This release is a bugfix release. Major changes in fplll-5.2.1 compared to fplll-5.2.0:
- Possibly resolved the Gram bug by Koen de Boer
- Fix Solaris issues by Jeroen Demeyer
- Unify randomness on 32-bit and 64-bit systems by Martin Albrecht
- Adding Windows 10 instructions to the readme by Thijs Laarhoven
- Can use mpir instead of gmp by Laurent Grémy
- Dot product by Laurent Grémy
- make MatGSOGram.row_swap public by Martin Albrecht
- dd wrapper by Gilles Villard
Major changes in fplll-5.2.0 compared to fplll-5.1.0:
- make distcheck for test_bkz by Laurent Grémy
- Bkz dump_gso in json by Laurent Grémy
- Improve pruner by Léo Ducas
- BKZ with Z_NR<long> by Michael Walter
- Z_NR<long> support by Martin Albrecht
- Allow BOUNDED_LLL in Slide reduction and remove LLL from postprocessing by Michael Walter
- Add Gram-matrix object by Koen de Boer
- Allow the recursive enumerate feature to be disabled at configure time by Erik Bray
Major changes in fplll-5.1.0 compared to fplll-5.0.3:
-
CVP is not experimental any longer by implementing the resetting feature (Guillaume Bonnoron #191)
-
pruner rewrite providing faster pruning, a new greedy pruning strategy, support for multiple solutions and a simpler API (Léo Ducas #228)
-
support for external enumeration libraries paving the way for multithreaded/MPI/GPU enumeration libraries. See https://github.com/cr-marcstevens/fplll-extenum for an example external library (Marc Stevens #227)
-
Gaussian Heuristic improved and renamed to ‘adjust_radius_to_gh_bound‘ (Léo Ducas #231)
-
Reduced calls to LLL during BKZ (Martin Albrecht #226, Damien Stehlé and Shi Bai #230)
-
Greatly improved OSX support (Martin Albrecht and Isuru Fernando #205)
-
Evaluator rewrite/refactoring now supporting different evaluation strategies (Marc Stevens #224)
-
We now enforce coding conventions (Tancrède Lepoint #215, #219)
making 5.0.3 for distributions
Assets
4
lets call 5.0.3rc0 5.0.3rc0
Merge pull request #197 from malb/precision run self-dual BKZ in at least dd precision
version bump to 5.0.1
Merge pull request #184 from malb/blocksize-90 strategies up to (incl) block size 90
make distcheck passes