Spin 6 supports formulas such as []<>(a < b) so that atomic properties
need not be specified using #define. Of course we don't want to
implement all the syntax of Spin in our LTL parser because other tools
may have different syntaxes for their atomic propositions. The
lenient mode tells the scanner to return any (...), {...}, or {...}!
block as a single token. The parser will try to recursively parse
this block as a LTL/SERE formula, and if this fails, it will consider
the block to be an atomic proposition. The drawback is that most
syntax errors will no be considered to be atomic propositions. For
instance (a U b U) is a single atomic proposition in lenient mode, and
a syntax error in default mode.
* src/ltlparse/ltlparse.yy, src/ltlparse/ltlscan.ll,
src/ltlparse/parsedecl.hh, src/ltlparse/public.hh: Add a
lenient parsing mode. Simplify the lexer using yy_scan_string.
* src/bin/common_finput.cc: Add a --lenient option.
* src/ltltest/lenient.test: New file.
* src/ltltest/Makefile.am: Add it.
* src/neverparse/neverclaimparse.yy: Parse the guards in lenient mode.
* src/tgbatest/neverclaimread.test: Adjust.
* src/ltlvisit/tostring.cc: When outputing a formula in Spin's syntax,
output (a < b) instead of "a < b".
* src/misc/escape.cc, src/misc/escape.hh (trim): New helper function.
This somehow revert changes from 2010-01-30 which killed this use of
star to make room for the Kleen star. Here we only allow '*' in the
temporal formula, so that it can still be the Kleen star in SERE. The
motivation for '*' available as And is better compatibility with Wring
and VIS.
* src/ltlparse/ltlscan.ll: Distinguish * from [*].
* src/ltlparse/ltlparse.yy: Allows * to be used as AND between
temporal formulae.
* src/ltltest/equals.test, src/ltltest/parse.test: Add a few
tests.
* doc/tl/tl.tex: Document it.
* src/ltlparse/ltlparse.yy: Keep the left operand of binary operator,
if the right one is erroneous. Also keep the sane beginning of
parenthesized blocks.
* src/ltltest/parseerr.test: Adjust test cases.
* NEWS: Mention it.
* src/ltlparse/ltlparse.yy: Make all the above operators
right-associative. Also let `:' have precedence over `;'.
* src/ltltest/reduccmp.test: Adjust for the `:' precedence.
* doc/tl/tl.tex, NEWS: Document this.
It was a mistake to try to overload And/Or LTL operator for these when
trivial simplification are performed. The reason is so simple it is
embarassing: And(f,1)=f is a trivial identity that should not be
applied with AndRat. E.g. AndRat(a;b, 1) is equal to 0, not a;b.
* src/ltlast/multop.hh, src/ltlast/multop.cc: Add the AndRat and OrRat
operators.
* src/ltlparse/ltlparse.yy: Build them.
* src/ltlvisit/mark.cc, src/ltlvisit/simplify.cc,
src/ltlvisit/tostring.cc, src/tgba/formula2bdd.cc,
src/tgbaalgos/eltl2tgba_lacim.cc, src/tgbaalgos/ltl2taa.cc,
src/tgbaalgos/ltl2tgba_fm.cc, src/tgbaalgos/ltl2tgba_lacim.cc:
Adjust all switches.
* src/ltlparse/ltlparse.yy: Cleanup the names used in the grammar.
* src/ltlparse/public.hh (parse_ratexp): Rename as...
(parse_sere): ... this.
* src/ltltest/consterm.cc: Adjust call to parse_ratexp().
* src/ltlparse/ltlparse.yy (rationalexp): Allow ->, <->, and xor,
in rational expressions as long as they apply only to Boolean
formulae.
* src/tgbaalgos/ltl2tgba_fm.cc (ratexp_trad_visitor): Adjust
assert in handling of unop::Not.
This bitfield is easily updated as the formulae are constructed.
Doing so avoids many AST recursions to compute these properties
individually. This patch removes the eventual_universal_visitor,
as well as the kind_of() function.
