Is there any reason not to use class or id?

After some tinkering, it’s possible, but not sure its desirable.

Would have a quicker template engine if it didn’t have to look up template variables for every class or id it sees.

The tinkering code is at http://homepage.ntlworld.com/jared.williams/php/template.php

2 classes, TemplateCompiler which compiles a html template down to a Template instance which could be cached to save the XML parsing/processing.

Several limitations in the implementation, like no attribute values can be changed (due to limitations in XmlWriter)

There is really no need to use standard attributes in the template source files as long as it conforms to the XML standards. Those attributes are removed at render time and do not get rendered to the browser anyway. There could be an argument to use namespacing but for now it seems to work ok without all that over-complication.

The main aim was to produce a template language that a designer could learn fast then create/maintain templates for the developer with minimal developer consultation.

With the three tag language it is easy to explain how things work:

* VAR: text and attribute values are replaced
* CHOICE: Element blocks are shown or hidden
* REPEAT: Used for all repeating data (tables. lists, results, etc)

@uhh
XLS solves a different problem really. And quite frankly, I think that it is a good example of over-engineering.

@Godar
I took a short look at your project. I see that you use non-standard attributes in the HTML. Is there any reason not to use class or id?

Its aim is to ensure designers can work in along side the developers.

The main advantage is the HTML prototypes produced by web designers are actually usable and can even be maintained by the same people during production, without the need of any php knowledge.

Designers around the world will undoubtedly rejoice. Finally creating templates wil be p*** easy for those hallowed and somewhat mythical creative types who are masters in the dark arts of CSS and HTML/XHTML standards with which they regularly transform simple paragraphs and headers into veritable disneyland like user interfaces, nonchalantly float divs with pinpoint accuracy through the gauntlet of CSS positioning, bestow their Utopian semantic vision upon us with weighty intellectual authority, but somehow lack the mental capacity or nous to handle a couple of echo statements.

I’ve used XPATH queries to select elements where content should be injected. ID’s are also supported.

The main advantage is the HTML prototypes produced by web designers are actually usable and can even be maintained by the same people during production, without the need of any php knowledge.

I (even more) recently posted an updated version of the pQuery library that allows you to keep the presentation generation code and the presentation markup in the same file, here’s the post: http://loveandtheft.org/2008/09/20/php-templating-in-53/

I still relies on the DOM-structure, meaning you’ll be doing something like: $root->li->something = "Something else"; But I have a local development version that is complete with a CSS selector, so you techincally can do this instead: $root("#post-list li")->title = "My title for something"; which puts an abstraction layer in form of dom #id’s and css-classes allowing you to move them around (in effect changing where the output gets pushed into the template without changing the php attached to it).

/ Fredrik Holmström