The original fix to the OSID code is good.

The secondary issues that came up (see the comments for details) really are out-of-scope for this issue.

The change described in my comment of 19-Oct-2006 appears to work as intended. The "bad" data (the record with the missing title) is simply passed up to the consumer, and the OSID continues to retrieve search results.

However, I'm not altogether happy with the final product. The search described (looking in Blackwell for the string "pinecone") now produces an estimate of 67 hits, but only 36 records are displayed in the search result list.

Specifically, with a page size of 10, the final page of results shows: "Viewing 31 - 36 of 67". The "next page" arrow is disabled.

What happened? The search results now contain 31 identical records. The "OpenURL duplication check" in the Citation BaseSearchManager code discards these duplicate records, but no indication of this is provided to the user.

This is probably out-of-scope for SLIB-25, but I'll continue to leave this issue open for now.

I've changed the OSID code to log a warning and continue when empty "mandatory" fields are encountered.

The record with the missing data will be provided to the caller "as is". We no longer throw an exception.

This will be included in the phase-1-003 build. If the change behaves well, I'll mark the issue as resolved.

There will be times when we cannot return any more results though the search engine tells us that there should be more results to return. It may be that the search timed out, or the search source is temporarily off-line, or any number of possible problems. Gaurav will send Mark and Sharon a list of the problems that get reported back to us. The question then is what should be reported to the user and where in our interface it should be reported.

I'm assigning to Steve and to Gaurav. If you can identify any cases in which the OSID impl is bailing out of a search (throwing an exception) where we might be able to continue, please deal with those for the next tag. If you find that the SearchManager is misinterpreting any exceptions, please let me know.