* src/ltlast/formula.hh (is_boolean, is_sugar_free_boolean,
is_in_nenoform, is_X_free, is_sugar_free_ltl,
is_ltl_formula, is_eltl_formula, is_psl_formula, is_eventual,
is_universal, is_marked): New methods to query formula
properties in constant time.
(get_props, ltl_prop): A method and structure for
implementation as a field bit in an unsigned, for fast
computation.
(print_formula_props): New function.
* src/ltlast/formula.cc (print_formula_props): Implement it.
* src/ltlast/atomic_prop.cc, src/ltlast/binop.cc,
src/ltlast/bunop.cc, src/ltlast/constant.cc, src/ltlast/multop.cc,
src/ltlast/unop.cc, src/ltlast/automatop.cc: Compute the
properties as instances are constructed.
* src/ltlparse/ltlparse.yy: Update to use is_boolean() instead
of kind_of().
* src/ltltest/kind.cc: Update to use print_formula_props().
* src/ltltest/kind.test: Adjust to test eventual and universal
properties.
* src/ltlvisit/kind.cc, src/ltlvisit/kind.hh: Delete these files.
* src/ltlvisit/Makefile.am: Remove kind.hh and kind.cc.
* src/ltlvisit/reduce.cc (recurse_eu, eventual_universal_visitor):
Remove, no longer needed.
(reduce_visitor, is_eventual, is_universal): Adjust to
use formula::is_eventual(), and formula::is_universal().
* src/ltlvisit/reduce.hh (is_eventual, is_universal): Declare as
deprecated.
* src/ltlparse/ltlscan.ll (OP_SQBKT_SEP): Accept ":" and "to"
in addition to ".." and ",".
(OP_UNBOUNDED): Recognize "$" for the rule below.
* src/ltlparse/ltlparse.yy: Accept [OP1:$] as a synonym
for [OP1:], for people used to SVA's syntax.
* src/ltltest/equals.test: Test these syntaxes.
* src/ltlast/bunop.hh: Declare bunop::Goto
* src/ltlast/bunop.cc: Handle it.
* src/ltlparse/ltlparse.yy,
src/ltlparse/ltlscan.ll: Add rules for [->min..max].
* src/tgbaalgos/ltl2tgba_fm.cc: Handle bunop::Goto in
the translation.
* src/ltltest/equals.test: Test trivial identities.
* src/tgbatest/ltl2tgba.test: Test two more formulae using [->].
* src/ltlast/bunop.hh: Declare bunop::Equal
* src/ltlast/bunop.cc: Handle it.
* src/ltlparse/ltlparse.yy,
src/ltlparse/ltlscan.ll: Add rules for [=min..max].
* src/tgbaalgos/ltl2tgba_fm.cc: Handle bunop::Equal in
the translation.
* src/ltltest/equals.test: Test trivial identities
for [=min..max].
* src/tgbatest/ltl2tgba.test: Add new formulae to test.
* src/ltlast/bunop.hh, src/ltlast/bunop.cc: New files for
bounded unary operators.
* src/ltlast/Makefile.am, src/ltlast/allnodes.hh: Add them.
* src/ltlast/predecl.hh (bunop): Declare.
* src/ltlast/unop.hh, src/ltlast/unop.cc (Star): Remove
declaration of Star and associated code.
* src/ltlast/visitor.hh: Add visit(bunop* node) methods.
* src/ltlparse/ltlparse.yy, src/ltlparse/ltlscan.ll: Add parse
rules for LTL. This required passing the parse_error list
to the lexer, so it can report scanning errors when it reads
a number that does not fit in an unsigned int.
* src/ltlparse/parsedecl.hh (YY_DECL): Take error_list
as third argument.
* src/ltltest/consterm.test, src/ltltest/tostring.test,
src/ltltest/equals.test, src/tgbatest/ltl2tgba.test: More tests.
