Christoph Mallon [Fri, 14 Nov 2008 15:16:27 +0000 (15:16 +0000)]
Teach cparser how to parse (f)(void);, i.e. global declarations can start with a (.
[r23658]
Matthias Braun [Fri, 14 Nov 2008 12:55:25 +0000 (12:55 +0000)]
delete temp files
[r23653]
Matthias Braun [Fri, 14 Nov 2008 12:40:15 +0000 (12:40 +0000)]
more cleanup
[r23651]
Matthias Braun [Fri, 14 Nov 2008 12:37:57 +0000 (12:37 +0000)]
use ushort wchar_t on mingw
[r23650]
Matthias Braun [Fri, 14 Nov 2008 12:36:23 +0000 (12:36 +0000)]
missing break
[r23649]
Matthias Braun [Fri, 14 Nov 2008 12:35:45 +0000 (12:35 +0000)]
most of the options are handled in main.c now
[r23648]
Matthias Braun [Fri, 14 Nov 2008 12:34:35 +0000 (12:34 +0000)]
- Try to move OS/target specific stuff to lang_features.h and main.c
- In this case: initialize types in types.c and set stuff like short_wchar
in main
[r23647]
Matthias Braun [Thu, 13 Nov 2008 14:02:44 +0000 (14:02 +0000)]
fix local labels
[r23632]
Matthias Braun [Thu, 13 Nov 2008 13:36:42 +0000 (13:36 +0000)]
pass correct modes to 64bit lowerer
[r23626]
Matthias Braun [Thu, 13 Nov 2008 12:36:22 +0000 (12:36 +0000)]
no need for a separate local_label stack, the scoping rules are exactly the same as for the normal environment_stack
[r23622]
Matthias Braun [Thu, 13 Nov 2008 11:23:04 +0000 (11:23 +0000)]
avoid warnings and fix potential bugs (if someone manages to get a % into a function name
[r23615]
Matthias Braun [Thu, 13 Nov 2008 10:26:42 +0000 (10:26 +0000)]
fix assert
[r23612]
Matthias Braun [Tue, 11 Nov 2008 19:56:32 +0000 (19:56 +0000)]
c frontend doesn't case about modulo_shift, rename modes to uppercase so it dumps nicer
[r23580]
Christoph Mallon [Mon, 10 Nov 2008 09:38:26 +0000 (09:38 +0000)]
In C++ the result type of !, <, <=, >=, >, == and != is bool.
[r23539]
Christoph Mallon [Mon, 10 Nov 2008 09:23:31 +0000 (09:23 +0000)]
Parse C++ bool, false and true.
[r23538]
Matthias Braun [Fri, 7 Nov 2008 08:18:40 +0000 (08:18 +0000)]
more work towards expansion of macros with arguments
[r23505]
Michael Beck [Thu, 6 Nov 2008 15:09:52 +0000 (15:09 +0000)]
- renamed option end-melt to shape-blocks
- let the switch lowerer run after the block shaper
[r23490]
Christoph Mallon [Tue, 4 Nov 2008 15:00:36 +0000 (15:00 +0000)]
When creating an error typedef, set its type to the error type, so nobody encounters a null pointer later on.
[r23434]
Christoph Mallon [Tue, 4 Nov 2008 14:51:53 +0000 (14:51 +0000)]
Do not panic, when finding a typedef name, when an expression is expected.
[r23433]
Christoph Mallon [Tue, 4 Nov 2008 14:37:49 +0000 (14:37 +0000)]
use expect() to find the ] of an array access expression.
[r23432]
Christoph Mallon [Tue, 4 Nov 2008 14:34:01 +0000 (14:34 +0000)]
Set the source position in allocate_expression_zero().
[r23431]
Christoph Mallon [Tue, 4 Nov 2008 14:28:10 +0000 (14:28 +0000)]
Tell the ast printer a thing ot two about EXPR_REFERENCE_ENUM_VALUE.
