原文：The Next-Gen Web: Browser Storage Support。

下一代的 web 已经开始上路了，就在这个星期，MySpace 集成了 Google Gears，雅虎发布了新的 BrowserPlus，Google 的浏览器版三维地球也上线了。类似 AIR、S…

http://www.appcelerant.com/lipstick-on-cgi.html

For another open source effort in this direction, check out:

http://jnext.com

Now, on the client, how far are we from eschewing the need for a heavy OS if the most popular browsers can add these common services, while the VM vendors provide bootload and hypervisors and minimal console.

The real promise of RIA is to provide the advantages of client/server architecture with added reach and without the ownership costs, because you just update your app on a Web server and everybody is automatically running the new version. So then the next frontier is getting a powerful enough client-side platform to be able to run a client/server-class app. AJAX is looking a bit ragged around the edges for this purpose, and Adobe, Microsoft (Silverlight), and Curl are all competing to be used as this platform.

Back to the future, with Javascript is the C/S programming language (but terrible to debug with) and the browser as the C/S programming host (as opposed to the Windows class libs or VB Forms stuff).

Richard: I also think that standards adoptions has been very good nor quick enough. The whole reason Google built Gears was because they were frustrated with browser development and wanted to leap ahead to HTML5. So thankfuly we might see plugins for browsers to fill the gaps where a standard implementation is missing.

Also at least Firefox, Opera and Webkit seem to be reading from the same song sheet now, which is a solid start.

I will check out Curl shortly - I have heard of it but thanks for the pointer and your comment.

http://www.whatwg.org/specs/we.....#security7

but it deals more with application security (eg. SQL injection etc.).

In terms of the local security you guys have mentioned, the model is the same as what it is with cookies. In the same way that when you are at a net cafe you wouldn’t use a cookie to remember your logged in session, you also wouldn’t enable Gears (or it wouldn’t be installed) so you would revert back to a non-Geared app.

The database itself is stored within the users operating system profile (eg. on OSX or Vista/XP), so by default only you when you are logged in should have any access to it.

In saying that though, I do expect that with these new applications you will see a re-hash of the same security and privacy concerns that cookies experienced years ago. It might take 2-3 years for most users to be comftable with both web apps integrated with the desktop, and also with placing their files and content up in the cloud.

App developers just need to be wary of these concerns and not ‘force’ Gears as a requirement for using a web app. They will also need to allow users to not use Gears even if it is installed (in the same way as when you login to a site, you can choose ‘remember me’ or not..). I expect that a whole set of best practices in this regard will start to form and the terminology (eg. ‘enable desktop’ or something like that, on login) will begin to form.

If you look at how Google Reader uses Gears, they have some of this. Interestingly, I went back to MySpace and they don’t let you switch it off again- which could be a concern as per what has been said above.

How do most people feel about having something like Gears store their data?

My confidence in the ability of commercial vendors to properly support browser standards is pretty low. Just look at Ajax which consists of standards like JavaScript (ECMA), HTML/XHTML (W3C), DOM (W3C), and CSS (W3C). Ajax requires libraries to bridge the incompatibilities of these technologies as implemented among all of the browsers (e.g. IE, Firefox, Safari, Opera, etc.). It’s been several years since these technologies were introduced yet browser compatibility is still poor - better than the DHTML days but not much.

I believe in standards but only if the vendors and open source projects that implement them do so in a consistent manner. Without consistent implementations a standard has limited utility and terrible portability.

Products such as Java (Sun), Flex (Adobe), Silverlight (Microsoft), and Curl (Curl, Inc.) all use plug-ins defined and distributed by a single vendor because that seems to be the only model that works if you want to get a portability across browser vendors. We learned this the hard way with Java applets and DHTML in the mid to late 1990’s. I suspect that Web Applications 1.0, while it may be technically sound, will never be implemented consistently across browsers and won’t be the silver bullet many people think it will be.

I wish I was wrong about the failure of browser vendors to implement standards, but I’m not. They just can’t do it. You have to have a plug-in provided by a single vendor or open source project in order to get consistent behavior across browsers.

All the best,

Richard
VP of Developer Relations
Curl, Inc.

Excellent start to what should be a very good series of articles. I would encourage you to take a close look at Curl which is a RIA platform that has offered desktop storage for years. Curl is both a browser RIA solution as well as a desktop runtime similar to Adobe AIR only a lot more secure and much, much faster. It’s also the second most mature RIA framework available today - only Java has been around longer.

Anyway, please check out Curl before writing other articles as I think you will find it really is an excellent platform.

All the best,

Richard Monson-Haefel
VP of Developer Relations
Curl, Inc.