DataAndLogic

For typesetting, it is generally recognized as a good practice to separate style from content. Some programs make the same kind of distinction to separate the data from the logic that operates on the data.

In the past, I’ve though of this as “internally” and “externally” defined logic.

Examples that separate data from logic:

• The UNIX philosophy
• XML/XSLT

Examples that do not separate data from logic:

• $\LaTeX$
• Spreadsheets
• TiddlyWiki, and tons of other notetaking software.

Lisps seem to be their own things… (Often claims “code is data.” Aren’t S-expressions similar to XML?)

For a long time, I thought that that separation was wise.

• Data is often useful for longer than a host application. Giving data an application-independent existence gives it more permanence.
• Independent data can be used by multiple applications.

Some exceptions come to mind:

• It is difficult to define application-independent data. Think of how long it takes to codify a standard.
• Coupling data to application gives you the opportunity for ad hoc logic.

What about when data is primary and logic is secondary?

To do:

Dig into this discussion between Alan Kay and Rich Hadley.