Supplemental: Re: [osis-core] type on identifier and subject, syntax of content

[Chris Little]

I don't think I understand what purpose type serves on <identifier> if 
we're going to constrain it to be either "OSIS" or semantically vacuous.

I really don't understand the purpose of the prefix proposed as part of 
the content of identifier/subject, so I'll just identify some sample 
usage and you can let me know if it makes sense or correct me if not....

Take, for example, the book "The Apostolic Fathers" by JB Lightfoot. 
(You can follow along with me at catlog.loc.gov using the LCCN 
73173713.)  I should have (IMO) the following:

<identifier 
type="OSIS">private:Monograph.Lightfoot_JB.Apostolic_Fathers</identifier>
<identifier type="LCCN">73173713</identifier>
<identifier type="ISBN">3487046881</identifier>
<identifier type="LCCall">BR60.A62 L5213 1973</identifier>
<subject type="Dewey">270.1 20</identifier>
<subject type="LCSH">Christian literature, Early.</identifier>

I'm fine with the prefix in cases like my type="OSIS" identifier above. 
  There, I'm essentially identifying a work ID, and using "ccl:" to say 
that this is the work ID that _I_ claim.  This could be replaced with 
prefixes from groups who want to act as naming authorities if they 
should become established.

 From Todd's message....
> Either 
> 1) type="ISBN" plays no role in interpreting the system of identifiers
> that the body of <identifier> comes from AND a prefix is required
> OR 
> 2) type="ISBN" acts in the role of a work prefix and a work prefix is
> not necessary
> 
> In case #1, the type attribute only serves as a guide but does NOT add
> to the interpretation of the value in the body of <identifier>.
> 
> In case #2, the type attribute serves two purposes/roles, a work prefix
> is required in cases where type="OSIS" is used, AND conflicting
> "contexts" can be encoded (<identifier
> type="LOC">isbn:123456789</identifier>).

I have a particular problem with the example at the end here. 
Identifiers like this aren't of a "type" which should be identified as a 
naming authority.  ISBN numbers are identifiers of type "ISBN". 
Similarly, type="OSIS" doesn't really mean that the identifier is 
registered by BTG, it means it's of the OSIS "type".  That is, it 
conforms to whatever standards for identifier creation BTG defines.

Except for identifiers of type "OSIS", the contents of identifier and 
subject elements are unlikely to bear any resemblance to IDs, osisIDs, 
or any other ID-like value.  There's no guarantee the won't include 
colons, and many of them include spaces & other punctuation.  I don't 
feel there's any good reason to treat them /sort of/ like they're IDs by 
prepending prefixes.

They're just strings, and they should be easily parsed and make sense as 
simple strings without need of stripping prefixes.  I feel we should 
just be enclosing the data itself (the identifier value) in the 
<identifier> element--not the identifier plus a prefix.

--Chris