- From: Norman Walsh <ndw@nwalsh.com>
- Date: Wed, 27 Aug 2008 08:49:56 -0400
- To: public-xml-processing-model-comments@w3.org
- Message-ID: <m23akqd3m3.fsf@nwalsh.com>
"Florent Georges" <fgeorges@fgeorges.org> writes:
>> The only place they're allowed is in a subpipeline where they
>> must have a corresponding declaration (as they must be atomic
>> steps).
>
> So you've chosen to not reuse the concepts of "user-defined
> data elements" [1] and "extension instructions" [2,3] from XSLT
> (both 1.0 & 2.0)? :
It's trickier in XProc because sequence matters more so than in XSLT.
> The former has proved very useful for little "extensions" to
> the language, for example regarding documentation, or meta
> information (for instance mapping from the business rules
> documents).
You can put anything you want in p:pipeinfo and you can put that
anywhere you want.
> About the later, I don't understand comprehensively and
> exactly the extension mechanism of XProc, but I can't think you
> didn't allow the ability of declaring extension steps defined
> in an implementation-defined way (read: in Java or whatever.)
Sure, you can do that.
<p:declare-step type="px:my-extension"
px:class="org.xproc.extensions.MyExtension">
<p:input port="source"/>
<p:output port="result"/>
</p:declare-step>
Be seeing you,
norm
--
Norman Walsh <ndw@nwalsh.com> | Birds are taken with pipes that imitate
http://nwalsh.com/ | their own voices, and men with those
| sayings that are most agreeable to
| their own opinions.--Samuel Butler
Received on Wednesday, 27 August 2008 12:50:42 UTC