Malthe Borch wrote:
Martijn Pieters wrote:
I object as well, and have asked for Malthe to provide his reasoning here at the Plone Performance Sprint in Bristol, but so far his only motivation is that he wants to see if he can get this to work without a C-extension. I am sceptical he'll be able to, and am not convinced it'll be worth introducing risks.
The obvious motivation for this is to:
* Reduce code complexity * Allow operation in a pure-Python environment
As for cons, any change is a risk and I believe the concensus seen in this thread is that it outweighs the above mentioned motivation.
Allowing operation in a pure-Python environment is a worthwhile goal, which I support. Unless it can be clearly demonstrated that the new method is equivalent in both performance and security, talk of dropping the C extension seems somewhat premature. A pure Python fallback for this module would however be interesting to everybody, I think. My suspicion from observing the discussions in this thread so far indicate that a drop in code complexity doesn't seem to be a necessary consequence of rewriting to Python either. Regarsd, Martijn