XML News from Friday, June 24, 2005

The Mozilla Project has posted the first alpha of Firefox 1.1, a.k.a. Deer Park. "it is being made available for testing purposes only for developers and the testing community. Current users of Mozilla Firefox 1.0.x should not download or use Deer Park Alpha 1. Note: Deer Park Alpha 1 is not an official mozilla.org final release, it has been made available for testing purposes only, with no end-user support. If that sounds scary, you'd probably be better off with the latest final release." New features in 1.1 for XML developers include:

• CSS @-moz-document selector for matching on site/document URL, useful in user stylesheets
• SVG support is now turned on by default.
• XML Events in JavaScript
• An extension to support XForms
• CSS 2 quotes support
• CSS 2 counters support
• Lots of neat CSS 3 features
• URIs are now always encoded in UTF-8

Other new featurs in 1.1 include:

• Sanitize "provides an easy way to quickly remove browsing history, cookies, cache, saved form information, and other personal data. The items to be removed can be customized, and the feature can be activated using either a keyboard shortcut or through a menu item."
• When viewing images, tab icons now display thumbnails of the displayed image.
• "Much faster session history navigation. The feature is off by default but can be enabled for testing purposes by setting the browser.sessionhistory.max_viewers preference to a nonzero number."
• FTP users are prompted for a name and password if anonymous access fails.
• Report a broken website wizard
• Changes made in the Preferences window now apply immediately
• Searchable download actions manager
• Searchable cookie manager

I tried out one of the earlier nightly builds on my Mac. It felt a little slower than 1.0.4 but otherwise quite stable.

Google has released version 0.1 of AJAXSLT, an open source implementation of XSLT in JavaScript, of all things. It's intended to be used in browsers. I'm skeptical. Why not just use the browser's built-in XSLT engine and maybe DOM level 3 XPath? Except for Opera and Lynx, I think pretty much all major browsers have XSLT support these days. Still, according to the README:

Safari/2.0 has XSL-T built in, but it is not exposed to JavaScript, but is only applied to XML documents that have a stylesheet declaration when they are loaded.

Internet Explorer exposes XSLT via the transformNode() method on the XML DOM. However, this is not avaiable if ActiveX is disabled.

Firefox exploses XSLT via the XSLTProcessor() object, however XPath is not exposed in the DOM. TODO(mesch): verify this.

Still, before reinventing the wheel, I'd prefer to work on improving the XSLT/XPath support in at least Firefox and Konqueror, the open source base for Safari. Implementing XPath alone is non-trivial, much less XSLT doing it in browser JavaScript strikes me as insane and unlikely to succeed. Just scanning the source code I can see lots of unimplemented functions, and lot so fplaces they made the same mistakes we made in Jaxen. For instance they've got the same bug in counting characters in strings that I fixed a couple of days ago in Jaxen. XPath and XSLT have a lot of weird corner cases to deal with.