This actually performs three related changes, but separating them
would be quite inconvenient.
1) rename tostring.hh to print.hh a welcome side-effect is that
I could fix several files that included this file for not reason.
2) de-overload some of the to_string functions, and rename them
as follow:
to_string -> print_psl, print_sere, str_psl, str_sere
to_utf8_string -> print_utf8_psl, print_utf8_sere,
str_utf8_psl, str_utf8_sere
to_spin_string -> print_spin_ltl, str_spin_ltl
to_wring_string -> print_wring_ltl, str_wing_ltl
to_lbt_string -> print_lbt_ltl, str_lbt_ltl
to_latex_string -> print_latex_psl, str_latex_psl
to_sclatex_string -> print_sclatex_psl, str_sclatex_psl
Now it is clearer what these functions do, and their restrictions.
3) all those print_* functions now take the stream to write onto
as their first argument. This fixes#88.
* src/ltlvisit/tostring.cc, src/ltlvisit/tostring.hh: Rename into...
* src/ltlvisit/print.cc, src/ltlvisit/print.hh: ... those, and make
the changes listed above.
* doc/org/tut01.org, src/bin/common_output.cc,
src/bin/common_trans.cc, src/bin/ltl2tgba.cc, src/bin/ltl2tgta.cc,
src/bin/ltlcross.cc, src/bin/ltldo.cc, src/bin/ltlfilt.cc,
src/bin/randltl.cc, src/ltlparse/ltlparse.yy,
src/ltlvisit/Makefile.am, src/ltlvisit/mark.cc,
src/ltlvisit/relabel.cc, src/ltlvisit/simplify.cc,
src/ltlvisit/snf.cc, src/ta/taexplicit.cc, src/ta/tgtaexplicit.cc,
src/taalgos/tgba2ta.cc, src/tests/equalsf.cc, src/tests/ltl2tgba.cc,
src/tests/ltlrel.cc, src/tests/randtgba.cc, src/tests/reduc.cc,
src/tests/syntimpl.cc, src/tests/tostring.cc, src/twa/bdddict.cc,
src/twa/bddprint.cc, src/twa/taatgba.cc, src/twa/taatgba.hh,
src/twa/twagraph.cc, src/twaalgos/compsusp.cc, src/twaalgos/lbtt.cc,
src/twaalgos/ltl2taa.cc, src/twaalgos/ltl2tgba_fm.cc,
src/twaalgos/neverclaim.cc, src/twaalgos/remprop.cc,
src/twaalgos/stats.cc, wrap/python/ajax/spot.in, wrap/python/spot.py,
wrap/python/spot_impl.i: Adjust.
The difficulty is not the example, but setting up org-mode to allow
Python and C++ example that use the local libraries, not those
installed system-wide.
* doc/org/.dir-locals.el: Rename as...
* doc/org/.dir-locals.el.in: ... this, so we can easily define
PYTHONPATH and other environment variables.
* doc/org/init.el.in: Enable C++, and make sure but Python
and C++ use the local libraries.
* doc/org/g++wrap.in, doc/org/tut01.org: New files.
* doc/Makefile.am, configure.ac: Adjust.
* wrap/python/spot.py (to_str): Take a parenth argument.
* src/bin/common_trans.cc: Skip leading directories, and
do not require that the prefix is followed by space of 0.
* src/tests/ltldo.test: Add a test.
* doc/org/ltlcross.org: Update.
This is particularly important for src/tests/satmin.test, where ltl2tgba
might be killed while writing a huge temporary file used for SAT-based
minimization. Before this patch, the temporary files would remain in
src/tests/satmin.dir/, easily overflowing the 100GB limit of the docker
containers we use on the build farm.
* src/bin/common_setup.cc: Catch termination signals for all tools,
even those that do not yet clear temporary files.
* configure.ac: Check for sigaction.
This way -S means --state-based-acc like with other tools
producing automata. This fixes#82.
