Re: [closed] Re: p:validate-with-schematron report defect

"Florent Georges" <fgeorges@fgeorges.org> writes:

>> The WG agreed to suggest that the output SHOULD be SVRL.
>
>   That's better than nothing.  On the other side, that means that one
> can not rely on the step's output format, thus she couldn't use the
> report output to transform it in HTML for instance, nor to report any
> meaningful info to the user in any way.

I understand your concern, but the WG is trying to be pragmatic.
Realistically, implementors are going to want to incorporate
off-the-shelf implementations of as many steps as they possibly can.
I'm using Saxon, for example, for the XSLT and Query steps, Kohsuke's
MSV for RELAX NG validation, etc.

I haven't figured out how I'm going to support Schematron, but
whatever package I pick to do it, is going to produce some sort of
output. I don't want to be prevented from using what might be my only
practical option just because it produces output that isn't in SVRL.

>   Maybe we could have an optional option to set the expected report
> format (html, svrl, or an implementation-defined value) ?

Again, unless it's clear that off-the-shelf toolkits offer this
capability out-of-the-box, I don't think the WG will be persuaded to
go there.

                                        Be seeing you,
                                          norm

-- 
Norman Walsh <ndw@nwalsh.com> | Between the ages of twenty and forty we
http://nwalsh.com/            | are engaged in the process of
                              | discovering who we are, which involves
                              | learning the difference between
                              | accidental limitations which it is our
                              | duty to outgrow and the necessary
                              | limitations of our nature beyond which
                              | we cannot trespass with impunity.--W.
                              | H. Auden

Received on Friday, 3 October 2008 13:35:59 UTC