]> git.ipfire.org Git - thirdparty/sqlalchemy/sqlalchemy.git/commit
Don't double-process ResultMetaData for BufferedColumnResultProxy
authorMike Bayer <mike_mp@zzzcomputing.com>
Wed, 27 Apr 2016 16:37:58 +0000 (11:37 -0500)
committerMike Bayer <mike_mp@zzzcomputing.com>
Wed, 27 Apr 2016 16:37:58 +0000 (11:37 -0500)
commitf3bc60bdd809235cbeb3f414717ac0e273269cf9
treeeec41353d94d79aae07ebae6c42fe0df66526702
parent84ba09a7d7b0160d8e3e52269519d0df0cd175ac
Don't double-process ResultMetaData for BufferedColumnResultProxy

Fixed a bug in the result proxy used mainly by Oracle when binary and
other LOB types are in play, such that when query / statement caching
were used, the type-level result processors, notably that required by
the binary type itself but also any other processor, would become lost
after the first run of the statement due to it being removed from the
cached result metadata.

Change-Id: I751940866cffb4f48de46edc8137482eab59790c
Fixes: #3699
doc/build/changelog/changelog_10.rst
lib/sqlalchemy/engine/result.py
test/sql/test_resultset.py