Pagination will probably have to move to being built by build_pagination for accessability reasons.

The page was being first hit using the 'id' parameter, pieforms was building the form based on this. But the hidden element in the form was called 'viewid'. When the page was being hit again after form submission, the form was built again assuming 'viewid' held the view ID, which it did not anymore. Thus the form was built with and ID of 0, and the submit handler behaved appropriately - by creating a new view.

The small bug was that the 'save' button said 'new' when you were saving an existing view.

Now uses the centralised method of building the artefact chooser content, and uses extradata to make sure the correct page of results is being generated.

A much simpler template for the artefactchooser element now that the pagination deals with the number of results and each result is a different smarty template.

 * It now takes a 'class' parameter to set the DOM class of the pagination
 * It now takes an 'extradata' parameter for more stuff to send through to the paginator when javascript is enabled.
 * It now outputs the number of results in the paginated set. Not configurable yet, might be soon.

Centralised the building of the pagination related stuff for the artefact chooser, since it is used in two places - the artefactchooser pieform element and the json response script.

Broke out the pagination building into the view class, so the artefactchooser json script can also use it.

There is a bit of a hack to make sure the javascript for the paginator is available, by storing it in a global variable and accessing that later in the appropriate function. This will probably cause problems if more than one artefact chooser is being rendered on the same page, but I think that support for that should be disabled anyway. It will be fixed later when Pieforms goes to a more OO element format.

Now documented and cleaned up somewhat. Also takes an 'extradata' parameter of stuff to send back to the json script.

Corrected an oversight in the change to allowing the elements themselves to define their javascript.

Now that is happening, the element really has to be built first to give it the best chance to know what javascript it needs. This is definitely true in the case of the artefact chooser.

Added the artefactchooser pieform element, smarty templates and javascript file. Really rough first drafts.

First go at the paginator javascript module. It's pretty small, as most of the processing is done server side.

The parameter function makes sure that the variable matches /\d+x\d+/, and each digit is a multiple of five.

The breadcrumb function builds configurable breadcrumbs, with consistent HTML. A javascript library to make the links ajax-ified is coming shortly.

Added a method for building the url for a block editing page. Used as a base for pagination urls, and as a target for canceled configuration forms

Be more explicit about our checking for database types. This prevents random fatal errors if the appropriate database extension isn't installed, because ADODB isn't smart enough to test for this itself.

Also deprecate the 'mysql' database type, in favour of 'mysql5'. This brings it into alignment with the 'postgres8' type. The 'mysql' type will issue a log_environ error in 1.0 and crash in 1.1.

Added a new api function to return the javascript required to intialise the editor. Used in the views interface when configuring blocks and javascript is on

Fix a bug in setting the configdata for a block, and provide a better way for elements to export any startup javascript they need.

Because the configdata was not being set in the actual object (it was just being set directly in the database, the rebuilding of what artefacts were in the block was thinking that the existing ones were still in there. Now the config data is set()'ed and then the object is commit()'ed so that everything gets set properly.

Also improved the way elements export javascript they need to run. It can now be set in the context of an element, meaning the javascript can be personalised to the element itself.

Changed the signature of the get_icon method - it now takes the ID of the artefact to render, which makes generating little thumbnails of images possible

Implemented get_icon for the internal artefacts. Also changed the profileicon's render_self to render the image, for the views stuff.

The render_self method, which used to be an api method, is now used when rendering image/profileicon block types, to return the HTML for them.

Also implemented get_path() to allow retrieving of images of different sizes.

Used in the artefact chooser currently to show little thumbnails, but could be used elsewhere, in particular in the 'my files' section.

It displays any one chosen image fine, although the javascript pagination causes checkboxes to appear rather than radio buttons.

Currently there's a hardcoded limit of three images per page, for pagination testing.

Improved the error reporting of the XMLRPC auth plugin, by catching more specific exceptions and logging unknown ones. Added a TODO about a bad catch elsewhere