Related to IR-365 ("Investigate, possibly remove "skipped records" in IRSpy toroid") When probing for record syntaxes, try an all-records search. This gives us a chance of finding something in Russian-language targets such as http://irspy.indexdata.com/full.html?id=Z39.50%3Ans1.gbs.spb.ru%3A212%2Fnmagpl in which we can't find any records using the standard English-language searches.
Record::Fetch test registers a fail when the syntax actually retrieved does not match that requested. Fixes bug IR-324.
Attempt XML before GRS-1.
Reorder record-syntaxes probed to reduce likelihood of early failure.
@syntax made global.
_fetch_record() now sets the temporary value of the preferredRecordSyntax option into the connection rather than the result-set, avoiding a permanent override. This fixed bug #3534.
Remove extraneous comments.
enable OPAC record syntax test again, see 8e8489f6c6191b3335adb1f1b285d9c7ad724023
no redundant timeouts logs
Stop testing a target if we got to many timeouts (>= 3). See bug #3382
remove stale $Id$
CHange 1016 search to "water"
Add new test-search.
Use connection-level diagnostic if there is no record.
Use Task::Search->render_query()
Fix result-set leak.
Cope with $record not set at all (though IIRC this should never happen. But it did, and broke a looong run.) _/|_ ___________________________________________________________________ /o ) \/ Mike Taylor <mike@indexdata.com> http://www.miketaylor.org.uk )_v__/\ He's right you know -- the lad's got a point!
Use ZEND only for callbacks, rather than RECV_*
Better logging for failed searches.
Fix variable-name typo, caused infinite retry loop after bad search.