Re: getPath proposal

[Len Gallagher <LGallagher@nist.gov>]

Query Team,

Attached are some comments on the proposed getPath() method definition 
distributed to this Query team last Friday.

   http://lists.oasis-open.org/archives/regrep-query/200110/pdf00003.pdf

It pains me to raise these issues, because I don't want to oppose this 
proposal, but some of these comments reflect serious unresolved issues:

Line 31 - minor
Shouldn't "Package" be "ClassificationNode"?

Lines 38-40 - minor
The reference to being similar to "file system paths" is not necessary and 
should probably be removed.

Lines 48-50 - Major
The BNF grammar for "ID" is very restrictive. It only allows the use of 26 
letters of the alphabet (both uppercase and lowercase), underscore, and 10 
digits.  But an "ID" is supposed to represent the "code" attribute of a 
ClassificationNode and its data type is "ShortName", which is defined in 
ebRIM to be an arbitrary text string of up to 64 characters, presumably 
from the Unicode character set.  How do we represent "codes" that use 
characters other than the 63 characters provided?  And if we do allow 
arbitrary characters in the code attribute for ClassificationNode 
instances, how do we treat the special XPATH characters (e.g. "/" and "*").

Lines 43 and 51 - Major
The value of the schemeId element in the BNF production for canonicalPath 
is defined to be the value of the "id" attribute of the 
ClassificationScheme instance that describes the classification scheme with 
this node. Some of us believe that it will be impossible to enforce an 
assumption that this "id" will be the same for all Registry 
implementations. See the discussion and responses at URL:
      http://lists.oasis-open.org/archives/regrep-query/200109/msg00062.html
Thus we will likely have getPath() representations that differ in the first 
element depending on which Registry it comes from! On top of this the value 
of "id" is a UUID (128 bits) that only a machine can process.

Line 65 - minor
Shouldn't the "Geography" reference be "Geography-id"?

Line 66 - minor
Shouldn't the "NorthAmerica" reference be "NorthAmerica-id"?

Line 68 - minor
Shouldn't the "Geography" reference be "Geography-id"?

Line 69 - minor
Shouldn't the "Asia" reference be "Asia-id"?

Line 70 - minor
Shouldn't the "Japan" reference be "Japan-id"?

I hope these issues can be resolved before we have to vote on this proposal.

-- Len


At 10:16 PM 10/5/01, Farrukh Najmi wrote:
>Team,
>
>In todays con call I was asked to re-submit my original proposal for the
>syntax of ClassificationNode getPath method and Path expression for
>filter queries as two separate proposals. The original proposal may be
>found at:
>
>  http://lists.oasis-open.org/archives/regrep-query/200109/msg00075.html
>
>A related comparison of path filters and other approached is at:
>
>http://lists.oasis-open.org/archives/regrep-query/200109/msg00047.html
>
>Attached are both proposal for your consideration.
>
>--
>Regards,
>Farrukh
>
>
>

**************************************************************
Len Gallagher                             LGallagher@nist.gov
NIST                                      Work: 301-975-3251
Bldg 820  Room 562                        Home: 301-424-1928
Gaithersburg, MD 20899-8970 USA           Fax: 301-948-6213
**************************************************************