This simplify all laters invocations, because we do not have to pass
the automaton the run was generated on.
This fixes#113 by allowing the __str__ function to be implemented on
runs.
* src/twaalgos/emptiness.cc, src/twaalgos/emptiness.hh (twa_run):
Store the automaton.
(prin_twa_run): Rewrite as an overloaded <<.
* src/twaalgos/reducerun.cc, src/twaalgos/reducerun.hh (reduce_run):
Do not like the automaton as a parameter.
* src/twaalgos/replayrun.cc, src/twaalgos/replayrun.hh (replay_twa_run):
Likewise.
* src/bin/common_aoutput.hh, src/bin/ltlcross.cc,
src/tests/complementation.cc, src/tests/ikwiad.cc,
src/tests/randtgba.cc, src/twaalgos/gtec/ce.cc, src/twaalgos/gv04.cc,
src/twaalgos/magic.cc, src/twaalgos/ndfs_result.hxx,
src/twaalgos/se05.cc, src/twaalgos/projrun.cc: Adjust.
* wrap/python/ajax/spotcgi.in: Add a __str__ function to twa_run_ptr.
* wrap/python/spot_impl.i: Adjust.
Doing so also work around some differences between Swig 3.0.2 and 3.0.7
observed on our build farm.
* wrap/python/spot.py: Here.
* wrap/python/spot_impl.i: Recognize None as a null formula
on input.
Fixes#114.
* src/parseaut/public.hh: Add support for a trust_hoa option.
* src/parseaut/parseaut.yy: If trust_hoa is set, recognize the
"inherently-weak" and "stutter-invariant" properties.
* src/bin/common_conv.cc, src/bin/common_conv.hh (read_automaton):
Move...
* src/bin/common_hoaread.cc, src/bin/common_hoaread.hh: ... in this
new file, that also handle the --trust-hoa option.
* src/bin/Makefile.am: Add them.
* src/bin/autfilt.cc, src/bin/dstar2tgba.cc, src/bin/ltlcross.cc,
src/bin/ltldo.cc: Use them.
* src/tests/parseaut.test, src/tests/ltldo.test: Adjust, and test
--trust-hoa=no.
* src/tests/complement.test, src/tests/prodor.test,
src/tests/sbacc.test: Adjust.
* wrap/python/spot.py (automata): Add option trust_hoa.
* NEWS: Update.
* src/parseaut/public.hh, src/parseaut/parseaut.yy: Make it easier to
pass new options to the parser.
* src/tests/ikwiad.cc, wrap/python/spot.py: Adjust.
This occured when parsing the HOA automaton generated by Rabinizer3 for
a very long LTL formula with many nested U. State labels could easily
be more than 40k characters.
* src/parseaut/scanaut.ll: Fix that.
* src/tests/parseaut.test: New test case.
* NEWS: Mention the fix.
* wrap/python/spot.py: Implement it. Also report non-zero
exit using CalledProcessError.
* wrap/python/tests/automata-io.ipynb, wrap/python/tests/piperead.ipynb:
Adjust.
* NEWS: Mention the change.
I have to remember that org-mode does not evaluate something unless it
is eventually exported.
* doc/org/hoa.org, doc/org/ioltl.org, doc/org/tut21.org,
doc/org/tut30.org: Adjust exports commands.
* src/tests/unambig.test: New test case. Reported by Ming-Hsien Tsai.
* src/twaalgos/sccfilter.cc: Always create an initial state.
* src/twaalgos/isunamb.cc: Speed up on empty languages.
* NEWS, THANKS: Update.
This simplifies the use of the spot.postprocessor object.
* wrap/python/spot.py: Add it.
* wrap/python/tests/automata.ipynb: Use it.
* NEWS: Mention it.
This is a bug:
% ltlfilt -f 'a W b' --unabbreviate=WR
a U (b | (a W b))
* src/tl/unabbrev.cc: Here.
* src/tests/unabbrevwm.test: Harden test case.
* wrap/python/tests/randltl.ipynb: Adjust expected output.
* NEWS: Mention the fix.
* wrap/python/spot_impl.i (get_name, set_name): New methods for twa.
* wrap/python/tests/remfin.py: Test them.
* wrap/python/ajax/spotcgi.in: Use set_name().
* NEWS: Mention it.
* wrap/python/spot.py: Add the unabbreviate() method, and
remove unabbrivate_ltl() and get_literal().
* wrap/python/tests/randltl.ipynb: Demonstrate unabbreviate().
