Sounds like MarkLogic is not conforming to the spec then, as it clearly applies language to attributes as well.
But who am I to complain, maybe it is an omission in the spec.. :) Cheers, Geert On 6/25/15, 1:26 PM, "David Lee" <david....@marklogic.com> wrote: >>> I know the order of attributes has no semantic difference. I was >>>merely suggesting it as an attempt to bypass the bug. > >I understand, and it definitely points to something being wrong or >atleast suspicious. I am suggesting that the test case itself is >potentially in the 'not defined' camp - so the results being dependent on >the attribute order may also be 'not defined' (although it helps locate >where in the code this is affected). > >The docs are pretty clear that the xml:lang affects the language of the >*child text* of elements, >All the samples give have no child text - so this is not (documented as) >in-scope with the xml:lang behavior. This may be irrelevant - but it >would be useful to see if the same behavior is reproducible using defined >behavior - i.e. using child text and testing for the language to the text >node as opposed to empty elements. _______________________________________________ General mailing list General@developer.marklogic.com Manage your subscription at: http://developer.marklogic.com/mailman/listinfo/general