Fix compilation error on platforms where real64_t is mapped onto 'long double'

Correctly handle structured variables that include FB and external FBs (example: FB1.FB2.extFB3.FB4.extFB5.extFB6.structvar.field1 := 42;) (this commit fixes only ST code. TODO: IL and SFC)

Remove call to type_is_complex() in ST code generation.

A little code cleanup (reduce number of possible variable generation states)

Finish: Change order of SET_xxx() macros. (this will allow me to simplify the print_setter() methods later on)

Change order of SET_xxx() macros. (this will allow me to simplify the print_setter() methods later on)

More static declarations in POUS.c in order to avoid conflicting symbol definitions with new POUS.c inclusion scheme

Fix multiple resources support, POUS.c now only contain static definitions. POUS.c is included in configuration C code as well as in resources to allow GLOBAL FBs to be declared in configurations

Disable stage4 options with on windows since getsuboptr isn't supported in mingw

merge

Track of line number correctly when input file has code with very long lines.

Fix memory leak in lexical parser.

Make generation of #line directives optional.

Insert #line directives in generated C code.

Add support for parsing of REF_TO keyword (defined in IEC61131-3 v3). This support is currently inactive (in flex) until we get to add C code generation later on.

Update files to be ignored by mercurial: ignore files generated by autoconf, directory test/, and *.orig

Fix bug: do not print IL 'label' when generating inline functions (JMPxx label was generating erroneous C code)

Add limited support for the REF() operator (defined in v3 of IEC 61131-3)

Remove assertion being failed by IL labels (IL labels do not yet have specific datatypes).

Added stage1_2/Makefile.am weird rule to keep compatible with previous version of automake

Merge

fix definition of pragma.

Fix the state machine that became broken 2 commits ago (when adding support for nested comments)

Add option to control support for nested comments (default is off, as defined in IEC 61131-3 v2)

Add support for nested comments

Fix bug: when checking compatibility between GLOBAL and EXTERNAL variables, must only enforce GLOBAL CONSTANT => EXTERNAL CONSTANT.
Reverse CONSTANT implication is not enforced, and neither is RETAIN compatibility.

merge

Fix bug: correctly generate code when accessing external variables declared inside FBs as a structured variable (realvar := fb1.fb2.extvar1.realvar)

Fix bug in standard: standard does not allow multiple VAR_GLOBAL ... END_VAR constructs in configurations and resources. This is probably a bug, so we allow it.

Code cleanup (part 3): generate_c_typedecl_c is no longer needed for code generation in POUS.c It is now only needed for datatype declaration in POUS.h

Code cleanup (part 2): generate_typedecl_c now only prints to POUS.h !

Code cleanup (part 1): subrange check functions are now declared in POUS.h (as static inline functions or #define)

Fix get_datatype_info_c::is_subrange(), which did not work when using base type! (we now use get_equivtype() instead of get_base_type() )

fix a couple of typos.

Fix bug - correctly declare struct members whose type is a directly defined array (e.g.: STRUCT x: ARRAY of XXX; END_STRUCT)

Start using the called_fb_declaration annotation when generating C code from FB calls in ST.

Code cleanup: move datatype analysis to get_datatype_info_c

Fix bug-fix of previous commit.

Fix bug: allow use, as lvalues, structures/arrays inside FBs (e.g. fb1.struct1.r := 33.3).

Fix bug/issue #33 (correctly access struct/array variables declared inside a FB -> r:=FB1.FB2.struct1.array1[3] )

Add assertion suggested by Manuele.

merge

Use get_datatype_info_c::is_type_valid() to determine datatype validity
(some datatype bugs in source code were going unreported!)

Recursively check the datatype compatibility of values/expressions passed in function/FB invocations.

Change error message text so as to become more suitable to where they might occur in the source code.

Add code to check if an IN_OUT variable is being passed an IL list in formal IL FB/function invocations.

Generate correct error message when encountering IL lists embedded in IL formal invocations.

Fill in the 'datatype' anotation in the identifiers of symbolic variables.
(even though currently it is not being used).

Fix C code generation of FB invocation in IL.

Fix detection of datatype errors on IL conditional flow control operators (JMPC, RETC, ...)

Allow array_dimension_iterator to accept an array_spec_init_c.
Fixes bug activated by generating C code from IL code containing array variables.

Small code cleanup (move common code to a function)

Fix datatype analysis of conditional IL operators (CALC, CALCN, RETC, RETCN, JMPC, JMPCN, S and R)

Fix datatype analyses of S and R IL operators.

make sure all IL operands are narrowed (datatype checking algorithm)

cosmetic change only - fix code alignment.

Add support for FB call semantics of 'S' and 'R' IL operators!
Remove segfaults when analysing buggy IL code (IL operators with no operands).

