zodb / relstorage / 391

74%

master: 92%

push

travis-ci

Initial pass at two-generation cache.

Still need to probably segment the main portion into protected/probation
to get the right behaviour.

The code is unoptimized but all tests pass (with minor modifications to
account for the frequency-based preservation, which is actually kind of
cool---although we will ultimately need more specific tests).

The mixed micro-benchmark is actually faster. Populating is notably
slower. Reading is basically unchanged (slightly slower).

Start using the frequency information to avoid writing out objects that
haven't been accessed in a long time. We should probably run an aging pass
over these objects before we write to make it more accurate.

103 of 103 new or added lines in 1 file covered. (100.0%)

4757 of 4979 relevant lines covered (95.54%)

7.74 hits per line