Re: [xslt] Identical separators in format-number() and decimal-format

[Daniel Veillard <veillard redhat com>]

On Wed, Jul 13, 2005 at 08:56:41PM +0200, Bjorn Reese wrote:
> Tony Marston wrote:
> 
> >Which interpretation is correct? Which interpretation is the most 
> >logical? Which interpretation is easiest to implement?

  For the record Bjorn is the initial author of the number and format support
in libxslt :-)

> There are a number of ways in which we can try to address this matter:
> 
>  1. We can seek clarification in the JDK.
> 
>     The JDK 1.1 specification of DecimalFormat is rather vague, so we
>     have to resort to a later version of JDK. However, the XSLT spec
>     deliberately seems to discourage this (section 12.3)
> 
>       "The format pattern must not contain the currency sign (#x00A4);
>        support for this feature was added after the initial release of
>        JDK 1.1"

  Refering to the JDK was a serious mistake, they recognize that now :-)

>  2. We can seek clarification from an authorative source (the editors of
>     the XSLT standard).
> 
>     I have previously attempted to seek clarify from the XSLT editors
>     on other borderline cases, but without success, so I am not too
>     optimistic about this option. See:
> 
>     http://lists.w3.org/Archives/Public/xsl-editors/2001JanMar/0075.html

  The Working Group moved to 2.0, a lot of the original 1.0 contributors
are not there anymore, not sure tehy are interested.

>  3. We can adopt the behavior of (the majority of) other XSLT processors
>     out there.
> 
>     This is how we have resolved borderline cases concerning infinity
>     and not-a-number. This, however, requires that somebody with access
>     to other XSLT processors examine and report their behavior.

  I think William looked at Saxon's behaviour before making an answer to
Tony in bugzilla. We should also check the 2 XSLT processors from IBM/Apache
the C and the Java ones.

>  4. We can reason about what kind of behavior we prefer.
> 
>     Although it is always good to apply a dose of reasoning to our
>     decisions, we should be careful not to invent our own behavior
>     when it comes to standards. For example, hardcoding the values of
>     decimal-point and thousand-separator can be confusing for users
>     who are not familiar with the values that we select.

  Agreed

> I am unsure how we best approach this (although our approach ought to
> include the spanking of the XSLT editors :)

  Ahum, to me 3. is the most pragmatic approach. 2. would be the correct
one, it doesn't cost much except time.
   Asking the xsl-list and checking Michael Kay's book on the topic are
also reasonable things to do.

Daniel

-- 
Daniel Veillard      | Red Hat Desktop team http://redhat.com/
veillard redhat com  | libxml GNOME XML XSLT toolkit  http://xmlsoft.org/
http://veillard.com/ | Rpmfind RPM search engine http://rpmfind.net/