Stop lvalue check from segfaulting when coming across buggy IL code (IL operator with no operand!)

Allow get_datatype_info_c to be called with NULL parameters.

Do lvalue check of function output parameters (since they may contain expressions inside array subscripts!)

Handle buggy source code gracefully (do not bork when non-array variable is used as an array. e.g: int_var[42]:= 33)

Delete debugging messages left in by mistake.

Fix segfault when doing lvalue check of buggy IEC 61131-3 source code

Fix datatype analysis of structured variables that contain arrays in their fields (e.g. var.x1[var2 + 42].y1).

Add capability of returning array subscript list while decomposing a struct/array variable.

Code cleanup: Remove unused option = Delete dead code.

merge

Declare that YYLSTYPE is not the default trivial datatype used by bison. (was causing compile errors in bison ver >= 2.7)

Fix build on Ubuntu 13.10

Fixed warning when compiling generated program containing external variable assignments

Enlarged flex buffer size in order to accept larger C pragmas

Missing braces arround extern declarartions in __SET_EXTERNAL define was making problem with GCC in switch case statements

Moved __INIT_* defines within type definition headers files

Fixed warning in accessor __SET_EXTERNAL

Fixed bug when defining task with single_data_source instead of interval_data_source

Merged Mario's changes

Fix bug related to FB invocation of FB instances declared as VAR_EXTERN.

Merged Mario's changes

Partial fix to VAR_EXTERN vs VAR_GLOBAL check. Now considers globals declared in resources too. TODO: recursively check FB declared in the program.

Allow variables declared inside resources to be referenced by res_name.var_name

Add a new node to the abstract symtax tree, which will let us do datatype checking of FB variable declarations using the standard algorithm, and no special cases.

Print filename when debugging abstract symbol table.

Consider fb_name_decl_c a non base data type class!

Fill symbol->datatype anotation for FB declarations.

Fill in the symbol.datatype annotation in symbol classes used in derived data type declarations.

Use simple_spec_init_c inside en_param_declaration_c (will reduce need to handle it as a special case in the future).

Merged

Fixed bug in string format when translating TIME and DATETIME to STRING

Fixed bug with SFC transition datatype checking

Stripped generated files again. http://stackoverflow.com/questions/3290908/which-files-generated-by-autotools-should-i-keep-in-version-control-repository

Fixed bug in case statement with enumerated type variable as case expression

Merged

Set the symbol->datatpe of function_declaration_c

Fill in the symbol->datatype of elementary datatypes.

merge

merge

Fix divide by 0 bug.

grand merge

disable not yet complete constant propagation algorithm (currently brocken and producing incorrect results),

merge

Fix constant propagation for repeat_statement_c class.

Remove unused variable.

Implement Mario's suggestions:
- Clean up code replace merge algorithm with a function.
- Fix for_statement setting control variable to not a constant.
- Fix for_statement checking BY expression to get correct value for optimization
branch.

Thanks Mario.

Fix constant_folding missing call. [Bug found by Mario.]

Fix divide by 0 bug.

merge

Fix CPPFLAGS Makefile.am for all stage4.

Remove redundant class for get variable name.

Fix constant propagation alg. in for statement like Mario suggestion.

Add note about How Constant Folding class is extended with a implementation constant propagation algorithm
by Mario de Sousa.

Add constant propagation algorithm for loop cycles.

Update symbol_c::first_file/last_file entries in list_c::add_element()

moved class to obtain datatype name (in char *) to get_datatype_info_c

merge

Fix comment about meet semilattice rules.

Complete fix initialize const_value.

Start to implement constant propagation algorithm.

Fix const_value field initialized.

Remove Ambiguous enumerate value in enum conversion functions.

Fix enum conversion function to string.

Fix struct generic_function_call_t initialize style.

Merge with Mario's changes.

Merge with Mario's repository.

Fix warning about call fprintf without format string.

merge

Remove Ambiguous enumerate value in enum conversion functions.

Fix enum conversion function to string.

Fix struct generic_function_call_t initialize style.

Small code cleanup.

Allow bison to ask for an END_OF_INPUT token multiple times!

Protect against redefining macros

merge

Fix flag to enable conversion functions.

Code cleanup: move tracking functions to flex.

Insert the auto generated enum datatype conversion functions after the TYPE...END_TYPE, instead of at the end of input file.

merge

Set ENO to false when conversion function fails.

Fix typo in functionDataType array.

Add files missing from a few commits ago...

merge

Refix automake files.

Add some comments to argument new code.

Start to change code like Mario suggestions.

Create functionDataType array in derived_conversion_functions_c.

Add conversion functions option.

Start enumerate conversion functions.

merge

Be sure to not redefine TRUE and FALSE macros.

merge

Removed unused __attribute__((__packed__)) only gcc compatible.

