Open fire-eggs opened 6 years ago
Preliminary testing suggests the benefit of caching (e.g.) surnames is outweighed by the cost.
There is a memory overhead from the cache proper, especially from the default Dictionary expansion allocation. There is the performance cost of the cache lookup, aggravated by parallelism support.
I'm going to leave this open for now, pending further measurement and experiments.
dotMemory tracing found repetition of surnames. E.g. for (TomsTreePriv.ged?) the surname "Skillman" was repeated 12,996 times
Possible candidate for lookup table? See issue #38