XML in PHP5

[Note: Again, this list is for the development of the PHP5 XML extension and related ideas. This is not a user support forum, ask user support questions on php-general@lists.php.net] Hi, This list was created to discuss the design and implementation of a "new" XML extension for PHP5. Christian and I were talking, and we have many of the same ideas for how this extension should work. This message however, reflects my views alone. Don't blame Christian if I offend you, or say something incorrect. :-) In the past, PHP's XML support has been focused around "a new extension for everything." You want DOM, ext/domxml, SAX, ext/xml, XSLT, ext/xslt. However, XML and the related programmatical API's are meant to integrate with each other seemlessly. ext/domxml has started to take this direction, it supports in some variety or another, DOM, Xpath, Xpointer and XSLT. However, the current extension is flawed in both design and implementation: - Design. The API does not comply with the W3C specification, although it does mimic it. XSLT, XPath and XPointer support are not implemented coherently, but rather are an afterthought. - Implementation. The extension is poorly designed to take advantage of the new features in Ze2. The Ze1 implementation is leaky and a bit more bloated than necessary. For example, memory is not free'd "on-destruction" but rather on request shutdown, which makes parsing large XML documents (or a large number of XML documents) quite hard. For PHP5, I'd like to take the domxml extension, and rework it (still using the libxml and libxslt libraries) and create a "über-extension," which incorporates libxml and libxslt to become the default (and only) XML extension distributed with PHP. The first step is planning and outlining development. Let's start by brainstorming. After that, we'll worry about creating an API to support our ideas. In this extension, I think the following features should be supported: - SAX 2 compliant API. - DOM 2 compliant API. - XPath 1.0 compliant interface. - XSLT 1.0 compliant interface. - XPointer 1.0 compliant interface. - XInclude 1.0 compliant interface. - Streams Support. Streams for accessing documents should be fully supported so that, for example, someone could easily access and parse a compressed XML document using the bz2 stream wrapper. - Extension level interface (the libxml and libxslt libraries should be seperated from the extension itself). This comes with another idea, bundled libraries should be separated into php4/bundle and bundling should be a separate option from enabling or disabling a particular extension. - Encoding Options and MbString Integration. - Compatibility API. DOM 3 and Xpath 2 and XSLT 2 are all coming out "soonish," and they will be breaking compatibility. We need to have a built-in way of dealing with these evolving standards. - Proxy Objects. Provide the user the ability to hook into transformation stages. For example, when performing an XSLT transformation, they should have the ability to modify the DOM object *while* the translation is occuring. Thoughts? Comments? Questions? Additions? -Sterling -- "A business that makes nothing but money is a poor kind of business." - Henry Ford