Print

Print


We expect to reference two or more different versions of a schema from a
single instance document and all our systems are designed to support
that.  The use case goes like this:  event metadata for an object is
cumulated in blocks in a METS wrapper; each block reflects the version
of the event metadata schema in use at that time the event was
generated.  We might in the future migrate the embedded metadata to a
later version, particularly if the version is not backward compatible
but we aren't promising to do it instantly so we expect to have
different versions in the same METS wrapper.  Another example might be
technical MD; we wouldn't necessarily discard the earlier metadata if we
created new metadata to a newer standard so there could well be two
blocks using different versions of the same schema.  

Because of this, we strongly favor being absolutely clear about what
version of a schema is being referenced at any given point.  All our own
internal identifiers include version information.  Our sanity will
suffer if we don't know exactly what is intended at all times. 

Evan 

==================================
Evan Owens, Chief Technology Officer
Portico  www.portico.org
[log in to unmask]  (609) 258 8230
228 Alexander Street, Princeton NJ 08550
   

 

 

-----Original Message-----
From: PREMIS Implementors Group Forum [mailto:[log in to unmask]] On Behalf Of
Christopher Vicary
Sent: Friday, July 08, 2005 3:49 PM
To: [log in to unmask]
Subject: Re: [PIG] namespaces and versions (fwd)

Actually I was saying that keeping the same namespace identifer across
versions is OK because parsers will identify any definition conflicts if
the same namespace is imported more than once from different schema
locations. It really depends on what we're trying to accomplish. I can't
come up with a good reason to reference 2 or more different versions of
a namespace from a single instance document or schema, so I would rather
disallow it. Maintaining the same namespace across versions facilitates
this. But if the goal is to reference different namespace versions from
a single document, then I agree that versioning the namespace identifier
is a better approach. If so, the METS people might want to take note
since they keep the same namespace identifier across versions (versions
only differ according schema location).

Chris

Ray Denenberg, Library of Congress wrote:

>I think maybe we're saying the same thing.  I got into this discussion 
>late, Rebecca was telling me that premis was tentatively considering 
>using a single namespace across versions (even major versions) and I 
>told here I thought it was a bad idea. Seems you do to.  Please see
separately posted
>message: (subj: versioning/namespacing/location).   --Ray
>
>
>From: "Christopher Vicary" <[log in to unmask]>
>
>
>>Off the top of my head, I'm not sure that versioning namespaces as 
>>described below (using mods) presents any *new* challenges. Let's say 
>>that namespaces are required. When a schema imports elements from 
>>other schemas, you must include the external namespace prefix for 
>>those elements in instance documents. Since the namespace identifiers 
>>and their prefixes are unique, there is no problem here in determining

>>which element definition to use. Again, I mocked up a test for this 
>>situation using XMLSpy and there was no problem because the validating

>>parser required that namespace prefixes be used.
>>
>>Let's say for the sake of argument that there is a way avoid using 
>>namespace qualification. In this situation I believe there would be a 
>>problem, there would be no way to tell which definition of the element

>>should be used. The important thing to remember, however, is that this

>>would always a problem, regardless of whether the referenced 
>>namespaces are related versions. So the situation described below in 
>>which we have
>>2 namespaces, mods1 and mods2, with like-named elements being 
>>referenced from each namespace is no different from having 2 unrelated

>>namespaces, say foo and bar, that contain like-named elements that are

>>referenced from a 3rd shema (eg. foo:title and bar:title). 
>>Semantically, they present the same problem. That's probably one of 
>>the reasons namespaces are used in the first place.
>>
>>Now the situation described below is actually different from what was 
>>proposed for PREMIS (and what is used by METS). Below, each namespace 
>>identifier is unique, http://www.loc.gov/mods1 and 
>>http://www.loc.gov/mods2, and so are the schema locations. What we 
>>were proposing for PREMIS is to maintain the same namespace identifer,

>>but version the schema locations. The possibility for confusion here 
>>seems greater, but I covered why I think even that situation is not a 
>>problem in my previous email.
>>
>>Again, I want to reiterate that the tests I created to prove the 
>>concepts were run using only XMLSpy which is my XML editor du jour 
>>(hey it's free and seems to work well). I have noticed cases where 
>>XMLSpy handles namespaces differently than say Xerces, but if anything