* src/bin/randaut.cc: Rename -S as -Q, rename --state-acc as
--state-based-acc (with --sbacc as a synonym), and declare -S as the
short version of --state-based-acc.
* doc/org/autfilt.org, doc/org/oaut.org, doc/org/randaut.org,
src/tests/isomorph.test, src/tests/randaut.test,
src/tests/randtgba.test, src/tests/readsave.test, src/tests/uniq.test,
wrap/python/tests/randaut.ipynb: Adjust all calls to randaut.
gcc version 6.0.0 20150516 (experimental) [trunk revision 223239]
(Debian 20150516-1)
* src/hoaparse/hoaparse.yy (state_info): Initialize used and declared,
otherwise they are uninitialized after a vector resize.
* src/bin/randaut.cc: Replace the --acc-type and --acc-sets
options by a more general --acceptance option, that take either
an acceptance formula, or an acceptance name parametred by ranges.
Also accept a range for the number of atomic propositions.
* src/twaalgos/randomgraph.cc (random_acceptance): Move...
* src/twa/acc.cc, src/twa/acc.hh (random): ... here.
(parse_acc_code): Generalize to accept ranges instead of
numbers whenever sensible, and accept a 'random' acceptance.
* src/tests/randaut.test: Adjust tests and add more.
* wrap/python/tests/randaut.ipynb: Adjust call to randaut.
According to https://github.com/adl/hoaf/issues/46
* src/twa/acc.cc (parity): Adjust generation.
* src/tests/hoaparse.test, wrap/python/tests/accparse.ipynb:
Adjust existing test cases.
* wrap/python/tests/accparse2.py: New test cases.
* doc/org/satmin.org: Document the new DTωA-minimization procedure.
* doc/org/tools.org: Fix link.
* src/bin/autfilt.cc: Pass -S to sat_minimize().
* src/twa/twagraph.hh: (state_acc_sets) New method.
* src/twaalgos/dotty.cc: Use it to correctly display co-Büchi automata.
* src/twaalgos/dtbasat.cc: Set the deterministic property on the result.
* src/twaalgos/dtgbasat.cc: Likewise, and preprocess the input automaton
in sat_minimize().
* src/twaalgos/dtgbasat.hh: Fix documentation, and take the state-based
information as an argument.
* src/twaalgos/postproc.cc: Do not call simulation-based reduction
on non-separated acceptances.
* src/tests/satmin2.test: Use -S rather than 'state-based'.
* NEWS: Update.
* src/twa/acc.cc: Allow duplicate and reordered pairs. Also recognize
the single-pair cases.
* src/twaalgos/hoa.cc: When Rabin or Streett is detected, canonicalize
the Acceptance: line.
* src/tests/hoaparse.test, wrap/python/tests/accparse2.py: More tests.
* src/tests/sbacc.test: Adjust.
It has two modes: strict or not. In strict mode (tested in
hoaparse.test), the acceptance formula has to match exactly the one
given in the HOA spec. In non-strict mode (tested in accparse2.py)
any equivalent formula is accepted.
* src/twa/acc.cc, src/twa/acc.hh (acc_cond::is_parity): New method.
* src/twaalgos/hoa.cc: Use it.
* src/tests/hoaparse.test: Test it.
* wrap/python/spot_impl.i: Bind it.
* wrap/python/tests/accparse2.py: New file.
* wrap/python/tests/Makefile.am: Add it.
So far the HOA output would emit an acc-name only
for generalized-Buchi or inferior types (Buchi, all).
It now knows about none, co-Buchi, generalized-co-Buchi,
Rabin, Streett, and generalized-Rabin as well.
* src/twa/acc.cc, src/twa/acc.hh: Add detection code.
* src/twaalgos/hoa.cc: Use it.
* src/tests/remfin.test, src/tests/maskacc.test,
src/tests/complete.test, src/tests/sim3.test,
src/tests/ltl2dstar.test: Adjust tests.
* src/tests/hoaparse.test: Adjust and add more tests.
