[Om] Hexadecimal integers?

[Lars Hellström]

I first posted the following a week ago, before subscribing to the 
list, but since it hasn't shown up yet I suppose I'd better repost it...

While I'm not entirely sure, I think the OpenMath2 specification has a 
bug in that the base-16 form of integers is missing from the Relax NG 
schema. The same issue also seems to exist in OpenMath3 (as found in 
the SVN repository).

Concretely, Section 3.1.2 (Informal description of the XML Encoding) of
the OpenMath 2.0 specification speaks of base 16 integers matching the
regular expression -?x[0-9A-F]+ and gives <OMI> xA </OMI> as an
example, but the schema in Section 3.1.1 (A Schema for the XML
Encoding) only gives the regexp

   \s*(-\s?)?[0-9]+(\s[0-9]+)*\s*

for the string in an OMI element, and unless I'm mistaken that only
covers base 10 integers. (It also seems to disallow multiple whitespace 
characters in sequence, which strikes me as unnecessarily restrictive; 
consider indentation in an integer wrapped over several lines.)

The inconsistency between these two sections appears to have come about
between versions 1.1 and 2.0, since a regexp allowing base-16 integers
can be found in the former. The informal description is the same in both:
http://www.openmath.org/standard/om20-2004-06-30/omstd20html-diff.xml

Since this change is not noted in the list of changes, I presume it is
the schema that has an error.

As the 2.0 standard is over four years old, I found it a bit unlikely
that such an error would have gone unnoticed for this long, but have
been unable to find any other mention of it. This did however uncover
some other issues, perhaps primarily regarding the www.openmath.org
website:

* There is no errata document for the 2.0 standard linked to from
   http://www.openmath.org/standard/om20-2004-06-30/
   although at least one error has previously been reported on
   this mailing lies (something concerning # in href in some example,
   if I recall my skimming of the archives last night correctly).

* There appears to be no information on how to report errors.
   (Unlike the case with the MathML standards, which carry a notice
   asking that errors are reported to www-math at m3.org.)

   From googling I've discovered that there is an OpenMath 3
   trac server at https://trac.mathweb.org/OM3 which could be
   appropriate, but I have no way of knowing how authorative that
   might be.

* There is an om3 at openmath.org mailing list, but this is not listed
   on http://www.openmath.org/lists/index.html.

Lars Hellström