Post ATvbVuGyXymwgoQAPw by dbs@code4lib.social
 (DIR) More posts by dbs@code4lib.social
 (DIR) Post #ATvbVuGyXymwgoQAPw by dbs@code4lib.social
       2023-03-24T01:58:58Z
       
       0 likes, 0 repeats
       
       Four days ago I contacted OCLC and Library and Archives Canada (LAC) to tell them that the Z39.50 server for Canada's national union catalogue (Voilà) has been corrupting every MARC record it transmits since at least October.1st response from OCLC was "You're not allowed to use that for copy cataloguing, you have to have a full Worldcat subscription!"🙄
       
 (DIR) Post #ATvbVuvO7j7Ai9KRPM by dbs@code4lib.social
       2023-03-24T02:03:41Z
       
       0 likes, 0 repeats
       
       1st response from LAC was "It works for us, see?" with a screenshot showing their connection. To Worldcat, not the Voilà Z39.50 server. 🙄After reassuring OCLC that I was trying to use it for the allowed purpose of searching for interlibrary loan items, they relaxed and started to investigate.3 days later, my OCLC contact said that their technical team suggested I try setting the server character encoding to Unicode.
       
 (DIR) Post #ATvbVvQEH3miFnl4To by dbs@code4lib.social
       2023-03-24T02:09:54Z
       
       0 likes, 0 repeats
       
       So they ignored the part where I pointed out that every record has an 8 char date in the 008 fixed field that only allows 6 chars, thus corrupting the rest of the 008. They also ignored the test case I gave them, using yaz-client to be nice and reproducible, that shows every non-ASCII character getting turned into a literal <U+00fd> 8-char string.
       
 (DIR) Post #ATvbVw4dqo6wH8fLTE by dbs@code4lib.social
       2023-03-24T02:12:41Z
       
       0 likes, 0 repeats
       
       And they ignored my suggestion that this corruption was likely being introduced by whatever middleware they use to strip the 001 and 035 field, and whatever else they're doing to prevent their precious OCLC numbers from falling into the hands of commoners.Nope. Apparently they're hoping that changing the server charset will somehow prevent the 008 from getting managed. And apparently it's up to me to test it, because they can't or won't?
       
 (DIR) Post #ATvbVwaXwBdDs5apCS by misty@digipres.club
       2023-03-24T03:26:49Z
       
       0 likes, 0 repeats
       
       @dbs Uuuugh.
       
 (DIR) Post #ATvbVx2uEkJhI2rTP6 by dbs@code4lib.social
       2023-03-24T03:30:47Z
       
       0 likes, 0 repeats
       
       Changing the marccharset does not change the overly long date in the 008.However, the binary MARC that I downloaded is MARC8. So that's why the characters don't display as expected.Here's why: Voilà's Z39.50 server runs on the FirstSearch host. Unlike the WorldCat Z39.50 server, which supports both UTF-8 and MARC8, FirstSearch is silent about encoding: https://help.oclc.org/Discovery_and_Reference/FirstSearch/Z3950_access/Z3950_configuration_for_FirstSearch_databasesSo in 2023, OCLC runs a Z39.50 server that is MARC8 only. Surprise!