* src/twa/acc.cc, src/twa/acc.hh: Add method to create
standard acceptance conditions, and adjust the parse_acc_code
to recognize them
* wrap/python/spot_impl.i (acc_cond::acc_code): Add a printer.
* wrap/python/tests/accparse.ipynb: New test file.
* wrap/python/tests/Makefile.am: Add it.
* src/tests/satmin2.test: Use the new syntax.
If the target acceptance is Fin(0)&Inf(1), there is no need to
distinguish between an history of {0,1} and an history of {0}, as a
cycle with either history will be rejected. This implements this
simplification. If both the canditate and reference automata
are Rabin automata with n pairs, then we now use at most
Q * Q' * Q * Q' * 3^n * 3^n
variables to encode the partial cycles histories, versus
Q * Q' * Q * Q' * 4^n * 4^n
before.
* src/twaalgos/dtgbasat.cc: Implement this.
* src/tests/satmin2.test: More tests.
* src/twaalgos/dtgbasat.cc (sat_minimize): Use any arbitrary
acceptance condition passed via the "acc" option.
* src/twa/acc.hh (mark_t::max_set): New method.
* src/tests/satmin2.test: Add two test cases with DRA as input
and DSA as output.
When the reference acceptance condition is complex enough, some
accepting SCCs may not use all acceptance sets. In that case
we don't have to encode all possible histories for this SCC.
* src/twaalgos/dtgbasat.cc: Improve the encoding by omitting
histories involving sets that are not used in a reference SCC.
This is still missing tests.
* src/bin/autfilt.cc: Add a --sat-minimize option.
* src/misc/optionmap.cc, src/misc/optionmap.hh: Allow passing strings.
* src/twa/acc.cc, src/twa/acc.hh: Add helper functions needed
by the SAT-encoder.
* src/twaalgos/complete.hh: Typos.
* src/twaalgos/dtbasat.hh: Adjust comment.
* src/twaalgos/dtgbasat.cc, src/twaalgos/dtgbasat.hh: Generalize
to take the target acceptance as input.
* src/twaalgos/postproc.cc, src/tests/ltl2tgba.cc: Adjust calls.
* src/twaalgos/sbacc.cc, src/twaalgos/sbacc.hh: Don't pass
the pointer by reference.
* src/tests/acc.cc, src/tests/acc.test: More tests
for the acceptance helper function.
and make it easier to extend and use.
* src/twaalgos/postproc.hh, src/twaalgos/translate.cc,
src/twaalgos/translate.hh: Incorporate the Unambiguous option
with the other preferences. It's cleaner this way, and the
previous setup did not work well with Python.
* src/bin/ltl2tgba.cc: Adjust to this change.
* wrap/python/spot.py (translate): Rewrite.
* wrap/python/tests/automata.ipynb: Adjust existing cases, and
add more as well as some comments.
Producing state-based acceptance is now part of the postprocessing
routines. That means we can more easily simplify automata with
state-based acceptance (using autfilt -S --small --high, for instance)
and as as side-effect, ltl2tgba can produce GBA. However the result of
ltl2tgba -S is often larger than that of ltl2tgba -B.
* src/twaalgos/postproc.cc, src/twaalgos/postproc.hh: Implement
the SBAcc option.
* src/bin/common_post.cc, src/bin/common_post.hh: Implement -S.
* src/bin/autfilt.cc, src/bin/dstar2tgba.cc, src/bin/ltl2tgba.cc,
src/bin/ltl2tgta.cc, src/bin/ltldo.cc: Adjust.
* src/tests/sim3.test: Augment test case.
* NEWS, doc/org/ltl2tgba.org, doc/org/autfilt.org: Document it -S.
* src/twaalgos/simulation.cc, src/twaalgos/simulation.hh: Adjust
to work on TωA. This only require separate acceptance sets.
* src/tests/sim3.test: New test.
* src/tests/Makefile.am: Add it.