[r23430]
Matthias Braun [Tue, 4 Nov 2008 10:04:31 +0000 (10:04 +0000)]
fix parsing of define argument lists, fix subtle problems with first token of included file not expanded, fix skip_spaces ignoring comments
[r23425]
Matthias Braun [Tue, 4 Nov 2008 10:03:34 +0000 (10:03 +0000)]
modifier to print preprocessing token
[r23424]
Matthias Braun [Tue, 4 Nov 2008 10:03:12 +0000 (10:03 +0000)]
print tokens without surrounding ''
[r23423]
Matthias Braun [Tue, 4 Nov 2008 09:43:14 +0000 (09:43 +0000)]
more tricky preprocessor problems
[r23421]
Matthias Braun [Mon, 3 Nov 2008 16:25:11 +0000 (16:25 +0000)]
if(n)def implementation
[r23414]
Matthias Braun [Mon, 3 Nov 2008 01:22:50 +0000 (01:22 +0000)]
implemented #include
[r23399]
Matthias Braun [Sun, 2 Nov 2008 16:58:26 +0000 (16:58 +0000)]
add beginnings of preprocessor code (and ugly hack to activate it) to svn
[r23396]
Matthias Braun [Sun, 2 Nov 2008 16:36:26 +0000 (16:36 +0000)]
EXPR_INVALID gets constructed in case of errors
[r23392]
Matthias Braun [Sun, 2 Nov 2008 16:33:58 +0000 (16:33 +0000)]
improve flexible array member handling, assign error_type to error_entity
[r23390]
Matthias Braun [Sun, 2 Nov 2008 14:41:31 +0000 (14:41 +0000)]
set has_implicit_size flag
[r23385]
Matthias Braun [Sun, 2 Nov 2008 11:44:30 +0000 (11:44 +0000)]
TODO update
[r23382]
Matthias Braun [Sat, 1 Nov 2008 19:18:58 +0000 (19:18 +0000)]
Move more C tests
[r23372]
Matthias Braun [Sat, 1 Nov 2008 19:18:31 +0000 (19:18 +0000)]
move C frontend tests
[r23371]
Matthias Braun [Sat, 1 Nov 2008 17:39:06 +0000 (17:39 +0000)]
add main, fix warnings not intended
[r23357]
Matthias Braun [Sat, 1 Nov 2008 17:32:22 +0000 (17:32 +0000)]
this should not be here
[r23356]
Matthias Braun [Sat, 1 Nov 2008 17:20:48 +0000 (17:20 +0000)]
write a testapp that really test functionality (so we can actually detect the implementation bugs)
[r23355]
Matthias Braun [Sat, 1 Nov 2008 17:12:31 +0000 (17:12 +0000)]
local labels should be fixed
[r23354]
Matthias Braun [Sat, 1 Nov 2008 17:11:17 +0000 (17:11 +0000)]
partly fix local labels, %P already prints %P an at no need for additional ones in the messages
[r23353]
Matthias Braun [Sat, 1 Nov 2008 17:05:58 +0000 (17:05 +0000)]
conditional expressions true case may be NULL as a GNU extension
[r23352]
Matthias Braun [Sat, 1 Nov 2008 17:00:35 +0000 (17:00 +0000)]
declaration_t -> entity_t
This breaks up the monotolithic declaration_t monster in a bunch of different
structs like variable_t, function_t or label_t
Not everything back to the way it was, but SPEC seems to work now. Just some obscure gnu extensions missing.
[r23351]
Matthias Braun [Sat, 1 Nov 2008 16:55:14 +0000 (16:55 +0000)]
fix warning
[r23349]
Matthias Braun [Sat, 1 Nov 2008 15:45:01 +0000 (15:45 +0000)]
this is a gnu extension I guess
[r23348]
Matthias Braun [Sat, 1 Nov 2008 15:17:47 +0000 (15:17 +0000)]
test implicit type for k&r functions
[r23347]
Michael Beck [Tue, 28 Oct 2008 15:54:18 +0000 (15:54 +0000)]
- add option -end-melt
[r23282]
Michael Beck [Mon, 27 Oct 2008 18:03:02 +0000 (18:03 +0000)]
- further fixed r23189
[r23242]
Christoph Mallon [Mon, 27 Oct 2008 15:44:09 +0000 (15:44 +0000)]
Fix anchor token underflow while parsing for-statements.
[r23236]
Christoph Mallon [Mon, 27 Oct 2008 15:40:15 +0000 (15:40 +0000)]
Parse sizeof(T){...} correctly, i.e. not sizeof type, but sizeof expression starting with a compound literal.
[r23234]
Christoph Mallon [Mon, 27 Oct 2008 12:17:13 +0000 (12:17 +0000)]
Add the C++ :: token.
[r23226]
Christoph Mallon [Mon, 27 Oct 2008 11:40:23 +0000 (11:40 +0000)]
Parse and check semantic of delete expressions.
