Space optimization for genshi/output.py

[Carsten Klein <carsten.klein@…>]

I am currently reviewing genshi and looking for provable (and smaller) space and time optimizations.

In the output module I have found that on each call, a local cache will be build.

Comparing execution times, both with caches enabled and without caches, yielded the same results, approx. 140 milliseconds for rendering a rather complex (no loops, but multiple matches) template.

In-between, there were rendering times of approx. 400 milliseconds, but I reckon that this has to do with garbage collection taking place, and reinstantiation of objects.

This request basically boils down to this:

• remove local caching of input/output fragment in the serializers as they will consume space with larger documents
• remove the cache parameter from the api, since it does not yield better results other than consuming more memory

What are your opinions on this?

[Carsten Klein <carsten.klein@…>]

Darn me, it was the wrong installation I was working on. Local caching actually yields approx 10..50 milliseconds less time processing the template.

I will invalidate this ticket for now.