[saxon] XQJ: module resolver

[Florent Georges <darkman_spam@ya...>]

  Hi

  I'm using the XQJ implementation of Saxon, in the package
net.sf.saxon.javax.xml.xquery.  I've looked for a way to set
a "module resolver" (either from the namespace URI or the
location hint), but didn't find anything.

  Did I miss something?

  PS: I'm open to Saxon-specific, even if I'd prefer to
remain in XQJ.

  Regards,

--drkm























      
	
		
___________________________________________________________________________ 
Découvrez une nouvelle façon d'obtenir des réponses à toutes vos questions ! 
Profitez des connaissances, des opinions et des expériences des internautes sur Yahoo! Questions/Réponses 
http://fr.answers.yahoo.com

[saxon] XQJ: module resolver

[Florent Georges <darkman_spam@ya...>]

  Hi

  I'm using the XQJ implementation of Saxon, in the package
net.sf.saxon.javax.xml.xquery.  I've looked for a way to set
a "module resolver" (either from the namespace URI or the
location hint), but didn't find anything.

  Did I miss something?

  PS: I'm open to Saxon-specific, even if I'd prefer to
remain in XQJ.

  Regards,

--drkm























      
	
		
___________________________________________________________________________ 
Découvrez une nouvelle façon d'obtenir des réponses à toutes vos questions ! 
Profitez des connaissances, des opinions et des expériences des internautes sur Yahoo! Questions/Réponses 
http://fr.answers.yahoo.com

Re: [saxon] XQJ: module resolver

[Michael Kay <mike@sa...>]

Generally I've tried to resist the temptation to implement enhancements =
to
the published XQJ interface, until such time as there is a final spec =
and we
can see what's really missing. It's all been a lot slower than I was =
hoping,
but I think I'm going to hold out and see what emerges. If I start =
adding
things it will almost certainly make the forwards transition more =
difficult
when we do see a definitive spec.

Michael Kay
http://www.saxonica.com/=20

> -----Original Message-----
> From: saxon-help-bounces@...=20
> [mailto:saxon-help-bounces@...] On Behalf=20
> Of Florent Georges
> Sent: 11 May 2007 15:01
> To: Saxon Help SF list
> Subject: [saxon] XQJ: module resolver
>=20
>   Hi
>=20
>   I'm using the XQJ implementation of Saxon, in the package=20
> net.sf.saxon.javax.xml.xquery.  I've looked for a way to set=20
> a "module resolver" (either from the namespace URI or the=20
> location hint), but didn't find anything.
>=20
>   Did I miss something?
>=20
>   PS: I'm open to Saxon-specific, even if I'd prefer to remain in XQJ.
>=20
>   Regards,
>=20
> --drkm
>=20
>=20
>=20
>=20
>=20
>=20
>=20
>=20
>=20
>=20
>=20
>=20
>=20
>=20
>=20
>=20
>=20
>=20
>=20
>=20
>=20
>=20
>=20
>      =20
> =09
> 	=09
> ______________________________________________________________
> _____________
> D=E9couvrez une nouvelle fa=E7on d'obtenir des r=E9ponses =E0 toutes=20
> vos questions !=20
> Profitez des connaissances, des opinions et des exp=E9riences=20
> des internautes sur Yahoo! Questions/R=E9ponses=20
> http://fr.answers.yahoo.com
>=20
> --------------------------------------------------------------
> -----------
> This SF.net email is sponsored by DB2 Express Download DB2=20
> Express C - the FREE version of DB2 express and take control=20
> of your XML. No limits. Just data. Click to get it now.
> http://sourceforge.net/powerbar/db2/
> _______________________________________________
> saxon-help mailing list
> saxon-help@...
> https://lists.sourceforge.net/lists/listinfo/saxon-help

Re: [saxon] XQJ: module resolver

[Florent Georges <darkman_spam@ya...>]

Michael Kay wrote:

> Generally I've tried to resist the temptation to implement
> enhancements to the published XQJ interface, until such
> time as there is a final spec and we can see what's really
> missing.

  That makes sense.  I should be able to use a precise
location hint as for now.  If not, I will have to use the
net.sf.saxon.query package I guess.

  I'm not familiar with the XQJ development process, but
I'll try to see if it could be interesting to report the
issue as a user.

  Thank you for your response.  Regards,

--drkm























      _____________________________________________________________________________ 
Ne gardez plus qu'une seule adresse mail ! Copiez vos mails vers Yahoo! Mail 

Re: [saxon] XQJ: module resolver

[Michael Kay <mike@sa...>]

>   I'm not familiar with the XQJ development process, but I'll 
> try to see if it could be interesting to report the issue as a user.

Expect to wait up to a year for a response if you do (you don't even get an
acknowledgement, let alone a tracking number, or someone telling you that
your custom is really important to them...)

Michael Kay

Re: [saxon] XQJ: module resolver

[Florent Georges <darkman_spam@ya...>]

Michael Kay wrote:

> Expect to wait up to a year for a response if you do

  I've just sent it, we will see...  The first weird thing is that I
didn't find any issue list nor public mailing list, only a private
email address for comments.

  I'll forward you the response if I ever get one.

  Regards,

--drkm





















	

	
		
___________________________________________________________________________ 
Découvrez une nouvelle façon d'obtenir des réponses à toutes vos questions ! 
Profitez des connaissances, des opinions et des expériences des internautes sur Yahoo! Questions/Réponses 
http://fr.answers.yahoo.com

Re: [saxon] XQJ: module resolver

[Florent Georges <darkman_spam@ya...>]

Florent Georges wrote:

> Michael Kay wrote:

  Michael,

  This is a followup to an email from the 11th of May this year,
about the ability to provide a custom resolver for module imports
through the XQJ API.

