[Dev] New version of CLARIN-FCS endpoint conformance tester

[Oliver Schonefeld]

Hi Olha,

Am 14.11.2013 12:01, schrieb Olha Shkaravska:
[..]
> And now a technical question. There is a random "abrakadabra" search
> request generated by your tester, on which our search library produces an
> error:
>
> 2013-11-14 11:51:22,347 INFO
> [nl.mpi.annot.search.cqlsearch2.SRUSearchResultSetImpl#handleRequest:112]
> - SEARCHING FOR:xVdMdmBhAryLoJJo
> 2013-11-14 11:51:22,372 ERROR [Searchlib.MultiPostgresQuery#getHits:282] -
> Returned 0 real hits while -2147483646 were requested, offset: 2147483646
> (caller too optimistic about stats?)
> 2013-11-14 11:51:22,374 INFO
> [eu.clarin.sru.server.SRUServer#handleRequest:155] - Sending fatal
> diagnostic 'info:srw/diagnostic/1/61' with message 'First record position
> out of range'
>
> Is it something what you would expect?

Short answer: yes

Long Answer: the tester deliberately sends some requests to provoke
   error conditions in the endpoint to check if it handles these
   correctly.
   A assume the snippet from the log is for the test case, where it's
   testing, if first record out of range is handled correctly. It does
   so, by sending a query that (hopefully) never matches (random string)
   and asks the endpoint to deliver records starting with a ridiculously
   high offset for the first record. The correct behavior in this
   situation is to send the "first record out of range" diagnostic.
   From the SRU point-of-view, your endpoint behaves correctly, because
   it sends this diagnostic. The error message is raised in you search
   library code. You can either "just live with it" or try to change the
   search library not to emit an error message in that case.

Best,
   Oliver
-- 
Oliver Schonefeld
Institut für Deutsche Sprache, Zentrale Forschung
R5, 6-13, D-68161 Mannheim
+49-(0)621-1581-451 | http://www.ids-mannheim.de