>>XMLSpy seems to fall on the strict side of things. If I have some 
>>time, I may try out some other parsing tools.
>>
>>It might be nice to get another opinion on this topic, Jerry are you 
>>out there?
>>
>>-Chris
>>
>>
>>Ray Denenberg, Library of Congress wrote:
>>
>>
>>
>>>The scenario I had in mind is represented by this sample schema:
>>>
>>>
>>>
>>......................................................................
.....
>>
>>
>.
>
>
>>>...............
>>><xsd:schema
>>>
>>>xmlns:mods1=http://www.loc.gov/mods1
>>>xmlns:mods2="http://www.loc.gov/mods2"   ......>
>>>
>>><xsd:import namespace="http://www.loc.gov/mods1"
>>>schemaLocation="mods1.xsd"/>
>>><xsd:import namespace="http://www.loc.gov/mods2"
>>>schemaLocation="mods2.xsd"/>
>>>
>>><xsd:element name="root">
>>><xsd:complexType>
>>><xsd:sequence>
>>><xsd:element ref="mods1:titleInfo"/>
>>><xsd:element ref="mods2:titleInfo"/>
>>><xsd:element name="otherElement1"/>
>>><xsd:element name="otherElement2"/>
>>><xsd:element name="otherElement3"/>
>>><xsd:element name="etc"/>
>>></xsd:sequence>
>>></xsd:complexType>
>>></xsd:element>
>>>
>>></xsd:schema>
>>>.........................................................
>>>
>>>In this case someone has constructed a (hypothetical) schema that 
>>>mixes elements from two mods versions (along with other elements),
>>>
>>>
>specifically,
>
>
>>>titleInfo, where (hypothetically) the titleInfo definition has 
>>>changed
>>>
>>>
>from
>
>
>>>(hypothetical) version 1 to (hypothetical) version 2.
>>>
>>>Does this help?
>>>
>>>--Ray
>>>
>>>
>>>
>>>
>>>
>>>
>>>
>>>
>>>>---------- Forwarded message ----------
>>>>Date: Wed, 29 Jun 2005 18:50:03 -0400
>>>>From: Christopher Vicary <[log in to unmask]>
>>>>Reply-To: PREMIS Implementors Group Forum <[log in to unmask]>
>>>>To: [log in to unmask]
>>>>Subject: Re: [PIG] namespaces and versions
>>>>
>>>>Sorry it's taken me so long to respond to this. I want to be sure I 
>>>>understand the scenario Ray is setting up. From what I gather from 
>>>>the final paragraph, there is a concern that a single schema may 
>>>>reference another namespace more than once with different schema 
>>>>locations attached to each reference. If we use METS as an example, 
>>>>there might be a namespace declaration in a schema's root element
that takes the form:
>>>>
>>>>xmlns:METS="http://www.loc.gov/METS/"
>>>>
>>>>and a corresponding location:
>>>>
>>>>xsi:schemaLocation="http://www.loc.gov/METS/
>>>>http://www.loc.gov/standards/mets/mets.xsd"
>>>>
>>>>Later in the same schema there might be a conflicting reference to 
>>>>the same namespace in an import statement:
>>>>
>>>><xs:import namespace="http://www.loc.gov/METS/"
>>>>schemaLocation="http://www.loc.gov/standards/mets/version13/mets.xsd
>>>>"/>
>>>>
>>>>Notice that while the namespace identifiers are the same, the 
>>>>locations are not.
>>>>
>>>>I wasn't sure exactly how a validating parser would handle this, so 
>>>>I created a test document and validated it using XMLSpy. It looks 
>>>>like the XMLSchema designers (or perhaps XML parser developers) have

>>>>already considered this situation, because the referencing schema
was invalid.
>>>>The validator recognized that there are conflicting definitions 
>>>>within the same namespace. It appears as though we don't have to 
>>>>worry about this specific situation.
>>>>
>>>>I took it a step further and created a schema (test1.xsd) that 
>>>>references the METS namespace at the location 
>>>>http://www.loc.gov/standards/mets/mets.xsd. It also references 
>>>>another schema, test2.xsd. test2.xsd also references the METS 
>>>>namespace, but at the location 
>>>>http://www.loc.gov/standards/mets/version13/mets.xsd. In this 
>>>>hierarchical scenario, test1.xsd is invalid, again due to the 
>>>>definition conflict. test2.xsd, which only sees one reference to the
METS namespace, is valid. To me, this does not seem to be problematic.
>>>>
>>>>I admit that I may not completely understand the situation outlined 
>>>>by Ray, if so, can someone provide a concrete example of the
problem?
>>>>Another caveat, I only had time to test this using one tool, XMLSpy,

