[tei-council] Oddity with Roma

[Martin Holmes]

On 12-11-15 05:43 AM, Lou Burnard wrote:
> On 15/11/12 01:01, Martin Holmes wrote:
>> I have an ODD file here:
>>
>> <https://hcmc.uvic.ca/people/martin/test/london_all2.odd>
>>
>> When I generate an RNG schema from it, the schema seems to disallow some
>> attributes which I haven't explicitly removed. Specifically, it doesn't
>> allow:
>>
>> @rend on <hi>
>> @xml:lang on <foreign>
>
> It seems suspiciously coincidental that these two attributes used to be
> locally defined on their specific elements, rather than being inherited
> from the att.global class. I wonder if you're picking up an old version
> of something somewhere along the way?

Judging by the svn log, this must have happened quite a while ago. The 
first time I worked on this schema would have been the summer of 2011, 
and I've tweaked the ODD file repeatedly since then, but I've never used 
Roma to do it. Yesterday, at a TEI workshop we were giving, one of our 
RAs took my ODD file and uploaded it into Roma to add some new elements 
and attributes he needed. The result was the broken schema, although as 
far as I can see there's nothing wrong with the ODD file itself.

Other people at the same workshop managed to generate broken schemas 
too; a problem I saw twice was duplicate definitions of att.ascribed. We 
had no trouble with schemas where people just added modules, but once 
they started removing elements and attributes from their modules, they 
had some problems. I hadn't really thought before how easy it is to 
generate something irrational if you don't know exactly what you're 
doing -- you can for instance remove <biblStruct> and leave <analytic>, 
and then the sanity checker will rightly point out that you can't get to 
<analytic> from anywhere; that sort of thing isn't Roma's fault, and I 
don't think it causes a broken schema in itself, but it all made me 
realize that Roma isn't really a tool for novices, once you go beyond 
selecting your modules.

When I get a chance to check out the latest Byzantium I'll try running 
it through there.

Cheers,
Martin

-- 
Martin Holmes
University of Victoria Humanities Computing and Media Centre
(mholmes at uvic.ca)