- From: Larry Masinter <masinter@adobe.com>
- Date: Tue, 2 Oct 2012 17:05:10 -0700
- To: "Michael[tm] Smith" <mike@w3.org>, "Martin J. Dürst" <duerst@it.aoyama.ac.jp>
- CC: Robin Berjon <robin@w3.org>, Noah Mendelsohn <nrm@arcanedomain.com>, W3C TAG <www-tag@w3.org>
I wanted to add that the "precision and error handling" in HTML5 is *not* mainly about parsing -- the parsing algorithm is great to debate about, but it's kind of a distraction. Yes, there's a parsing framework there, but much (most) of HTML5 is really a set of APIs, and when documenting an API and the operational behavior of the virtual environment it operates on *also* has a wide framework of precision and error handling. It's the precision and error handling, the attempt to gain specificity of operational behavior by giving pseudo-code algorithms, that is questionable in technical specifications. Java, JVM, Smalltalk and the Smalltalk virtual machine, and so forth, IEEE floating point by reference implementation... these are all precedents, but I can't think of any system as complex as the open web platform which has a "precise" specification written in such a way. Larry
Received on Wednesday, 3 October 2012 00:05:38 UTC