Re: Issues and imports

Hi Norm,

> Am 17.12.2016 um 14:33 schrieb Norman Walsh <ndw@nwalsh.com>:
> 
> Achim Berndzen <achim.berndzen@xml-project.com> writes:
>> I am currently having the same troubles. 
>> @Norm: Is it possible and would it help to raise the issues from
>> https://github.com/xproc/1.0-specification/tree/xproc20/langspec/xproc20
>> to xproc/1.1-specification? To my reading the two versions are almost
>> identical, so the issues on A should also be issues on B.
> 
> I raised some issues (per my action to make issues from the September
> minutes “pain points”). I don’t see a convenient way of moving the
> issues from the 1.0-specification repository to the 1.1 repo. :-(

I see. May be I will find the time to move some of them by hand, i.e. raising an issue with the text of the 1.0 issue in the 1.1 repo. Would that help or does it lead to confusion?

> 
>>> From my memory the biggest point with regard to the core specs is
>> interoperability of „p:import". The two processors may know the same
>> step/libraries, but use different uris to access them. I think we
>> should fix this, but currently I have different ideas how to do this.
>> Promise to come up with a proposal in Prague!
> 
> Would the problem be solved if we created a document, say,
> 
>  http://exproc.org/current/library.xpl
> 
> That contained the declarations for all the EXProc steps? Then
> pipeline authors could import that and all implementations would get
> the same declarations?
> 
> Implementations could go further and say that the /current/ URI is
> magic. An implementation, when it encounters that URI, can simply take
> as read that the current EXProc steps that it understands are
> declared. It wouldn’t have to hit the network at all.
> 
Yes, that is certainly a good solution for this special case. 

I thought about the more general case of an author creating a step library. Currently there is no way for her to say: Import my pipeline with this URI independently from where it is stored on the filesystem! Using the URI, an XProc processor might then look into am XMLCatalog or an EXPathPackage to actually resolve the file. We could think about an attribute „export-uri“ for p:library to declare this.

Did I make myself clear?  I have a glimpse of the problem and have to think more about it. Not to talk about a general solution.

Greetings from Germany,
Achim

------------------------------------------------
Achim Berndzen
achim.berndzen@xml-project.com

http://www.xml-project.com

Received on Saturday, 17 December 2016 14:09:33 UTC