> > Expect to wait up to a year for a response if you do

>   I've just sent it, we will see...  The first weird thing is
> that I didn't find any issue list nor public mailing list, only
> a private email address for comments.

>   I'll forward you the response if I ever get one.

  I get a response yesterday in the evening from Jim Melton.
Actually I got a first response on the 17th of May, but in
private and I didn't ask to publish his response (my fault).
Here is his response to the followup I sent to him.  Other
messages are below for reference.

  Hope that helps,

--drkm


From:    "Jim Melton"
To:      "Florent Georges"
Cc:      jsr-225-eg@...
Date:    Wed, 25 Jul 2007 12:53:42 -0600
Subject: Response to your comment on XQJ dated 2007-05-11 and
2007-05-17

Florent,

The JSR 225 (XQJ) Expert Group has considered the comment that you
submitted regarding module import support, and has directed me to
send you this response.

<comment>
It seems that XQJ 0.5.0, March 3, 2006 doesn't specify any way of
resolving a module import from its namespace URI and location hint to
an actual, real location (or source, or whatever).
</comment>

<additionalComment>
Although I've tried to understand the reasons behind this decision, I
cannot find any one.  Mainly for XQuery modules.  I understand that
by default the resolving is implementation-defined.  But I don't
understand why no standard way to change this behavior is provided to
the user.  As for instance a resolver that returns an equivalent of
the Source interface for XSLT in javax.xml.transforms.

I think this solution has proved its usefulness.

Could you please explain a little bit the rationale behind this
decision, or give a pointer to a public resource that explains it?
</additionalComment>

<moreAdditionalComment>
>XQJ defines the concept of an XQDataSource, and implementers must
>define the properties appropriate for their implementation.

So such a useful feature will be implementation-defined.  I guess
this will result in a mess each time a user will want to write an
implementation-agnostic program, won't it?
</moreAdditionalComment>


<response>
XQJ is designed to target both a client/server environment and a
"co-located" XQuery engine environment.  How the locations of XQuery
modules are resolved and how XQuery external Java function are
resolved and loaded must/should be handled by the underlying XQuery
engine and its managed XML data source. It is not the issue of an API
to handle this capability.  Consequently, it is beyond the scope of
XQJ 1.0.  However, a possible future version of XQJ may support an
extension API specifically for a co-located engine architecture in
which we may add module and XQuery external Java function API support.
</response>


You also asked "PS: My email was following a discussion on the Saxon
ML.  I don't know why I didn't put the list in copy of my
email.  Would you mind if
I forwarded your response to this (public) list?  To keep the guys
there informed."

We have no objection at all to your forwarding this response to that
public list.


Hope this helps,
   Jim


From:    "Florent Georges"
To:      "XQJ JSR 225 comments" <jsr-225-comments@...>
Date:    Fri, 11 May 2007 17:04:42 +0200
Subject: Module import resolver

  Hi

 I didn't find other email address (or any other way) to
report issues about the actual state of XQJ.  In particular,
I didn't find any public mailing list.  Please forgive me if
I missed something.

 It seems that XQJ 0.5.0, March 3, 2006 doesn't specify any
way of resolving a module import from its namespace URI and
location hint to an actual, real location (or source, or
whatever).

 Is it intentional?

 Regards,

--
Florent Georges


From:    "Jim Melton"
To:      "Florent Georges"
Cc:      "XQJ JSR 225 comments" <jsr-225-comments@...>
Date:    Thu, 17 May 2007 14:27:20 -0600
Subject: Re: Module import resolver

Thanks very much for your question.

It was intentional that XQJ does not define a mechanism to resolve
URIs to modules, documents, collection, etc.  XQJ defines the concept
of an XQDataSource, and implementers must define the properties
appropriate for their implementation. If it makes sense for your
implementation to offer the ability to hook in a module URI resolver
mechanism, then adding a property to your XQDataSource should be
considered. For example, a property "moduleURIResolver" with a string
value identifying a Java class might be such a property.

By the way, the email address that you used is the correct one to use
for comments on this JSR.

Hope this helps,
   Jim


From:    "Florent Georges"
To:      "Jim Melton"
Cc:      "XQJ JSR 225 comments" <jsr-225-comments@...>
Date:    Thu, 17 May 2007 23:27:46 +0200
Subject: Re: Module import resolver

On 5/17/07, Jim Melton wrote:

 Hi Jim,

 Thank you very much for your response.

> It was intentional that XQJ does not define a mechanism to resolve
> URIs to modules, documents, collection, etc.

 Although I've tried to understand the reasons behind this decision,
I cannot find any one.  Mainly for XQuery modules.  I understand that
by default the resolving is implementation-defined.  But I don't
understand why no standard way to change this behavior is provided to
the user.  As for instance a resolver that returns an equivalent of
the Source interface for XSLT in javax.xml.transforms.

 I think this solution has proved its usefulness.

 Could you please explain a little bit the rationale behind this
decision, or give a pointer to a public resource that explains it?

> XQJ defines the concept of an XQDataSource, and implementers must
> define the properties appropriate for their implementation.

 So such a useful feature will be implementation-defined.  I guess
this will result in a mess each time a user will want to write an
implementation-agnostic program, won't it?

 PS: My email was following a discussion on the Saxon ML.  I don't
know why I didn't put the list in copy of my email.  Would you mind if
I forwarded your response to this (public) list?  To keep the guys
there informed.

 Thanks again for your response.  Regards,

--
Florent Georges



      _____________________________________________________________________________ 
Ne gardez plus qu'une seule adresse mail ! Copiez vos mails vers Yahoo! Mail