As discussed with Jan Strejček.
* spot/twa/acc.cc, spot/twa/acc.hh (fin_unit_one_split): New function.
(fin_one_extract): Return the simplified acceptance condition as an
optimization.
* python/spot/impl.i: Bind this new function.
* tests/python/acc.py: New file, to test it.
* tests/Makefile.am: Add acc.py.
* spot/twaalgos/genem.cc, spot/twaalgos/genem.hh: Implement the
spot211 variant of the emptiness check.
* tests/python/genem.py: Test it.
* tests/python/acc_cond.ipynb: Adjust test for fin_one_extract.
With a formula like G(b1) & (GFi <-> GF(b1)), a direct strategy was
created while it is unrealizable.
* spot/twaalgos/synthesis.cc: here.
* tests/core/ltlsynt.test: add tests
merge_states is now hash-based, uses the new edge-sorting with
src first and can be executed in parallel.
* spot/twa/twagraph.cc: Here
* tests/python/mergedge.py: Test
sort_edge_srcfirst_ will sort the edge with respect
to the src state, then sort each sub list with respect
to the given predicate, possibly in parallel.
* spot/graph/graph.hh: Here
This largely speeds up the computation for conditions
like "Rabin n" sharing a lot of subtrees.
Also implement options to stop the construction if the shape is wrong.
* spot/twaalgos/zlktree.cc, spot/twaalgos/zlktree.hh: Implement the
cache and the options.
* tests/python/zlktree.ipynb, tests/python/zlktree.py: New tests.
* spot/twaalgos/complete.hh: Adjust documentation.
* spot/twaalgos/complete.cc: If the acceptance condition is a
tautology, delay the forcing of Büchi acceptance until we are sure it
is needed.
* NEWS: Mention the change.
The issue seems to be inside std::vector's copy constructor, but it
highlighted places in Spot were we could avoid this copy.
* spot/twaalgos/ltl2taa.cc: Avoid some copies of std::vector<formula>.
Fix#505, Reported by Edmond Irani Liu.
* spot/twa/twagraph.cc (copy): Deal with kripke_graph in a better way.
* spot/twaalgos/hoa.cc: Do not force the use of named-states since
when the input is a kripke_graph.
* tests/python/kripke.py: Adjust test cases.
* NEWS: Mention the change.
* THANKS: Add Edmund.
The issue seems to be inside std::vector's copy constructor, but it
highlighted places in Spot were we could avoid this copy.
* spot/twaalgos/ltl2taa.cc: Avoid some copies of std::vector<formula>.
This will later help improve the handling of different input types of
ltlsynt.
* bin/common_finput.hh (job_type): New enum.
(job::type): Use it.
* bin/autcross.cc, bin/autfilt.cc, bin/common_finput.cc,
bin/dstar2tgba.cc, bin/ltl2tgba.cc, bin/ltl2tgta.cc, bin/ltlcross.cc,
bin/ltldo.cc, bin/ltlfilt.cc, bin/ltlgrind.cc, bin/ltlsynt.cc: Adjust
to use the job_type enum instead of a boolean.
Fix#505, Reported by Edmond Irani Liu.
* spot/twa/twagraph.cc (copy): Deal with kripke_graph in a better way.
* spot/twaalgos/hoa.cc: Do not force the use of named-states since
when the input is a kripke_graph.
* tests/python/kripke.py: Adjust test cases.
* NEWS: Mention the change.
* THANKS: Add Edmund.
Optimization in Zielonka failed
under certain circumstances
todo: Devise a specialized test
for direct attr computation
* spot/twaalgos/game.cc: Correction
* tests/python/game.py: Test
Reported by Florian Renkin.
* spot/graph/graph.hh (sort_edges_of): Fix invalid read when sorting a
state without successor. Seen on core/tgbagraph.test.
* python/spot/impl.i (formula_from_bdd): Instantiate for twa_graph.
* spot/twa/twa.hh (register_aps_from_dict): Typo in exception.
* tests/python/except.py: More tests for the above.
* tests/python/bdddict.py: Typo in comment.
