the logic for why there isn’t feature detection. But once you have people in shops actually coding it, it might be that was a dumbass idea.