On Wed, Jul 14, 2004 at 09:59:40PM +0100, Tim Bunce wrote: > > > Make test, run against 8.0.5.2.1 created with ISO-8859-2 character > > set, is more interesting. The output is the same for both > > NLS_LANG=AMERICAN_AMERICA.UTF8 and NLS_LANG unset. I'm not sure > > how the 8bit characters get through but I assume the expected: > > values are those that are in the tests as expected values. > > > > Should I make some more tests? Would TEST_VERBOSE=1 help? > > Yes, please add "show_db_charsets( $dbh );" to line 22 of t/21nchar.t > then run "perl -Mblib t/21nchar.t" and send the output (just to > myself and "Lincoln A. Baxter" <[EMAIL PROTECTED]> to save the list)
Umm, I've just noticed that t/21nchar.t has code to skip all the tests if the client was built using Oracle 8. I'm not sure why that is, it may just be a hang over from early versions of the UTF8 support that only worked for Oracle 9. (Lincoln, can you remember.) Anyway, the fact the test ran implies that there's something wrong with the Oracle version detection. So please send me a log from the build. Thanks. Tim.
