Re: Multiple symbol spaces considered valuable

On Tue, 2005-12-13 at 20:20 +0000, Henry S. Thompson wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
> 
> Following up on the interesting analogy you drew during today's call
> between Python and XML with respect to the value of a single symbol
> space within a namespace, I offer the following analogy for
> consideration:
> 
>  XML Namespace  :  Element : Attribute
>  Python package :  Method  : Local variable

You're making a case for scopes, not for multiple top-level symbol
spaces.

It's straightforward to say that ...html#p refers to the
p element type in HTML, and ... ...os#curdir refers to
the curdir variable in the os module in python.

The tricky bit is what to do about html#title . To say
that attributes are scoped to elements is straightforward;
it makes them invisible at the outermost scope; that's a
little bit of a pain, but not too much. It just means
you can't refer to the title attribute using a URI.

The main benefit of the doc#term idiom is sharing across
formats:
 "Great multiplicative power of reuse derives from the facts that
 all languages use URIs as identifiers: This allows things written
 in one language to refer to things defined in another language."
 -- http://www.w3.org/DesignIssues/Architecture

So to say that html#title refers to the title element type
means you don't get a trivially easy way to refer to the title
attribute from, say, python. I can live with that.

[...]
> 
> It follows that there _is_ no QName-style name, that is, one made of
> of Namespace/Package name + local name, which distinguishes between
> the two Python variables, just as there is no QName-style name which
> distinguishes between the two XHTML attributes.

Well, sure there is: it's possible, though perhaps awkward, to mint
a URI for the space of attributes of the P element, or for the
space of local variables in the swap.formula.compareTerm method.

The doc#term idiom doesn't really apply in that case, though,
since the compareTerm method isn't published in the web as its
own information resource.

>   If you require a
> global unambiguous name for one of the variables, or one of the
> attributes, you require a more complex form of name (or URI), which
> encodes sufficient information about the context (method or element,
> in these cases), to disambiguate.
> 
> I don't think this is a bug, in Python or in XML -- I think it's a
> natural pattern of naming.
> 
> ht
> 
> [1] http://dev.w3.org/cvsweb/2000/10/swap/formula.py?rev=1.30&content-type=text/x-cvsweb-markup

-- 
Dan Connolly, W3C http://www.w3.org/People/Connolly/
D3C2 887B 0F92 6005 C541  0875 0F91 96DE 6E52 C29E

Received on Wednesday, 14 December 2005 17:27:27 UTC