lameiro / cx_oracle_on_ctypes

cx_Oracle on ctypes
22 stars 11 forks source link

cx_oracle_on_ctypes

What

cx_Oracle re-written in Python, using ctypes to interface with Oracle native code. 100% API-compatible with cx_Oracle.

Why?!

cx_Oracle is great code, but as many CPython extensions, it is not compatible with other Python implementations. I rewrote it in Python to be able to use PyPy and possibly Jython and IronPython. Also, ctypes is the recommended way to integrate C code with PyPy, as it can benefit from the JIT and no CPython emulation must be done.

How good is it?

Most of the Oracle datatypes are implemented and most of tests from cx_Oracle pass. Django works. Tested with Oracle 10 and 11. The performance is not as good as cx_Oracle yet, but it improves fast as PyPy's JIT improves.

Installation

As in cx_Oracle, you need the binary library from Oracle ( the instant client will suffice ) and you need to setup the environment variables so that Python can find it. As an example, take a look into https://github.com/lameiro/cx_oracle_on_ctypes/blob/master/test/setup_test_env_10.sh and adapt accordingly.

To install the python module, just put the cx_Oracle dir in your PYTHONPATH.

License

3-clause BSD, like cx_Oracle. See http://opensource.org/licenses/BSD-3-Clause

Differences

cx_oracle_on_ctypes is 100% API compatible with cx_Oracle. If you find any missing functionality, even if it is not in the DB-API specification, report a bug.

Contributing

Feedback and patches are welcome. Send them through the issue tracker or by e-mail.