Protect IEC_STD_LIB.H header agaist multi-include problems.

merge

Start to move enumeration declaration checking to its own class.

Fix comments and code alignment

Fix warnings previous change introduced when compiling with gcc.

Fix some important clang++ warnings.

Allow use of type#value syntax when defining default initial value inside a TYPE..END_TYPE declaration.

Do datatype verification of fully qualified enumerated values (e.g. ENUMTYPE#enumconst)

Add datatype checking of enumeration data type declarations.

Print const values of symbol.

Fix code generation of enum value constants (this is a temporary fix - it will be followed by a code cleanup).

Fix bug in datatype checking of enumerated datatypes delcarations.

Fix datatype checking of enumerated datatype declarations. We now follow the fill->narrow algorithm correctly.

Add method to determine the id (name) of a datatype.

Change the base datatype of enumerations to an enumerated_type_declaration_c (for named types!)

Change the public interface to debug functions.

Check for errors in enum datatype declarations. (This is not yet complete as it will not yet stop the compilation process if these errors are found)

Fix one more bug - upper_bound() works a little differently to what we want!

Fix more bugs in dsymtable_c (real semantics of std::multimap::lower_bound is not what was assumed!)

Fix counting bug in dsymbtable_c

Use duplicate symtable instead of symtable for enum constant value table (this will later allow us to detect semantic errors in IEC 61131-3 source code)

Move the enumerated_value_symtable to fill_candidate_datatypes_c, since it is only used there!

Make search_base_type_c a singleton!

Start datatype checking of variable declarations (for now, only seting symbol.datatype of enumerations)

Do datatype checking of enum values defined inside anonymous enumeration types (i.e. enum types defined inside a VAR ... END_VAR daclaration).

Added support for testing BCD value validity in BCD_TO_UINT function

Fixed bug in DATE_AND_TIME_TO_TIME_OF_DAY and DATE_AND_TIME_TO_DATE when date and time is before 1970-01-01-00:00:00

Fix warning message when declaring global variables in configuration

Removed unused variable

Replacing libc time functions for date <-> timespec translation by platform and localtime independant implementation

Removed files generedted by autoconf, should not be in repo

Fixed timegm usage incompatible with mingw, reverted to mktime in case mingw is detected

Fix in accessor undefined init methods for function block global instances

Fix undefined function toupper in iec_std_lib

Adding support for declaring function block instances as global and external variables

Fix DATE conversions standard functions

Fix definition of BCD to ANY_UINT and ANY_UINT to BCD conversion standard functions

Fix indentation in SFC SL qualifier code generated

Fix indentation and unused variable warning in SFC generated code

Fix pragma with double bracket token stripping

Fix some warning about loss of precision data in iec_std_lib.h file.

Fix typo.

Remove deprecated class search_constant_type_c

replace call to search_constant_type_c with symbol.datatype

Removing several warning in iec_std_lib.h

Code formating only.

Fix stage4 for compare expressions.

Prepare to delete search_constant_type_c -> Move the static variables from search_constant_type_c to get_datatype_info_c

Remove no longer needed search_il_operand_type_c

Also force_narrow IL instructions inside IL expressions, i.e. inside parenthesis.

Correctly generate C code for IL expressions, i.e. IL instructions inside parenthesis.

Fix bug when handling IL expressions (i.e. IL instructions inside parenthesis).

Delete debugging code, introduced by mistake in previous commit.

Fix handling of IL parenthesized expressions in generate_c_il_c

Fix processing of TIME literals (submited by Andreas)

Fix flow control analysis of IL simple instructions (i.e. IL instructions inside parenthises!)

Comment out invalid assertion.

Code cleanup: get the datatype of each entity off symbol->datatype (set in stage3 datatype checking), instead of determining it again in stage4. (NOTE: handling of IL parenthesized expressions is not working! Will be fixed later.)

Major code cleanup: IL code generation now uses symbol.datatype (filled in during stage3 data type checking) instead of determining the symbol's datatype by itself.

Do not repeat narrow algorithm for ST code, and better comments.

New visitor classes missing from previous commit.

Remove line printing debug info.

Add classes to print status of AST (use only for debugging!)

Remove all remaining NULL (undefined) datatypes from IL code.

Move is_datatype_valid/equal() to get_datatype_info_c

Fix bug related to support of SAFE datatypes.

Set datatype of il_intruction_c objects containing an ST operator.

Delete unused variables.

Use doubly linked connections for flow control graph (will be used later).

Remove deprecated search_expression_type_c

Deprecate use of search_expression_type_c (prepare for complete removal!).

Deprecate use of search_expression_type_c (prepare for complete removal!)

Deprecating use of search_expression_type_c, so it may later be deleted!

Merge