A web wizard is a web page or web application (group of pages) that is used to interactively construct or modify documents. The PIA is particularly well-adapted to this purpose because operations like <include> and <extract> make it easy to manipulate structured documents, and the ability to add specialized parsers means that such manipulations are not necessarily restricted to XML documents.
All of these tools would benefit from, indeed almost require, an overall framework for document management. This would be, essentially, a tree browser that lets the user browse to a directory, select documents, and apply tools to them.
In this category we have web wizards that build a document based on information supplied by the user via a form. The process will typically involve a template or skeleton document, with entities filled in from the form.
A web wizard that constructs the _subsite.xcf files,
Makefiles, HEADER.html files, and the like
needed for building and maintaining PIA-based applications.
A web wizard that builds customized but uniform home pages and secondary pages, for use in organizations (like CRC) that might want to give its employees home pages, but wants their pages to at least start out with a uniform look and feel. An example of this kind of form can be seen at advogato.org.
This kind of application requires a very customizable template, since each organization is going to want a different look and feel to their pages. In the PIA project we could easily start with a version of this for the Notes directory, and another for home pages under RiSource.org.
This one would build the ``chrome'' for a customized Mozilla-based web browser.
Many different kinds of configuration files (beyond the ones used in the PIA) would benefit from a form-and-template kind of approach. Note that web-based tools already exist for some; these are noted in parentheses below.
Makefiles
ant configuration files
In this category we have web wizards that modify an existing document ``in place.''
The simplest document editing wizard would be one that simply provides a text area for the entire body, with separate text boxes for the title and filename. This is, of course, crude, but it might be useful in some circumstances. It's closely related to Wiki servers.
More complex versions would include controls for adding tags, and an optional preview frame similar to the one in the Tutorial.
The difference between a web page editor and the page builder described in the previous section is that the page builder uses a form and a template, while the editor takes the page as-is and gives the user complete control.
The (existing) table of contents tagset (toc.ts) updates a table of
contents in a document. An associated wizard would make it easy to
add a TOC to a document.
A web wizard that constructs and edits pages full of links, in a couple of different formats (for example: single-column, annotated list (<dl>), tabular, ``quick-reference'', and so on. The ``quick reference'' format is exemplified by the following short extract from my personal start-up page:
- @ ^ [Java] docs ( JDK 1.3 api, JDK 1.2 api, JDK 1.1 api, JDK 1.0.2 api, Spec )
- ... Optional ( servlet (API) 3d ) /Gamelan/
- @ ^ [W3C] TR's ( DOM rec XML RDF HTML4 ) tags colors chars
- Open Dir: ( DOM XML server scripting )
- [Apache]( docs) [WebDAV]( [.org])
- /SGML/ [SP] [XP] [-tools] [us] /guide/ [SAX]
In this category we have web wizards that substantially transform or otherwise manipulate a document.
An indexer takes a document or collection of documents and prepares an index for use in a search engine or document browser. For example, the indices used in the PIA's directory index and tree browser.
In this category we have web wizards that build a template and its associated Web Wizard that, in turn, can be used to build documents.
In this category we have web wizards that output active XXML code. In its simplest form, this would just be an extension of a Page Editor with an extra set of buttons for adding DPS tags. That's less interesting than, for example, a tagset builder.
The simplest form of metaprogramming wizard would be a tagset builder: basically a page builder with templates for <define> tags and the other things that go into a tagset.
A more advanced form would allow different tag definitions, for example for decorations, to be included in order to customize the look and feel of the pages to which the tagset is going to be applied.
This category includes "builders" for simple forms, spreadsheets, worksheets, and calendars. The amount of ``programming'' permitted would be rather limited in most cases.
The output language of a metaprogramming wizard is not confined to the PIA's version of XML. It would be possible to build scripts in a variety of languages, including JavaScript and XUL, and in fact even Java and C++.