RE: Principle of Least Power

Umm.. that's not exactly my concern, Harry.  Sorry if it 
seemed that way from my posts.  It is more 
along the lines of how to apply the principle given 
some particular system decision.   This sort of thing 
becomes a problem for someone procuring a technology and 
the vendor that sells it to them.  See Sarbanes-Oxley. 
We have contracts wonks here and in the Beltway and 
its analog around the world who take the W3C very 
seriously and cite this stuff without necessarily 
doing due diligence technically.  Then it becomes my problem.

Power to the user isn't my problem.  Selling water by the river is.

Dan and Henry are doing a good job with this by exploring 
the metrics.  At some point, it needs to translate to 
something a proposal writer or consultant will understand, 
but first, the technogeekery pass is needed.  (Maybe we 
think programmers select technologies; that 
is not nearly as common as one might want to believe 
particularly as the system buys get bigger.)

BTW:  the comparisons of XSD, RELAX, Schematron are alive 
on the XML-Dev list simultaneously.  It is stated that RELAX 
is a superset of XSD.  XSD is a superset of DTD. If I applied 
the principle literally, I'd be writing DTDs as I did in the 
SGML days.  So in some way, the finding must express 'adequate 
to the job' and 'here's how you know if your language is 
adequate but no more than that'.

Wow.  This becomes the Peter Principle of Software.

len

-----Original Message-----
From: www-tag-request@w3.org [mailto:www-tag-request@w3.org]

And then to address Len's concerns, one could state something along the
lines of "By power we do not mean the subjective feeling of power
particular programming languages or data formats give their users, but
the amount of information a particular language tells us about the
programs or data expressed in that language. By setting constraints as
to what can be expressed and making themselves capable of being
analyzed, certain languages while being less powerful in terms of
Turing-completeness or other forms of power, actually give the Web more
information about themselves when they can guarantee certain behavior. "

And everybody wants more information about everybody else :)

Received on Wednesday, 8 February 2006 20:10:27 UTC