[r23225]
Christoph Mallon [Mon, 27 Oct 2008 11:08:09 +0000 (11:08 +0000)]
C++ banned implicit int.
[r23224]
Christoph Mallon [Mon, 27 Oct 2008 11:00:18 +0000 (11:00 +0000)]
C++ does not allow assignment from a void pointer type to a non-void pointer type.
[r23223]
Christoph Mallon [Mon, 27 Oct 2008 10:50:48 +0000 (10:50 +0000)]
C++ has not "tentative definitions".
[r23222]
Christoph Mallon [Mon, 27 Oct 2008 10:46:15 +0000 (10:46 +0000)]
C++ function declarations with an empty parameter list denote no parameters instead of unspecified parameters.
[r23221]
Christoph Mallon [Mon, 27 Oct 2008 10:32:09 +0000 (10:32 +0000)]
Parse and check semantic of throw expressions.
[r23220]
Christoph Mallon [Mon, 27 Oct 2008 06:48:38 +0000 (06:48 +0000)]
Remove if, whose condition is never true.
[r23219]
Christoph Mallon [Sun, 26 Oct 2008 20:55:23 +0000 (20:55 +0000)]
In C++ (most) character literals are of type char.
[r23218]
Christoph Mallon [Sun, 26 Oct 2008 20:38:34 +0000 (20:38 +0000)]
Add C++ keywords.
[r23217]
Christoph Mallon [Sun, 26 Oct 2008 19:49:53 +0000 (19:49 +0000)]
First bit of C++ support: Different precedence for ?:.
[r23216]
Christoph Mallon [Sun, 26 Oct 2008 19:48:46 +0000 (19:48 +0000)]
Add support for C++ in the compiler driver.
[r23215]
Matthias Braun [Sun, 26 Oct 2008 18:44:46 +0000 (18:44 +0000)]
more tests
[r23214]
Matthias Braun [Sun, 26 Oct 2008 18:04:37 +0000 (18:04 +0000)]
slightly improve testcase
[r23213]
Christoph Mallon [Sun, 26 Oct 2008 17:54:17 +0000 (17:54 +0000)]
Consistently name the directories should_*.
[r23212]
Christoph Mallon [Sun, 26 Oct 2008 17:50:49 +0000 (17:50 +0000)]
cp_error050: sizeof a compound literal.
[r23211]
Christoph Mallon [Sun, 26 Oct 2008 17:46:08 +0000 (17:46 +0000)]
Fix parsing of expression statements, which pointlessly start with sizeof or __alignof__, and sort the case list.
[r23210]
Matthias Braun [Sun, 26 Oct 2008 12:26:06 +0000 (12:26 +0000)]
fix typo
[r23207]
Matthias Braun [Sun, 26 Oct 2008 12:23:44 +0000 (12:23 +0000)]
this was not meant to be here
[r23206]
Christoph Mallon [Sun, 26 Oct 2008 12:11:03 +0000 (12:11 +0000)]
Change -w so that warnings can be activated again later on.
[r23205]
Matthias Braun [Sun, 26 Oct 2008 12:02:46 +0000 (12:02 +0000)]
more warning testcases
[r23204]
Matthias Braun [Sun, 26 Oct 2008 10:03:16 +0000 (10:03 +0000)]
source-position doesn't make any sense for types:
- atomic, pointer, function types are used throughout the program and always
defined "on-the-fly" it's not possible to assign a correct source_position
- for struct, union and enum types we have to create a declaration_t anyway
so they can get referenced by their name. In this case we can and should
use the source_position of the declaration
[r23203]
Christoph Mallon [Sun, 26 Oct 2008 08:30:31 +0000 (08:30 +0000)]
Prefix operators need no precedence.
[r23201]
Michael Beck [Sun, 26 Oct 2008 02:04:06 +0000 (02:04 +0000)]
- crashes, but should give an error
[r23200]
Michael Beck [Sun, 26 Oct 2008 00:24:57 +0000 (00:24 +0000)]
- BugFix: gcc optimizes def_nr + 1 < 0 away, arg\!
