Ticket #665 (closed task: fixed)

Opened 15 years ago

Last modified 13 years ago

[M] Schema namespace correction

Reported by: selatham Owned by: sdonegan
Priority: desirable Milestone: Reporting
Component: discovery Version:
Keywords: WS-Discovery2 MDIP Cc:

Description (last modified by selatham) (diff)

We cannot guarantee that Data Providers will put the right namespace declaration in their discovery records. We will have to try to accept whatever records they give us. We will have to replace any namespace declarations with a standard one that we know works in NDG, plus remove schemalocations so that exist doesn't try to validate and reject.

Change History

comment:1 Changed 15 years ago by selatham

  • Status changed from new to assigned
  • Description modified (diff)

Created a SchemaNameSpace? corrector. Applied in the ingest scripts. Note applies to all ingests from now on. Does DIF, needs extending for MDIP.

comment:2 Changed 15 years ago by domlowe

I've just encountered this exist validation issue too.

Why don't we just keep copies of the schemas (MOLES, CSML etc) in Exist, then documents can be validated while in Exist or during ingestion.

In the ingest script the schemalocation attribute would need to be corrected to point to the version in Exist rather than removed entirely. You would still need to remove rogue namespace declarations.

I think having the ability to validate documents in-situ in exist will prove useful, e.g. if we start editing documents in exist (either by hand or with software), but this validation is only possible if we store the schemas in exist.

We would probably have to be pretty strict on versioning and only update the exist schema with a svn tagged change in version number.

comment:3 Changed 15 years ago by selatham

We could do this. In discovery ingest we do need a stage where we validate records against the 'proper' schema and reject with some sort of report going back to DataProviders?. This could utilise exist validation, or I might want to do it in a separate script, haven't decided yet.

comment:4 Changed 14 years ago by selatham

  • Priority changed from blocker to desirable
  • Keywords India removed

MDIP bit is 'desirable'. Other bit gone to ticket #889

comment:5 Changed 13 years ago by lawrence

  • Owner changed from selatham to sdonegan
  • Status changed from assigned to new

Is this still relevant?

comment:6 Changed 13 years ago by sdonegan

  • Status changed from new to closed
  • Resolution set to fixed

As harvested records from data providers are now ingested into a postgres, so no problem with exist trying to validate (& in any case validation had been turned off on exist on the discvery service anyway). Effectively fixed, I think.

Note: See TracTickets for help on using tickets.