>>>>other validating parsers may treat this situation differently.
>>>>
>>>>Chris Vicary
>>>>
>>>>
>>>>Ray Denenberg, Library of Congress wrote:
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>>I just got subscribed to this list so pardon me if I'm out of
context.
>>>>>
>>>>>Different schemas need different namespaces unless you can 
>>>>>guarantee
>>>>>
>>>>>
>that
>
>
>>>>>any name occuring in both schemas will have a common definition. 
>>>>>Thus
>>>>>
>>>>>
>if
>
>
>>>>>
>>>>>
>>>you
>>>
>>>
>>>
>>>
>>>>>move from one version to another, take MODS for example, and you 
>>>>>change
>>>>>
>>>>>
>>>>>
>>>>>
>>>the
>>>
>>>
>>>
>>>
>>>>>definition of, say, titleInfo, then if you want to maintain the 
>>>>>same namespace, you need to change the name of element titleInfo.
>>>>>
>>>>>I can't recall the exact example but in one of the MODS revisions 
>>>>>we concluded that it was much more disruptive to change the element

>>>>>names
>>>>>
>>>>>
>>>>>
>>>>>
>>>(e.g.
>>>
>>>
>>>
>>>
>>>>>titleInfo --> newTitleInfo) than to change the namespace.
>>>>>
>>>>>A generic namespace with different locations doesn't get around
this.
>>>>>
>>>>>
>You
>
>
>>>>>cannot guarantee that sometime in the future some third schema 
>>>>>won't reference names from two different schemas with the same
namespace.
>>>>>
>>>>>
>When
>
>
>>>>>that happens, if the reference an element that has different
>>>>>
>>>>>
>definitions
>
>
>>>>>
>>>>>
>>>in
>>>
>>>
>>>
>>>
>>>>>the two schemas, you'd be in trouble.
>>>>>
>>>>>--Ray
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>
>>>>>
>>>>--
>>>>
>>>>
>>>---------------------------------------------------------------------
>>>----
>>>
>>>
>>>>Christopher Vicary: (352)392-9020 ext. 323
>>>>e-mail: [log in to unmask]
>>>>fax:    (352)392-9185
>>>>
>>>>
>>>---------------------------------------------------------------------
>>>----
>>>
>>>
>>>>   _/_/_/_/    _/_/_/  _/          _/_/
>>>>  _/        _/        _/        _/    _/
>>>> _/_/_/    _/        _/        _/_/_/_/   F C L A
>>>>_/        _/        _/        _/    _/    5830 NW 39th Avenue
>>>>_/          _/_/_/  _/_/_/_/  _/    _/     Gainesville, Fl 32606
>>>>
>>>>
>>>---------------------------------------------------------------------
>>>----
>>>
>>>
>>>>F l o r i d a   C e n t e r   f o r   L i b r a r y   A u t o m a t
i o
>>>>
>>>>
>n
>
>
>>>---------------------------------------------------------------------
>>>----
>>>
>>>
>>>>
>>>>
>>>>
>>>
>>>
>>>
>>--
>>----------------------------------------------------------------------
>>--- Christopher Vicary: (352)392-9020 ext. 323
>>e-mail: [log in to unmask]
>>fax:    (352)392-9185
>>----------------------------------------------------------------------
---
>>    _/_/_/_/    _/_/_/  _/          _/_/
>>   _/        _/        _/        _/    _/
>>  _/_/_/    _/        _/        _/_/_/_/   F C L A
>> _/        _/        _/        _/    _/    5830 NW 39th Avenue
>>_/          _/_/_/  _/_/_/_/  _/    _/     Gainesville, Fl 32606
>>----------------------------------------------------------------------
---
>>F l o r i d a   C e n t e r   f o r   L i b r a r y   A u t o m a t i
o n
>>----------------------------------------------------------------------
>>---
>>
>>
>
>
>

--
------------------------------------------------------------------------
-
Christopher Vicary: (352)392-9020 ext. 323
e-mail: [log in to unmask]
fax:    (352)392-9185
------------------------------------------------------------------------
-
    _/_/_/_/    _/_/_/  _/          _/_/
   _/        _/        _/        _/    _/
  _/_/_/    _/        _/        _/_/_/_/   F C L A
 _/        _/        _/        _/    _/    5830 NW 39th Avenue
_/          _/_/_/  _/_/_/_/  _/    _/     Gainesville, Fl 32606
------------------------------------------------------------------------
-
F l o r i d a   C e n t e r   f o r   L i b r a r y   A u t o m a t i o
n
------------------------------------------------------------------------
-