X-Git-Url: http://git.indexdata.com/?p=irspy-moved-to-github.git;a=blobdiff_plain;f=Changes;h=bc0b4e3fa5f4531e08f249ee54f9c795c6745d2a;hp=cad90359ba86b0bd0ad2180c0d4ce71d143879f5;hb=388b377406c84b03b6c4d9f9ddfed7e0c2679252;hpb=245354a09b5fca32e297406079cdf7c352421875 diff --git a/Changes b/Changes index cad9035..bc0b4e3 100644 --- a/Changes +++ b/Changes @@ -1,8 +1,113 @@ -$Id: Changes,v 1.17 2007-07-09 09:41:28 mike Exp $ - Revision history for Perl extension ZOOM::IRSpy. -1.00 Mon Jul 9 10:40:38 BST 2007 +1.03 [IN PROGRESS] + - Liberalise hostname validation regexp, so that hyphens are + allowed in the name. + - Add Search::Bath to Search::Main test. + - Assess reliability on the basis of the last sixty days only. + - Test for simple access-point-only searches when Bath-profile + searches fail. + - Introduce the notion of rules, which are applied under + specified conditions resulting in specified actions. As a + motivating example, when working on a server with an + implementation name (or ID or version) matching a specified + regular expression, skip a named test: this is used to avoid + having the test of catalog.dallaslibrary.org:210/PAC hang + when we attempt to test its CQL capabilities. + - Sample rules file etc/dallas.rules + - The Web UI sets $ZOOM::IRSpy::xslt_max_depth to 3500, just + like the overnight test-script. This is a hideous hack to + cope with the fact that XSLT can't iterate, and explodes if + you ask it to recurse across a list of more than a half + dozen elements. + - New script bin/irspy-xpath.pl -- a simple wrapper for + running an XPath with IRSpy-related namespaces established. + - Record::Fetch test registers a fail when the syntax actually + retrieved does not match that requested. Fixes bug IR-324. + - Record::OPAC test check that record returned in piggyback + after OPAC-record request really is an OPAC record. Fixes + bug IR-331. + - Add new test, Record::PiggyBack, with support for viewing, + editing and storing results. Fixes bug IR-333. + - IDs used in URLs for full-record links are properly + CQL-quoted. Fixes part of IR-303. + - Move aside the old, unused lib/ZOOM/Pod.pm. Will be removed + from distribution entirely in a subsequent release. + - Support new "irspy_data" log-level to register information + written to target description record. + - need a nagios alert script to check irspy updates, IR-336 + - Better logging for skipped tests. + - No failure within the Search::Bib1 test causes following + sibling tests to be skipped. Fixes bug IR-338. + - Some cosmetic improvements, e.g. better banner image and + Cabin font. + +1.02 Wed Jul 7 16:43:36 BST 2010 + - Enhance setrlimit program so that it can set maximum + address-space size as well as number of file descriptors. + - Add new restrict_modulo() method to restrict tests to those + targets whose ID hashes to a number satisfying + hashval mod n == i + - Add new command-line options -m , to invoke + restrict_modulo(). This enables us to set up cron jobs to + test one seventh of the targets each night. + - Modify Zebra configuration to expect Z39.50 query terms to + be encoded using UTF-8 (which is what the Web UI uses) + rather than the default ISO-8859-1. Fixes bug #3420. + - Add new test Record::OPAC which checks the ability to return + multiple OPAC records in a piggy-backed Search Response. + Test result is saved in the new element + + - Add new test Search::Bath which checks the ability to + respond to Bath-profile searches. Test result is saved in + the new element + for NAME in "author", "title", "subject" and "any". + - The web UI displays this new information in full-record + view. + - A target can be disabled from participating in testing by + setting the irspy:status/i:disabled xmlns element. This can + be set and unset from the Admin UI. It's useful for + omitting targets that make long test runs take too long. + - Improve input validation for new targets. + - New and enhanced admin scripts (in ./bin) for IRSpy + maintenance. + - Fix search for zeerex.reliability = 0, bug #3399 + - Improve timeout handling of targets. Stop testing a target + if we got to many timeouts (>= 3). Otherwise we have to wait + up to an hour to finish all tests for a broken target. A + complete tests of all targets is now much faster and more + reliable. + - Ignore case for record syntax field. This solves the problem that + you cannot search for record syntax "USmarc" if the server + response was in lowercase "usmarc". + - Run only one instance of the IRSpy script at a time. + - Use iconv character-set name "UTF-8" rather than "UTF8" as + the latter is for some reason not supported by the iconv + package in the Macports collection. + - _irspy_identifier2target() recognises and allows for + comma-separated ZOOM-style options at the front of IRSpy + identifiers. + - Fix bug #3534, which caused targets to be probed for + record-syntax support using the wrong + preferredRecordSyntax. + - Reorder record-syntaxes probed to reduce likelihood of early + failure. Ameliorates, but does not fix, bug #3548. + - irspy.pl and irspy-dump.pl are now installed as binaries. + - Add sample Apache2 configuration to distribution. + - robots.txt used to ameliorate web-site spamming. + +1.01 Mon Jul 16 12:51:19 BST 2007 + - Web UI pages to add or upload a new target are now publicly + accessible -- although admin privileges are still required + to edit the newly added records, since there is no simple + way to know that the person editing a record is the same + person who first added it. + - When adding new records via the Web UI, the port number is + guessed from the protocol if not explicitly stated. + - Note that Web UI's ZeeRex upload expects just the ZeeRex + record, not an entire SRU explainResponse. + +1.00 Mon Jul 9 10:42:06 BST 2007 - First public release. - Many, many changes since 0.02.