Questions
- If SGML and more specifically XML were attempting to formalize the transfer of structured objects between systems, why did they fail? What is the relationship to projects like ASN.1? Were a lack of great parsing and generation tools a factor? Was the actual syntax that XML chose too ill-designed? Was the flexibility of SGML too complicated?
Takeaways
- The entity definitions in SGML specify a schema for each entity and allow for omitting tags when the omission leaves things unambiguous. No wonder early HTML didn’t have closing P tags. In GML a closing P tag was forbidden!
SGML
Derives largely from an IBM project for building user interfaces. It’s in this “GML” (a backronym from the authors’ names’ initial letters) where we find the XML tag, which has remained until this time a mystery to me.
The tags and composition of a document follows GML’s design, but XML’s formalization of syntax (as suggested? by Wikipedia) was influenced by John McCarty’s work on abstract syntax.
GML and XMP
GML was mostly line-oriented but allowed a number of group blocks to contain other elements. “Tags” started with a : and an ending tag copied the name of the opening tag with an additional e prefix. Tags ended in a period, and content followed. Most tags were line-oriented or allowed continuing their content on successive lines, but did not allow nesting. Special text formatting tags can be embedded in specific other tags, like the p tag; these described limited ASCII and ANSI display formatting options, not document descriptions. The UL and OL lists appear as does the SL (simple list) with no bullets, but with list-like alignment.
:h1.A guide to HTML's origins
:p.Paragraphs cannot contain an end tag.
:xmp.
Example code pauses the normal parsing rules so that
other content won't mask as tags.
:exmp.
This solved a longstanding issue for me because I had no idea why XMP requires the closing of any open P elements, but I believe it’s because right in its origin it was characteristically different from the other tags.