No description of how numeric property values are obtained.

[w3c]

No description of how numeric property values are obtained.

[chaals]

I am inclined to suggest that processors may convert numerical values to a matching datatype, without obliging them to do so...

[gkellogg]

So, the text in the JSON serialization section is unnecessary, then.

[danbri]

Haven't thought about this much, but would declaring "meter" as allowing xsd:integer or xsd:double work?

[danbri]

SGTM.

[danbri]

p.s. can you talk me through how we handle surface details like "." vs "," e.g.a value of "1,00" vs "1.00"?

[gkellogg]

I believe HTML 5 addresses valid number syntax: https://www.w3.org/TR/html52/infrastructure.html#numbers, and how they're interpreted.

[chaals]

Since we define the content as strings, we don't have to do any handling. Processors can use whatever information they like to decidce that something is a number - e.g. a page in spanish that has <meter itemprop="rating" value="3,5"><img src="estr.gif"><img src="estr.gif"><img src="estr.gif"><img src="mestr.gif"><img src="estrv.gif"></meter> provides a value according to the spec of "3,5" as a string. But it is possible for a processor to decide that this is actually the number three-and-a-half. Although it is not required - and incidentally is not valid HTML.

[danbri]

incidentally is not valid HTML.

... I had missed that, and am now reassured.

[chaals]

We explain in the introduction that values are strings or URLs, and again in the algorithm for determining values. I think we can close this without a change to the text...

[chaals]

@gkellogg yep. I'll clean it up.

[chaals]

I trimmed some of the language about conversion in #76.

I will add some clarification in various places - while conversions may interpret content to add datatypes, microdata itself treats even dates taken from datetime as a string...

[danbri]

Closing as agreed/fixed above.