* src/ltlvisit/basicreduce.cc, src/ltlvisit/clone.cc,
src/ltlvisit/clone.hh, src/ltlvisit/consterm.cc,
src/ltlvisit/dotty.cc, src/ltlvisit/mark.cc,
src/ltlvisit/nenoform.cc, src/ltlvisit/postfix.cc,
src/ltlvisit/postfix.hh, src/ltlvisit/reduce.cc,
src/ltlvisit/syntimpl.cc, src/ltlvisit/tostring.cc,
src/ltlvisit/tunabbrev.cc, src/tgba/formula2bdd.cc,
src/tgbaalgos/eltl2tgba_lacim.cc, src/tgbaalgos/ltl2taa.cc,
src/tgbaalgos/ltl2tgba_lacim.cc: Adjust syntax to use
"bunop::Star" instead of "unop::Star".
* src/tgbaalgos/ltl2tgba_fm.cc: Likewise, but also adjust
the code to handle the bounds of the operator.
* src/ltlparse/ltlparse.yy: Recognize "*" as "1*".
* src/ltlvisit/tostring.cc: Abbreviate "1*" as "*".
* src/tgbatest/ltl2tgba.test: Use the new syntax.
* src/ltlparse/ltlscan.ll: Recognize "[]=>" (a.k.a "|=>") and "<>=>".
* src/ltlparse/ltlparse.yy: Support them by rewriting them using
"[]->" and "<>->".
* src/tgbatest/ltl2tgba.test: More tests.
* src/ltlast/multop.cc, src/ltlast/multop.hh: Declare AndNML
operator.
* src/ltlparse/ltlscan.ll: Distinguish "&" and "&&".
* src/ltlparse/ltlparse.yy: Handle them both as "And" for LTL
formula, use AndNML or And for rational expressions.
* src/ltlvisit/tostring.cc: Adjust to distinguish "&" and "&&" in
rational expressions. Also use {braces} to group rational
expressions.
* src/tgbaalgos/ltl2tgba_fm.cc
(ratexp_trad_visitor::ratexp_trad_visitor): Remove the possibility
to select the empty_word should act like true, and fix the rules
for Closure and NegClosure to rely on constant_term instead.
(ratexp_trad_visitor::visit) Adjust the And translation to also
support AndNML.
(ratexp_trad_visitor::recurse_and_concat): Introduce this new
method to simplify some calls to recurse(f, to_concat_).
* src/tgbatest/ltl2tgba.test: Add more test cases.
* src/ltlvisit/basicreduce.cc, src/ltlvisit/consterm.cc,
src/ltlvisit/contain.cc, src/ltlvisit/mark.cc,
src/ltlvisit/nenoform.cc, src/ltlvisit/syntimpl.cc,
src/tgba/formula2bdd.cc, src/tgbaalgos/eltl2tgba_lacim.cc,
src/tgbaalgos/ltl2taa.cc, src/tgbaalgos/ltl2tgba_lacim.cc: Add
missing cases in switches.
* src/ltlparse/ltlparse.yy (parenthesedsubformula): Extract these
rules from...
(subformula): ... here, and use it to recognize "{SERE}(formula)".
* src/ltlparse/ltlscan.ll: Recognize "|->" as "[]->".
* src/ltltest/equals.test: Test these two new syntaxes.
* src/ltlast/unop.hh, src/ltlast/unop.cc: Introduce Closure and
NegClosure operators.
* src/ltlparse/ltlparse.yy: Recognize {foo} as a Closure.
* src/ltlvisit/mark.cc: Consider NegClosure as a marked operator.
* src/tgbaalgos/ltl2tgba_fm.cc (ratexp_trad_visitor): Add option to
select whether the empty_word should act like true (for {SERE}
and {!SERE}) or false (for {SERE}<>->Exp or {SERE}[]->Exp).
(ltl_trad_visitor): Translate Closure and NegClosure.
* src/tgbatest/ltl2tgba.test: Add more tests.
