Re: xml:lang [was Re: Outstanding Issues ]

At 20:27 11/02/2002 +0200, Patrick Stickler wrote:
[...]
> > My question was: does anyone have a compelling reason to change this.  Do
> > you have one Patrick?

[...]

>A literal is not a pair ("string", "lang"). The M&S is wrong.

I was hoping for something a little more compelling than a bald assertion.

[...]


>So now,
>
>    <key xml:lang="en">pan</key>
>    <key xml:lang="sp">pan</key>
>
>do we get
>
>    xxx key "pan" .  ("en")
>    xxx key "pan" .  ("sp")
>
>or
>
>    xxx key ("pan","en") .
>    xxx key ("pan","sp") .


I do not understand the semantics of the difference between these two 
representations, so I can't answer that question.

>Now which represents tidy literals?

As above, I don't know.

>  And does
>that mean that for *every* query that compares
>literals one must specify language?

That seems to be an issue of query language design and out of scope of this 
discussion.


>And what about comparison of literals where one
>is specified for language and the other is not,
>do they match? No?

I would expect that we would define things such that they don't match.

>Why?

Because the language is part of the literal, and the languages don't match.


>Nope. I don't think that any of our discussions
>over the past few months have considered literals
>to be anything but the string.

Not since you joined, perhaps, but it has been an open issue all that time.

Brian