TEI address element is wrongly converted to HTML address element

[martindholmes]

The purpose of the HTML <address> element is very precise and constrained:

The address element provides contact information for a document or part of a document. 
Information provided by address may include the names of the document’s maintainers, 
links to the maintainers’ Web pages, e-mail addresses for feedback, postal addresses, 
phone numbers, and so on. The address element is not appropriate for all postal and 
e-mail addresses; it should be reserved for providing such information about the contact 
people for the document.

The TEI <address> element is much more generic, and appears all over the place, containing the addresses of people in personographies, places in gazetteers, and so on. However, the Stylesheets (in particular html/html_core.xsl, line 95) simply converts the TEI <address> to the HTML <address>, which is both semantically wrong and often invalid (because the range of contexts allowing the HTML address is quite constrained).

We need to decide on another default rendering approach for generic addresses (<span class="address">?) for most contexts, and also we need to see if we can figure out when an address would be a good candidate for conversion to the HTML <address> element (for instance when it's associated with a document author, a publisher, an editor, or something like that).

[lb42]

Ill bet there are plenty of places where html:address is used for its formattibg effect only. And where it really corresponds to something like tei:trailer

[jamescummings]

Where can the html:address appear or not? What are the constraints on it? I would agree that the fall back for tei:address is html:span with a class='address' makes sense since most of the places tei:address is allowed seem to be phrase-level. When you are rendering metadata (e.g. when ancestor::publicationStmt) then perhaps it makes sense to be doing something different with it, but surely that depends on whatever it is we're doing with publicationStmt in our teitohtml at the moment?

[martindholmes]

html:address can appear anywhere flow content is allowed (I know that's vague, but I can't find anything more specific). Here's a useful page on it:

https://www.w3.org/wiki/HTML/Elements/address

I think the semantic issues are more significant than the validation issues, although I first came across this through a validation problem in one of our test outputs.

[martindholmes]

The Stylesheets co-op group looked at this and decided that in html_core.xsl, the output from <address> should be <span class="address">, and that css/tei.css should have the following rule added to it:

span.address{
  display: block;
  font-style: italic;
}

That should mean that the output appearance should not change from what it now is, but the invalidity issues should go away. Two other issues came up, which will get separate tickets. One issue that remains for this ticket: are there contexts (such as <publicationStmt>, should that be processed into HTML) in which an HTML <address> element is appropriate as output from the TEI element?

[martindholmes]

See Martina's pull request here:

#265

Looks good to go -- as soon as anyone has the time to handle the expected-results fallout, do pull it!

[ebeshero]

I'm reviewing XSLT Stylesheets tickets, and saw we have at least a partial fix here thanks to @martinascholger . Where are we with this?

[ebeshero]

We may be finished with this: Tasks to double-check:

• Write up a test file with <address> and check how it works.
• check pull request

[ebeshero]

F2F Group: Pull request looks good. Questions: Should all TEI address elements be processed in this way? Check how we want <address> to be handled when we have ancestor::publicationStmt.

[ebeshero]

F2F: publicationStmt use of <address>does not neatly correlate to the HTML address. We think this fix is sufficient for all cases and can close the ticket.