Sdsl Lite Versions Save

This feature contains

• Fixes for some minor issues
• demangle detection
• compilation with gcc4.7.3
• Improved functionality
  • sd_vector can now be parametrized with other bit_vectors
• Small performance improvements
  • WT construction is more space-efficient in some cases

Most important changes to previous release: [x] sd_vector can now be constructed from an increasing sequence of integers [x] Port to Windows (compiles with the latest version of VC++) [x] Tests use CMake [x] Added a Full-Compressed Suffix Tree (cst_fully) [x] Added an alphabet partitioned WT (wt_ap.hpp) [x] Added the hybrid_bitvector (hyp_vector.hpp) of Kärkkäinen, Kempa, and Puglisi [x] External dependencies libdivsufsort and gtest are now included as submodules

This version

• adds a K^2-treap implementation (k2_treap)
• fixes a bug in dac_vector
• adds rank_0/select_0 to sd_vector
• adds expand method for wavelet trees (WT algorithms)
• adds the sequence representation of Golynski, Munro and Rao (wt_gmr)
• includes a preliminary memory mapped version of int_vector (int_vector_mapper)
• cst_sct3 can now be parametrized to support fast navigational operations also on very large alphabets

This version

• extends the functionality of bit_vectors: get_int extracts up to 64 continues bits
• adds a wavelet tree benchmark
• adds a LCP construction benchmark
• includes an Elias gamma code
• fixes a bug in the hugepage memory management
• adds an copy constructor to cst_sct3::node_type (it should be automatic generated by the compiler, but we had strange memory leaking problems in one application and adding it explicitly solved the problem)

This version includes

• a bug fix in the load method of select_support_scan
• a benchmark for Wavelet Trees
• a benchmark for Document Retrieval Solutions

This version

• simplifies the usage of the library
• features integer-alphabet version of all indexes
• contains a unified implementation of pointer based Wavelet Trees
• introduces a Hu-Tucker-shape strategy for WTs
• facilitates the user to switch between semi-external and in-memory construction
• enables the use of 1 GB hugepages (example)
• provides more sophisticated visualizations (construction resource graph and memory breakdowns)
• includes a tutorial slides set and a new cheatsheet
• comes with a set of fully automated benchmarks
• includes a comprehensive test suite

This version

• simplifies the usage of the library
• features integer-alphabet version of all indexes
• contains a unified implementation of pointer based Wavelet Trees
• introduces a Hu-Tucker-shape strategy for WTs
• facilitates the user to switch between semi-external and in-memory construction
• supports easy use 1 GB hugepages (example)
• provides more sophisticated visualizations (construction resource graph and memory breakdowns)
• includes a tutorial slides set and a new cheatsheet
• comes with a set of fully automated benchmarks
• includes a comprehensive test suite

This release is the last version which uses the class int_vector_file_buffer for buffered reading of int_vector data from a stream. The newer version will replace this class with int_vector_buffer, which allows reading and writing. Furthermore it also handles the loading and writing of blocks automatically and is therefore easier to use.