[log in to unmask]: wrote > You want your XML files spell-checked? You could > write an application which did a callout to a > spell-checker for each text node. > Sure, but this wasn't the question. [log in to unmask] wrote: > > PS: Instead of explaining NDR. - my English is too poor for this - I > would a= > > sk how I can say to a standard XML parser: "Check the nodes with > attribution= > > of OSE for well-formedness!" (Old Spelling English ;-) > > > This is an impossible request. Before the parser can determine that a > node exists, let alone what its attribute values might be, the document > must be entirely well formed. > Well-formedness is a precondition for a parser to operate on a document > in any way. > > Obviously I have to sharpen my terminology. You too? If 'well-formedness' in the given TEI context is to discuss only on the level of metalanguage, nodes are not to 'validate' (= parsed against the grammar defined by DTD or otherwise). But what means then the passage rightly picked up by Sylvain? [log in to unmask] wrote: > A strange statement in the Guidelines: > > By default, a <formula> is assumed to contain character data which > is not validated in any way: Underlining the strangeness of such remark I cite an example for a date structure element: _dateStruct value='26-10-1775'_ _day value='26'_26_/day_ _month value='10'_October_/month_ _year value='1775'_1775_/year_ _/dateStruct_ For me such examples are showing how TEI conformant markup allows semantic document processing without looking inside the nodes. Isn't it?