Re: Invalid intermediate results

You could also have variants of your schema that allow expected variations in the markup at these intermediate steps. For example, if you know cross references are broken until after a fix-up step, omit those from the variants of the schema(s) you use before that fixup step. This would allow you to catch other issues earlier in your pipeline if that would be helpful to your users. 

Regards,
David

> On May 14, 2024, at 5:31 AM, Erik Siegel <erik@xatapult.nl> wrote:
> 
> 
> Hi Denis,
>  
> Yes, that’s what I do too: ignore it. But as soon as you have valid XML, you can validate of course.
>  
> Erik SIegel
>  
> From: denis.maier@unibe.ch <denis.maier@unibe.ch>
> Sent: Tuesday, 14 May 2024 10:11
> To: xproc-dev@w3.org
> Subject: Invalid intermediate results
>  
> Hi everyone,
>  
> I’ve recently started learning xproc and as a learning task I’ve chosen to rewrite/stepify a monolithic xslt-stylesheet. Now, I’m wondering how you should deal with invalid intermediate results: Say, I’m going from markdown to jats xml using pandoc, but the xml produced by pandoc is not perfect and I need a couple of steps to produce the correct and valid (!) result. The intermediate steps consists mainly of renaming elements and moving them around. Most of the intermediate versions won’t be valid as they use elements that aren’t valid JATS.
>  
> How would you deal with that situation (that we produce invalid xmls along the way)? Just ignore it? (Which would be what I currently do since I mostly care about the final document and not so much about interdiary files.)
>  
> Best,
> Denis