* src/ltlvisit/basicreduce.cc, src/ltlvisit/consterm.cc,
src/ltlvisit/nenoform.cc, src/ltlvisit/reduce.cc,
src/ltlvisit/syntimpl.cc, src/ltlvisit/tostring.cc,
src/ltlvisit/tunabbrev.cc, src/tgba/formula2bdd.cc,
src/tgbaalgos/eltl2tgba_lacim.cc, src/tgbaalgos/ltl2tgba_lacim.cc,
src/tgbaalgos/ltl2taa.cc: Straightforward update to support or
assert on these new operators.
* src/ltlparse/ltlparse.yy (parse_ratexp): New function.
(START_LTL, START_RATEXP): Add these new tokens.
(result): Parse and LTL formula or a rational expression depending
on the start symbol.
* src/ltlparse/public.hh (parse_ratexp): New function.
* src/ltlparse/parsedecl.hh (flex_set_buffer): Add a new argument
to set the starting rule.
* src/ltlparse/ltlscan.ll (flex_set_buffer): Adjust.
(start_token): New global variable. Return this as first token if
it is set.
* src/ltlast/binop.hh, src/ltlast/binop.cc (EConcat, UConcat):
Add these new operators.
* src/ltlparse/ltlparse.yy, src/ltlparse/ltlscan.ll: Parse
these new operators.
* src/ltlvisit/simpfg.cc, src/ltlvisit/syntimpl.cc,
src/ltlvisit/tostring.cc, src/ltlvisit/basicreduce.cc,
src/ltlvisit/consterm.cc, src/ltlvisit/lunabbrev.cc,
src/ltlvisit/nenoform.cc, src/ltlvisit/reduce.cc
src/tgba/formula2bdd.cc, src/tgbaalgos/eltl2tgba_lacim.cc,
src/tgbaalgos/ltl2taa.cc, src/tgbaalgos/ltl2tgba_fm.cc,
src/tgbaalgos/ltl2tgba_lacim.cc: Add these new operators into the
switches.
Trivial simplifications rules (such as "FFa=Fa" or "x&1=x")
are performed any time a formule is instanciated.
* src/ltlast/constant.hh, src/ltlast/constant.cc
(true_instance, true_instance_): Declare the true_instance_ as a
static member, and move true_instance() into the .hh so it gets
inlined. Have true_instance_ as a class variable will ensure that
it is the first formula instantiated. Binop simplifications rely
on this to order arguments.
(false_instance, false_instance_): Likewise.
(empty_word_instance, empty_word_instance_): New method and static
member.
* src/ltlast/formula.hh (formula::formula): If max_count_ ever
loops, skip the first three values so that constants always have
smaller hash codes.
* src/ltlast/binop.hh, src/ltlast/binop.cc (instance): Add
simplifications and document them.
* src/ltlast/multop.hh (multop::Concat): New operator.
* src/ltlast/multop.cc (op_name): Handle Concat.
(instance): Inline Concat arguments without reordering. Handle
absorbent and neutral elements for all operators.
* src/ltlast/unop.hh (unop::Star): New operator.
* src/ltlast/unop.cc (op_name): Handle Star.
(instance): Handle Star, and add trivial simplifications for
other unary operators.
* src/ltlparse/ltlparse.yy (OP_CONCAT, OP_STAR, CONST_EMPTYWORD):
Declare these new operators and add rules for them.
* src/ltlparse/ltlscan.ll (OP_CONCAT, OP_STAR, CONST_EMPTYWORD):
Output these new operators.
* src/ltltest/equals.test: New tests.
* src/ltltest/parse.test: Remove redundant test.
* src/ltltest/tunabbrev.test, src/tgbatest/emptchk.test: Adjust tests.
* src/ltlvisit/basicreduce.cc, src/ltlvisit/contain.cc,
src/ltlvisit/nenoform.cc, src/ltlvisit/reduce.cc,
src/ltlvisit/syntimpl.cc, src/ltlvisit/tostring.cc,
src/ltlvisit/tunabbrev.cc: Complete visitors to handle new
operators.
* src/ltltest/nenoform.test: More tests.
* src/ltlvisit/lunabbrev.cc (unabbreviate_logic_visitor::visit):
Clone formulae before instance() function actually have a chance
to destroy them.