This additionally fixes#107.
* src/bin/ltlgrind.cc: Fix handling for FILEANAME/COL. Document FORMAT
in --help. Assume -F for arguments given without options.
* src/tests/ltlgrind.test: Add two tests.
* NEWS: Mention this.
This is motivated by an email from Fanda.
* src/bin/common_post.cc, src/bin/common_post.hh: Add variables to
detect when level or pref are sets.
* src/bin/autfilt.cc: Adjust default for pref/sets.
* src/tests/readsave.test: Add test cases.
* NEWS: Mention it.
* src/twaalgos/randomgraph.cc: Replace an assertion by an exception.
* src/bin/randaut.cc: Diagnose -Q0.
* src/tests/randaut.test: Test it.
* NEWS: Mention the bug.
* src/twaalgos/sccfilter.hh,
src/twaalgos/sccfilter.cc (scc_filter_states): Remove useless acceptance
marks while preserving state-based acceptance. Add a new argument
to specify if all useless mark have to be removed, like for scc_filter.
* src/twaalgos/simulation.cc: Use the new parameter.
* src/twaalgos/postproc.cc: Likewise. Also call do_scc_filter even
after WDBA simplification to cleanup trivial SCCs. Preserve state-based
acceptance for weak automata.
* src/tests/readsave.test: Add one test.
* src/tests/dstar.test, src/tests/prodor.test, src/tests/remfin.test,
src/tests/sim3.test, wrap/python/tests/automata.ipynb,
wrap/python/tests/piperead.ipynb: Adjust expected output.
* NEWS: Mention the change.
The previous code was sometime doing the work of remove_fin() in
addition to complementing the acceptance conditions. This separate
the two operations clearly. Also the specialized code for
complementing weak automata is now a specialized code for remove_fin()
on weak automata.
* src/twaalgos/dtgbacomp.hh, src/twaalgos/dtgbacomp.cc: Rename as ...
* src/twaalgos/complement.hh, src/twaalgos/complement.cc: ... these.
* src/twaalgos/Makefile.am: Adjust.
* src/twaalgos/complement.hh (dtgba_complement): Rename as ...
(dtwa_complement): ... this, and restrict the purpose to completion
and accetance complementation. Further acceptance simplification
can be done with remove_fin() and to_generalized_buchi().
* src/twaalgos/remfin.cc (remove_fin): Specialize handling of weak
automata using the code that was originally in dtgba_complement().
Also mark the output as state-based when the input has to Inf.
* src/twaalgos/postproc.cc, src/twaalgos/postproc.hh: Make sure
scc_filter is always called after to_generalized_buchi().
* bench/stutter/stutter_invariance_randomgraph.cc,
src/bin/ltlcross.cc, src/tests/ikwiad.cc, src/twaalgos/minimize.cc,
src/twaalgos/powerset.cc, src/twaalgos/stutter.cc: Adjust usage.
* src/tests/dstar.test, src/tests/ltl2dstar4.test,
src/tests/remfin.test: Adjust expected outputs.
* wrap/python/spot_impl.i: Export dtwa_complement().
* wrap/python/spot.py: Python 3.5 reports some unexpected SystemError
messages when the stack of iterator(...(iterator(generator))) we build
for random LTL generation raises a StopIteration. The messages are
attached either to delete_formula or delete_randltlgenerator, claiming
that these functions exit with an error; but I have checked that they
do not. I've been unable to understand the cause of the issue.
Replacing the generator by an iterator at least fixes the problem in a
way that is transparent for our API.
* wrap/python/tests/randltl.ipynb: Adjust expected formulas.
* src/twaalgos/postproc.hh: Allow the formula not to be specified. The
code already support that, as it is called with an explicit nullptr in
autfilt (for example), but not requiring the nullptr is better for the
Python bindings.
* src/twaalgos/complete.cc, src/twaalgos/complete.hh
(tgba_complete, tgba_complete_here): Rename as...
(complete, complete_here): ... these. Also fix useless output of
acceptance marks on transition leading to the sink when the automaton
does not use state-based acceptance.
* src/tests/ikwiad.cc, src/twaalgos/dtgbacomp.cc,
src/twaalgos/dtgbasat.cc, src/twaalgos/postproc.cc,
src/twaalgos/product.cc: Adjust.
* wrap/python/spot_impl.i: Export these function.
* wrap/python/tests/automata.ipynb: Test spot.complete().
