This is an archived snapshot of W3C's public bugzilla bug tracker, decommissioned in April 2019. Please see the home page for more details.
QT approved comment In 3.3.4.1, the definition of "constant" needs tidying up. The link is to a definition of "enumerated constant", which doesn't make it clear whether there are any non-enumerated constants; moreover, the definition says that constants are undefined, which isn't helpful. Related to this, In 3.3.4.1, is the distinction between INF and positiveInfinity intended (similarly NaN and notANumber, etc)?
Also on this topic: In 3.3.5.1, I thought "special values" were now called "constants". (I think I prefer "special values").
(In reply to comment #1) > In 3.3.5.1, I thought "special values" were now called "constants". (I think I > prefer "special values"). Generally, the values in the value space of float and double are ordinary numbers. There are five values that are, to most people's understanding, not ordinary numbers. (They might or might not consider the two zeros to be numbers, but probably not "ordinary".) Things in a value space are called "values"; these are values that are not "ordinary numbers". Yes, they are constants (as are a number of things that aren't members of value spaces).
(In reply to comment #0) > In 3.3.4.1, the definition of "constant" needs tidying up. The link is to a > definition of "enumerated constant", which doesn't make it clear whether there > are any non-enumerated constants; moreover, the definition says that constants > are undefined, which isn't helpful. There are no other constants as we use the word. (We don't use the word 'constant' linguistically to describe nouns such as '1' or 'one' (as opposed to variables, whose value may be reassigned without changing the mathematical structure under discussion). Perhaps having the definition read "(enumerated) constant". We call them enumerated because the only ones we have are those that get enumerated somewhere in the spec (as opposed to somehow being described without being explicitly mentioned). We call them undefined because they carry no preordained meaning other than the use to which we put them in the spec. Shall we explain these terms? > Related to this, In 3.3.4.1, is the distinction between INF and > positiveInfinity intended (similarly NaN and notANumber, etc)? Yes. positiveInfinity (the constant) is the value; 'INF' is its lexical representation. Similarly, notANumber is the value; 'NaN' is its lexical representation.
Thanks for the clarifications, but the real question is, how do you propose to change the spec so that people can understand it without looking at your explanations in bugzilla?
A wording proposal which would resolve this issue (by changing most occurrances of 'constant' to 'special value' and providing a definition of 'special value' modeled on the current definition of 'constant', with revisions) has been prepared; it's at http://www.w3.org/XML/Group/2004/06/xmlschema-2/datatypes.b3249.html (member-only link).
At its call today, the XML Schema WG declined to adopt the proposal mentioned in comment 5, and requested that the editors try again. Accordingly, I'm changing the status keyword to needsDrafting, and raising the severity from minor (easy) to normal (work).
At its call of today, the XML Schema WG accepted a revised wording proposal intended to resolve this issue. http://www.w3.org/XML/Group/2004/06/xmlschema-2/datatypes.b3249.html (member-only link). Michael, since you were there and voted to approve the proposal, I assume you will be willing to take the resolution back to QT or to close the issue on their behalf, or to reopen it if there is pushback. Thanks.