Memory leak in cprocessors UnciodeResultProcessor

[Former user]

(original reporter: ged) It needs a dealloc method. See #1978. I don't have time to fix this right now, but should be able to get to it next week, if nobody beats me to it...

[Former user]

(original author: ged) "Fixed" in 20418096d5862ed40f22b6c7b7cc53dd212bbd21. Though, in practice, I couldn't find any sign of a leak. Having the explicit dealloc method is recommended so it's better to have it anyway even if it doesn't make any difference here.

[Former user]

(original author: taavi) Yeah, I couldn't get the DecimalResultProcessor to leak string parameters in the tests either, but when running our pylons app without providing that dealloc method memory was still leaking. Providing a dealloc method for DecimalResultProcessor stopped the leak in my full-system load tests.

But I didn't even look to provide one to the UnicodeResultProcessor, because it wasn't leaking. Of course, that might have something to do with the db driver (python-mysql) and the unicode settings we have.

I suspect the implementation difference between the {{{"%%.%df" % x}}} dynamically created string fed to DecimalStringProcessor and the probably-static string provided to UnicodeResultProcessor also leads to one leaking in practice while the other doesn't.

But yeah, consistency and following the recommended code structure's a good thing. :)

[Mike Bayer]

Removing milestone: 0.6.6 (automated comment)