XMetaL: Wouldn’t it be loverly?

December 18, 1998

Liora Alschuler

SoftQuad’s XMetaL, shown at XML ’98 in alpha code and scheduled for release in the first quarter of 1999, takes great strides toward a structured editor that builds reusable content without sacrificing the cushy user interface expected in word processing GUIs. XMetaL uses CSS 1 and 2 for display and file export, and SoftQuad will be adding print support, although support for traditional publishing will be less than in a word processor such as Microsoft Word. XMetaL will support use of both SGML and XML DTDs and will export SGML, XML, and HTML.

XMetaL requires setup of a *.ux file for each DTD, although it will come with configurations for DocBook and a small number of additional doc types. The configuration file maps constructs in the DTD to familiar GUI objects like enumerated and bulleted lists. Tool icons and dialog boxes can be customized for each DTD so that they change when a different doctype is loaded. XMetaL has a COM interface to control the application, supports the DOM for document component access, and has a scripting engine for JavaScript and VBScript as well as an ActiveX engine that can interface with COM objects.

Peter Sharpe, Vice President of Development, described user-comforts such that when you hit the Enter key, "something has to happen", typically the insertion of a similar or subordinate element. Likewise, when you hit a character key on the keyboard, a character appears. This may not seem exciting to a generation bred on Speak-and-Spell, but SGML editors have been ruled by content models that don’t allow plain text at every point in the structure tree. XMetaL respects content models and inserts text at the next allowable spot. Tags can be changed using a drop-down box of tag names even if the element is not highlighted in the same manner in which a paragraph style can be changed in Microsoft Word.

A HoTMetaL-like Resource Manager provides drag-and-drop access to reusable content including text entities, graphics, boiler-plate text, macros and style sheets. Document preview is linked to either Netscape or Microsoft IE.

Soft Spots.
SoftQuad demonstrated simple, visual table construction with expected, MS Word-like behavior for the CALS table model and release one will also support the HTML table model. While, like everyone else in the industry, they recognize the limitations of these format-driven table models, they stated that the difficulty of setting up a visual editor for semantically-defined tables will prevent them from doing the right thing, at least in release one. As data exchange moves to XML, the inability to create and display even a simple no-spans-allowed data array strikes us as a serious limitation. Users can use scripting and the *.ux file to create dialog boxes for forms entry and perhaps that can provide a short term work-around for some applications.

Another soft spot in the alpha version that we saw is that it provide no structure view, not of the DTD, not of the document. Users of the current generation of SGML editing tools report that they often edit exclusively in the structure view. It seems that if one expects a writer to make the extra effort required to input tags and make a coherent, consistent document hierarchy, one ought to provide a graphic view of the result. A related soft spot is that there is no easy method of document navigation using the document’s table of contents or tree structure, although one can expand and contract tagged sections to create an outline of variable depth.

Priced at $495, SoftQuad is aiming XMetaL at three markets: first, they draw a concentric circle around the tech pubs departments representing the engineers, marketing managers and all those who contribute information to the publication, but are not full-time documentation specialists. These folks would not climb the SGML editorial learning curve, but could pick up XMetaL, customized for their DTD, and be productive, eliminating a data conversion step. Second, SoftQuad believes it can expand this basic market with a lower break-even point for XML documents by reducing the complexity of DTD-specific configuration. Third, they see a large, new audience for XML on the Web, although they approach this group of HTML users with caution because it is too early to say how XML will fit into Web publishing.

When the target is print, clearly XMetaL need to be pipelined with specialized layout tools. Starting with CSS layout strikes us as a good first step for Web authors increasingly familiar with its simple requirements supported by current generation browsers.

It would be nice if it did simple office tasks, like email documents with a generic or rtf-like DTD -- these seem to fit the defined market space because they don’t have high publishing and layout requirements, and don’t justify any setup time for DTD building or DTD-specific configuration. It would be nice to be able to build simple XML documents for email, office communication, and the Web that don’t have a DTD and to do it with an interface that mimicked expected word processing behavior.