Skip to content

v2.0.0

Latest
Compare
Choose a tag to compare
@LTLA LTLA released this 22 May 19:50
· 8 commits to master since this release
  • Updated to match the latest extractor interfaces in tatami.
  • Renamed the oracle-aware slab caches to OracularSlabCache and OracularSubsettedSlabCache.
  • Support caching of variable-size slabs via OracularVariableSlabCache.
  • Optimized oracle-aware and LRU caches for repeated requests from the same slab.
  • Cache classes can now report the maximum and current number of slabs.
  • Added the ChunkDimensionStats class to manage per-dimension chunk stats.
  • Replaced the TypicalSlabCacheWorkspace class with the simpler SlabCacheStats class.
  • Cap the computed number of cached slabs at the maximum number of possible slabs for a given matrix. This includes the situation where the slabs or data elements have zero size (e.g., for sparse extraction with no requested values or indices).
  • Added the DenseSlabFactory and SparseSlabFactory classes to create slabs from a contiguous buffer.
  • Created dedicated *Options classes for each of the custom chunked matrices.