Re: [xsl] Re: The Perils of Sudden Type-Safety in XPath 2.0

Subject: Re: [xsl] Re: The Perils of Sudden Type-Safety in XPath 2.0
From: Gunther Schadow <gunther@xxxxxxxxxxxxxxxxxxxxxx>
Date: Tue, 25 Feb 2003 15:24:47 -0500
Dimitre, David, Kurt,

Dimitre Novatchev wrote:
> "Gunther Schadow" <gunther@xxxxxxxxxxxxxxxxxxxxxx> wrote in message

>>Michael suggested I comment on the official w3c comment list. But
>>it could be a more powerful comment if it came from more than one
>>of us. We could draft a comment and then mass-mail it in from
>>differnent members of this list.

> Not too long ago David Carlisle expressed strong concerns on
> similar issues.
>
> And please add to the list of issues the "feature" of different
> variables having the same name in the same lexical scope...
>
> I have the feeling that a group of people has the goal of making
> XSLT a ridiculous language.


as far as this and other complaints goes that we could submit as official W3C comments, I would almost prefer if you, Dimitre, or David would write up such a comment. Dimitre has in the past exposed some weaknesses in my argumentation :-) and so, my words may not be as effective as yours. I am ready to help and sign a petition and I have contacts to a voting member of the working group.

I am also copying Michael for the fairness. Michael has spoken in
favor of some type safety and I trust him a lot, certainly he is
not someone who could be accused of trying to throw sand into the
gears of XSLT. I could see how parameter polymorphism would work
nicely even with automatic type casting. First one would attempt
at matching an function's implementation against the actual argument
types and only if no complete match exists, conversion would
be done on those actual arguments that do not macth the type of
the corresponding formal argument. The whole thing could be
disambiguated by a match score that would be

        n
score:=SUM(likeness(type(formalArg[i]), type(actualArg[i])) * (n - i))
       i=1

with n being the number of arguments, and the "likeness" function
between types being something that makes double and int more
related than double and boolean. Also I have factored in the
position of the argument just for the fun of it. This may be a
totally off base solution to the polymorphism issue. If so, forget
it and suggest a better one.


I am very anxious of having the practical experience and understanding of this group here influence the W3C working group. There's got to be a way for ernest and sophisticated users to have some influence even if against the Microsofts of this world.

regards,
-Gunther


-- Gunther Schadow, M.D., Ph.D. gschadow@xxxxxxxxxxxxxxx Medical Information Scientist Regenstrief Institute for Health Care Adjunct Assistant Professor Indiana University School of Medicine tel:1(317)630-7960 http://aurora.regenstrief.org



XSL-List info and archive: http://www.mulberrytech.com/xsl/xsl-list


Current Thread