- From: Norman Walsh <ndw@nwalsh.com>
- Date: Tue, 04 Nov 2008 07:47:53 -0500
- To: public-xml-processing-model-comments@w3.org
- Message-ID: <m2wsfjr72e.fsf@nwalsh.com>
Toman_Vojtech@emc.com writes: >> I think these should probably be dynamic errors and I actually think >> 43 subsumes 45. > > I agree that the codes could be merged together, but why do you want > them to be dynamic errors? Because in the general case, option values can be constructed at runtime. If we don't make 43/45 dynamic, then we'll need dynamic errors *as well* for the cases where they can't be checked statically. e.g.: <p:add-attribute attribute-value="5"> <p:with-option name="attribute-name" select="QName('http://example.com/',/*/@name)"/> </p:add-attribute> > (Just noticed that for p:add-attribute, the spec just says: "The value > of the attribute-name option must be a QName. The corresponding expanded > name is used to construct the added attribute." So, if the value is not > a valid QName, the spec does not say what error should be raised.) I think the idea was to state some of the really common errors (like invalid option names/types) in one place, rather than having to repeat them over and over again. Be seeing you, norm -- Norman Walsh <ndw@nwalsh.com> | Before doing someone a favour, make http://nwalsh.com/ | sure that he isn't a madman.--Eugéne | Labiche
Received on Tuesday, 4 November 2008 12:50:17 UTC