* src/tgba/formula2bdd.cc, src/tgbaalgos/eltl2tgba_lacim.cc,
src/tgbaalgos/ltl2taa.cc, src/tgbaalgos/ltl2tgba_fm.cc,
src/tgbaalgos/ltl2tgba_lacim.cc: Adjust switches to assert on new
operators.
* src/ltlast/binop.cc, src/ltlast/binop.cc: Add support for
these new operators.
* src/ltlparse/ltlparse.yy, src/ltlparse/ltlscan.ll: Parse them.
* src/ltltest/reduccmp.test: Add new tests for W and M.
* src/ltlvisit/basicreduce.cc, src/ltlvisit/contain.cc,
src/ltlvisit/lunabbrev.cc, src/ltlvisit/nenoform.cc,
src/ltlvisit/randomltl.cc, src/ltlvisit/randomltl.hh,
src/ltlvisit/reduce.cc, src/ltlvisite/simpfg.cc,
src/ltlvisit/simpfg.hh, src/ltlvisit/syntimpl.cc,
src/ltlvisit/tostring.cc, src/tgba/formula2bdd.cc,
src/tgbaalgos/eltl2tgba_lacim.cc, src/tgbaalgos/ltl2taa.cc,
src/tgbaalgos/ltl2tgba_fm.cc, src/tgbaalgos/ltl2tgba_lacim.cc:
Add support for W and M.
* src/tgbatest/ltl2neverclaim.test: Test never claim output
using LBTT, this is more thorough. Also we cannot use -N
any more in the spotlbtt.test.
* src/tgbatests/ltl2tgba.cc: Define M and W for ELTL.
* src/tgbatest/ltl2neverclaim.test: Test W and M, and use
-DS instead of -N, because lbtt-translate does not want
to translate these operators for tools that masquerade as Spin.
* src/ltlparse/ltlparse.yy, src/tgbaparse/tgbaparse.yy,
src/evtgbaparse/evtgbaparse.yy, src/eltlparse/eltlparse.yy: Use
token types for %destructor and %printer. Remove the yylex hack,
since %name-prefix is now honored by Bison. Also remove the
useless <token> type. Suggested by Akim Demaille.
* src/ltlparse/ltlparse.yy: Change the precedence of "->" and
"<->" so that "a & b -> c" is interpreted as "(a & b) -> c"
instead of "a & (b -> c)". The new interpretation is more
intuitive, and matches that of LBTT.
* HACKING: Mention that we require Bison >= 2.4 for developers.
* src/evtgbaparse/evtgbaparse.yy, src/tgbaparse/tgbaparse.yy,
src/ltlparse/ltlparse.yy: The sections "%{ ... %}" should now be
renamed "%code requires { ... }" or "%code { ... }" depending on
whether they should end up in the parser's header file or its cc
file. Also use %language, %locations, %defines, instead of
command-line arguments.
* src/evtgbaparse/Makefile.am, src/tgbaparse/Makefile.am,
src/ltlparse/Makefile.am: Remove the --locations, --defines
and --languages in the call to bison. Add -Wall -Werror to
catch more errors.
Thanks to Akim Demaille <akim@lrde.epita.fr> for the help.
(subformula): Recognize `ATOMIC_PROP OP_POST_POS' and
`ATOMIC_PROP OP_POST_NEG'.
* src/ltlparse/ltlscan.ll: Introduce the not_prop start condition,
to restrict the set of atomic propositions allowed in places
where they are not expected. Make `true' and `false' case insensitive.
* src/ltltest/parse.test, src/ltltest/tostring.test: More cases.
* src/ltlvisit/tostring.cc (to_string_visitor): Quote atomic
propositions equal to "true" or "false".
* src/ltlparse/ltlparse.yy: Simplify error handling now that Bison
will call destructors. Give each operator a full name, so that
Bison uses it in error messages.
* src/ltlast/binop.cc (binop::instance): Order operands for
associative operators, so that e.g. "a xor b" and "b xor a" are
mapped to the same formula.
* src/ltltest/equals.test: Check this.
