Re: missing start elementt
David Rosenborg <[email protected]>
| Newsgroups | gmane.text.xml.relaxng.general |
|---|---|
| Message-ID | <[email protected]> |
Um, I think not. Simplification rule 4.18 clearly states "A grammar must
have a start child element."
This is after the previous simplification rules has been applied. More
specifically "4.7. include element"
is interesting in this situation.
start = element foo { grammar { include "grammar-with-start.rnc" }}
Provided that the included grammar does have a start pattern,
this is a perfectly valid schema and this shows that you cannot
require a start in the grammar before the simplification.
Cheers,
David
David Tolpin wrote:
>>This is a general problem with RELAX NG: most of the language
>>restrictions can't be captured by the RNG schema for RNG.
>>
>>
>
>This one easily can, by the way. RNG is context specific, and specifying
>that the 'grammar' element has different content models in different contexts
>is possible.
>
>On the other hand, if includeContent is different from grammarContent, then
>topLevelContent can be different too. It is easy and natural in BNF, RNG
>and plain text. I would be glad to see this part of the specification updated --
>it would make my life easier.
>
>David Tolpin
>_______________________________________________
>relaxng-user mailing list
>[email protected]
>http://relaxng.org/mailman/listinfo/relaxng-user
>
>