Infrastructures for Information/Grif

July 5, 1998

Liora Alschuler

The Seybold Report on Internet Publishing
Special for XML.com

A view from SGML/XML Europe, '98

I4I was formed in 1993 as an SGML and database integrator by Michel Vulpe, an early SGML DTD designer who worked on the massive inventory project of the Smithsonian collection in the 1970s.

The core technology is not word processing, but what they call "S4." The quick fix on S4, according to I4I, is that it does for SG/XML encoded information what ODBC drivers do for tabular databases. Here's what we said about this engine when it debuted in 1996.

What they have produced that merits coverage in an XML editors roundup is an add-on, not a plug-in, for Microsoft Word. Unlike the four Word plug-ins that failed to make an impact on the structured editing market, the I4I technology provides real-time validation of content as it is created.

If writing with a WYSIWYG, Word-like interface while producing valid SG/XML is the Holy Grail of structured editors, I4I has identified the sword and the stone, has a method, and is just waiting for the right King Arthur to come along and execute the plan. S4 DT is a library or toolkit for developers that can customize it for a single implementation. Better yet, according to I4I, developers can buy the technology and make a product out of it. Discussions are under way with some strategic partners, but according to Kevin Burke, president of I4I, if no developer creates an end-user, off-the-shelf product, I4I will do so. The approach to editing seems to do some things right, from a writer's perspective, so we feel it is worth inclusion here, despite the half-cooked product strategy.

The S4-based implementation shown in Paris accepts any DTD, no pre-compilation required. It uses the OASIS Catalog to associate DTDs and documents. The familiar Word interface sports added icons and windows for structure view and attribute editing. The structure view and the text document are linked, so moving the cursor in one window is tracked in the other window. Unlike FrameMaker+SGML and some of the other editors, however, a user cannot edit a document directly in the structure view. Tags can be shown or hidden. Element insertion is context sensitive, according to the DTD. And, at the same time, the text can be formatted using any of the Word formatting capabilities.




So, how do they produce this WYSIWYG cake and structure, too? The I4I technology controls format in a template that contains Word styles and macros. Currently, it is an integrator-level task to create the Word macros that relate style information to structured text. These macros are DTD specific, so, while the add-on can load and parse an arbitrary DTD, the point-and-click styling must be associated through a separate template for each DTD. Of course, I4I is saying the right things about XSL: When they build it, we'll support it.

The template not only gives the user point-and-click, immediate styling gratification. It enables some of the capabilities of a full-fledged publishing tool, which Word has never claimed to be. For example, templates and macros can specify prefix text and can alter formatting depending on the document context. The styles themselves, that is, whether one uses Arial or Futura, can be edited in the Word interface.

Save options are as Word, as SGML, or as a hybrid that includes both the structure and the format information. Docs saved as Word can then be twiddled for print output. The intent of the hybrid is to give readers and reviewers who do not have the add-on the ability to view the explicit structure in a formatted version of the document.

Entities can be declared in the DTD or the declaration subset and inserted via a dialog box, but cannot be edited directly in the document. The S4 desktop uses ActiveX controls to communicate with a database.

Current implementations include a project with Toronto's Financial Post (I4I is not only a good-old SGML company, it is a good old Canadian SGML company, and one that has inherited staff from the best of the other companies.) The Post has an SQL database and they have modeled the SQL queries in SGML. The retrieved content is inserted into a document instance. The document then goes to FrameMaker+SGML and gets published as a CD-ROM, on the Web as HTML, and in print.

Will this ever become more than a development environemnt? Certainly, we like to see an editor with a Word interface and real XML out the back end. Whoever King Arthur or Prince Charming turns out to be, he will have to build a GUI for the I4I templates as well and face down who knows what other fire-breathing bugs. Switching metaphors, we view each of these attempts to marry writer-ease-of-use with structure-and-validation as important experiments, even if this one, too, fails to last.

Grif's SGML editor