Menu
▾
▴
#405 Wrong schema generated
closed-fixed
5(default)
2013-11-19
2012-11-06
No
In the new release of the TEI, there is, I guess, a mistake.
If one edits with ROMA the attributes of <title>, it can be seen that the attributes type and subtype are defined twice (see attached image).
This fact, I guess, produces the following error associated with ROMA:
If I create a NEW attribute of <title>, then the attribute SUBTYPE will disappear as possible attribute for <title> (!)
I guess it is due to the fact that the attribute subtype is defined twice for <title> AND one of them has no definition (see the attached image).
This seems to be a result of ongoing work to make it possible to apply local overrides to attributes provided through a class. <title> is a member of att.typed, but the @subtype attribute is then locally removed and @type is redefined.
This seems to have happened in August, in r10742:
"make <abbr> and <title> members of att.typed, then locally remove @subtype and redefine @type"
It looks as though Roma might not be supporting this.
I have put in a minimal emergency fix to avoid the attributes being show twice, but its not right yet, as @subtype still shows erroneously. Fixing this properly means redoing the Roma infrastructure, and thats too much at the moment.
I am not very happy about this effect ....
I am putting this ticket to "pending", as it depends on work on Roma,which has no resource at present
Changing to AMBER. We know it needs doing but are not currently developing ROMA.
At Oxford 2013-11 face-to-face MH to investigate if this has been solved by other changes.
I tested this by starting from a complete schema (tei_all), and adding an attribute called "testAttribute" to the
<title>element. I then confirmed that the @subtype attribute was still there in the Roma interface (it was). Then saved an ODD and generated a RelaxNG schema (XML) from Roma, and tested them. The @subtype attribute was available alongside the new @testAttribute.I conclude that if this ever was a bug, it is no longer a problem, so I'm going to close the ticket.