Re: p:for-each

2008/12/23 Norman Walsh <>:
> "Dave Pawson" <> writes:
>>> A sequence of documents is...a sequence of zero or more documents.
>> For which the spec has no definition of 'document'. My usage, of
>> a file within a directory, is a reasonable interpretation lacking
>> existing clarity within the spec as I see it.
> Really? You think the notion of XML document is insufficiently
> rigerous in the spec? Given what we say in paragraph 3 of 2.2 Inputs
> and Outputs, would we really improve things if we added something
> along these lines:

The phrase ' sequence of documents', along with sequence, is
not defined. On that basis I'm free to take any interpretation
that seems reasonable to me?
  Being a dochead, I see a document as a file on disk.
My point being that there is nothing in the spec to
disillusion me?

2.2 doesn't provide any definition Norm. It reduces the space
from any document to XML document?

>  XProc processes Infoset Document Information Items as XML documents
> The spec makes it perfectly clear, I think, that how you get from the
> outside world (files in a directory) into the initial pipeline step is
> implementation defined.

I may be alone in misinterpreting things. If so, great.

> In the case of Calabash, you can use -isource multiple times to make a
> sequence. Perhaps I should also allow -isource=*.xml, I think that
> makes sense.

I went looking for 'collection' as per XSLT 2.0, but failed to find it.

IMHO working with a collection of XML files on disk is sufficiently
common to warrant being addressed as a source of XML.


Dave Pawson
Docbook FAQ.

Received on Tuesday, 23 December 2008 10:30:25 UTC