Re: [search-ws] Back to Record Packing

["Ray Denenberg, Library of Congress" <rden@loc.gov>]

So, following up on yesterday's message ... according to this plan, 
recordPacking would have these values:

- string
- xml
- loose
-strict

There are the following issues/notes.

- First of all, string and xml would apply only when the serialization is 
xml. loose and strict would apply to any serialization.

- a value of string or xml (assuming an xml serialization) should not 
preclude a value of loose or strict, that is, string vs. xml and loose vs. 
strict are orthogonal.  This means that the parameter is repeatable, or can 
occur zero to two times.
More precisely:
o For an XML serialization it may occur zero, one, or two times. If the 
value 'string' is supplied then the value 'xml' may not be supplied, and if 
the value 'loose' is supplied then the value 'strict' may not be supplied.
o For a non XML serialization it may occur zero or one time. Neither of the 
values 'string' or 'xml' may  be supplied, and if the value 'loose' is 
supplied then the value 'strict' may not be supplied.

- Finally,  the orthogonality seems to me to make this parameter/element 
sematics overloaded and akward.

Nevertheless, I'm willing to go with this if everyone else is.   However I 
need feedback on this soon, please.

--Ray




----- Original Message ----- 
From: "Ray Denenberg, Library of Congress" <rden@loc.gov>
To: "OASIS SWS TC" <search-ws@lists.oasis-open.org>
Sent: Tuesday, April 13, 2010 5:50 PM
Subject: [search-ws] Back to Record Packing


> Trying to get  the "Record Packing" discussion back on the fast track, the 
> discussion veered off to issues that may be orthogonal, or may not, for 
> example how to pair the media type and schema.
>
> But Tony, is it fair to say that the essence of what you are trying to 
> achieve is (as you have put it) "allow for the possibility of the record 
> data to rattle around a fixed skeleton."   And, in essence, you're 
> proposing "unpacked" as a value of recordPacking to mean this?
>
> Is this a fair summary?
>
> If so, recall that in the last message on this, I suggested that  If we 
> allow the client to say "unpacked" we also need to allow it to say
> "packed", because we're going to want the server to be able to designate 
> (in Explain) the default, so that if it designates json/(SRU)/unpacked as 
> the default then the client doesn't have to say "unpacked" if that's what 
> it wants but it has to say "packed" if that's what it wants. Right?
>
> I also suggested that  we change "packed" and "unpacked" to "loose" and 
> "strict"?
>
> So if the client requests "loose" this means that the server is at liberty 
> to supply a variation of the schema; if "strict" is requested then the
> server should suppy the schema according to its definition.
>
> --Ray
>
> ---------------------------------------------------------------------
> To unsubscribe from this mail list, you must leave the OASIS TC that
> generates this mail.  Follow this link to all your TCs in OASIS at:
> https://www.oasis-open.org/apps/org/workgroup/portal/my_workgroups.php