]> git.ipfire.org Git - thirdparty/sqlalchemy/sqlalchemy.git/commit
- the behavior of String/Unicode types regarding that they auto-convert
authorMike Bayer <mike_mp@zzzcomputing.com>
Wed, 26 Sep 2007 14:55:44 +0000 (14:55 +0000)
committerMike Bayer <mike_mp@zzzcomputing.com>
Wed, 26 Sep 2007 14:55:44 +0000 (14:55 +0000)
commite37a3a961c9f5bf822f86682c0f96418281d3e42
treea47128eca7327fd6e4ce63bc655775eddd7cc846
parenta0838e0c476f7deab5b67f431c8ce107974b9313
- the behavior of String/Unicode types regarding that they auto-convert
  to TEXT/CLOB when no length is present now occurs *only* for an exact type
  of String or Unicode with no arguments.  If you use VARCHAR or NCHAR
  (subclasses of String/Unicode) with no length, they will be interpreted
  by the dialect as VARCHAR/NCHAR; no "magic" conversion happens there.
  This is less surprising behavior and in particular this helps Oracle keep
  string-based bind parameters as VARCHARs and not CLOBs [ticket:793].
CHANGES
lib/sqlalchemy/sql/expression.py
lib/sqlalchemy/types.py
test/dialect/oracle.py
test/sql/testtypes.py