- From: Innovimax SARL <innovimax@gmail.com>
- Date: Fri, 22 Dec 2006 23:00:44 +0100
- To: "XProc WG" <public-xml-processing-model-wg@w3.org>
== Load == I think the story of the load component, which we all agree is the component behind the <p:document href="..."/>, should be clearer about its validating status As the document is at least well formed, what is the application supposed to do * when there is a <!DOCTYPE ...> * when there is <root xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:noNamespaceSchemaLocation="mySchema.xsd"> We should take care that DTD and XML Schema for PSVI == Validate == The p:validate is referenced in the spec for the moment in example 2 but is not in standard component library It seems interesting to specify all the way to validate with current heavily used grammar so as to be interoperable, without mandating to support them all The fallback mechanism is highly interesting Because there is at least 5 kinds of fallback : a) The processor doesn't not support this kind of language b) The processor support but file is not accessible c) The processor support but the model is corrupted d) The processor support, the model is correct, but the file is not accessible e) The processor support, the model is correct, but the file doesn't validate We need to be able to distinguish amongst them b) and d) need to surround each p:input with try/catch c) could may be caught by validating the schema itself before (so p:validate should be able to do that) a) we need to have a construct for optional construct to detect them with a simple test. For this, I propose a p:pipeline source and configuration port stream which could be used like <p:parameter name="is-supporting-relax-ng" filter="/component/p:validate/supported-namespaces/ns = 'http://relaxng.org/ns/structure/1.0'"> <p:pipe step="p:pipeline" port="configuration"/> </p:parameter> [Note that it could be interesting to reserve step that are in the xproc namespace] By successive fallback e) should be the fallback of p:validate == NVDL == It is interesting at multiple level to look carefully at NVDL First, it is in the requirements Then, with a single document and an NVDL script we could end up with an undefined number of output from the NVDL perspective (or may be a fixed number (depending only on the script) of output sequence) Mohamed -- Innovimax SARL Consulting, Training & XML Development 9, impasse des Orteaux 75020 Paris Tel : +33 8 72 475787 Fax : +33 1 4356 1746 http://www.innovimax.fr RCS Paris 488.018.631 SARL au capital de 10.000 €
Received on Friday, 22 December 2006 22:01:02 UTC