Fix 'orafce' to work on GPDB.
The previous commit updated 'orafce' module to new upstream version. This commit fixes it so that it works with GPDB. We carried many of these diffs against the old version too, but some are new with the new orafce version. Old differences, put back in this commit: - In the 'finish no data found' NOTICEs, GPDB strips trailing spaces - GPDB doesn't allow SQL_ASCII as database encoding. Use 'iso-8859-1' in the 'nlssort' test instead. - Don't install dbms_alert. It doesn't work in GPDB. - GPDB doesn't have 'tts_tableOid' field in TupleTableSlot New differences: - Use init_file to suppress extra NOTICEs about distribution keys and such that GPDB prints. (Previously, we had changed the expected outputs to include the NOTICEs, but this seems nicer.) - GPDB has built-in DECODE() support, so don't install the orafce decode() compatibility functions. The DECODE() is transformed to a CASE-WHEN construct, so fix all the regression tests that had 'decode' as the result column name now into 'case' instead. A few tests are now erroring out because of missing casts between text and other datatypes, but that's expected / accepted because the GPDB implementation of DECODE() is different.
Showing
gpcontrib/orafce/init_file
0 → 100644
想要评论请 注册 或 登录