[r23196]
Michael Beck [Sun, 26 Oct 2008 00:08:44 +0000 (00:08 +0000)]
- BugFix: generation of default case number was completely broken
[r23195]
Michael Beck [Sat, 25 Oct 2008 22:37:06 +0000 (22:37 +0000)]
- fixed r23189: the AST-printer doubles the precesion to have one free
bit for right-to-left handling
[r23194]
Christoph Mallon [Sat, 25 Oct 2008 20:06:26 +0000 (20:06 +0000)]
Unify the precedence lists of the parser and the AST printer.
[r23189]
Matthias Braun [Sat, 25 Oct 2008 19:42:22 +0000 (19:42 +0000)]
improved switch.c
[r23188]
Christoph Mallon [Sat, 25 Oct 2008 19:26:26 +0000 (19:26 +0000)]
Replace magic numbers for expression precedences by an enum. Do not unnecessarily pass the precedence to the expression parsers.
[r23184]
Matthias Braun [Sat, 25 Oct 2008 17:10:15 +0000 (17:10 +0000)]
get rid of some fields in ast_t: outer_fkt_jmp, is_outer_ref, is_parameter
[r23181]
Christoph Mallon [Sat, 25 Oct 2008 17:01:13 +0000 (17:01 +0000)]
Fix some precedence stuff for expression parsing.
[r23180]
Matthias Braun [Sat, 25 Oct 2008 14:44:06 +0000 (14:44 +0000)]
some testcases I had lying around here
[r23179]
Christoph Mallon [Fri, 24 Oct 2008 17:30:32 +0000 (17:30 +0000)]
While our semantics of _Bool are wrong, at least make it the right size (for most platforms).
[r23175]
Christoph Mallon [Fri, 24 Oct 2008 17:26:56 +0000 (17:26 +0000)]
Use make_pointer_type().
[r23174]
Matthias Braun [Thu, 23 Oct 2008 21:59:02 +0000 (21:59 +0000)]
asm_name is not used
[r23149]
Christoph Mallon [Tue, 21 Oct 2008 10:11:07 +0000 (10:11 +0000)]
skip_typeref().
[r23060]
Christoph Mallon [Tue, 21 Oct 2008 08:13:29 +0000 (08:13 +0000)]
Improve unused variable check.
[r23059]
Christoph Mallon [Tue, 21 Oct 2008 06:11:05 +0000 (06:11 +0000)]
Improve unused variable check.
[r23057]
Christoph Mallon [Tue, 21 Oct 2008 06:07:12 +0000 (06:07 +0000)]
Implement a statement walker.
[r23056]
Christoph Mallon [Mon, 20 Oct 2008 10:47:54 +0000 (10:47 +0000)]
Remove a variable, which was found by the new, shiny -Wunused-variable.
[r23027]
Christoph Mallon [Mon, 20 Oct 2008 10:46:10 +0000 (10:46 +0000)]
Another attempt to solve the halting problem: Detect variables, which are never read. The test is cheap and could be improved by using dataflow, SSA and fixpoint magic, but of course it is lightyears ahead of GCC. For example in x = x + 1; x is not marked as read, because it is only used to assign to itself.
[r23026]
Christoph Mallon [Mon, 20 Oct 2008 08:40:04 +0000 (08:40 +0000)]
Make check_unreachable() more tail recursion friendly.
[r23021]
Christoph Mallon [Mon, 20 Oct 2008 08:35:58 +0000 (08:35 +0000)]
Resolve assertion, when using -fsyntax-only.
[r23020]
Christoph Mallon [Sun, 19 Oct 2008 17:30:24 +0000 (17:30 +0000)]
Remove stale (since r21011) comment.
[r23016]
Christoph Mallon [Sun, 19 Oct 2008 17:25:56 +0000 (17:25 +0000)]
Fix r23014.
[r23015]
Christoph Mallon [Sun, 19 Oct 2008 16:49:14 +0000 (16:49 +0000)]
Do not generate an error message about incompatible types in ?: if the error type is involved.
[r23014]
Christoph Mallon [Sun, 19 Oct 2008 16:42:19 +0000 (16:42 +0000)]
Resolve some warnings.
[r23013]
Christoph Mallon [Sun, 19 Oct 2008 16:27:40 +0000 (16:27 +0000)]
Add forgotten conversion specifiers.
[r23012]
Christoph Mallon [Sun, 19 Oct 2008 16:19:50 +0000 (16:19 +0000)]
Resolve some warnings.
[r23011]
Christoph Mallon [Tue, 14 Oct 2008 15:16:52 +0000 (15:16 +0000)]
-fopt-foo means activate, not deactivate.
[r22879]