On 15-05-19 11:29 AM, James Cummings wrote:
I was just bitten by the side effect of moving @rend to att.global.rendition. In taking an old and perfectly fine TEI ODD, and generating a fresh schema from it, the RNG was invalid because of a duplicate definition of @rend. Changing it so that I modified @rend on att.global.rendition rather than att.global solved this. Took me a good 10 minutes staring at it to realise what was wrong. doh.
I'm just mentioning it to put it in the back of your brain in case anyone asks in the future.
I've been bitten by it too, and I was the once who made the change. :-) I think any significant change to class structures like this will generate similar problems. I have tried to think of a way around this, but the only approach I can think of involves specially-coded traps in the schema creation process, and that's a bit ungainly. We could add Schematron rules to apply to the ODD, but for that to work, you would need to be validating your ODD against a fresh TEI schema at some point. Would that have worked for you in this case? Cheers, Martin