- 03 May, 2010 1 commit
-
-
Gerald Pfeifer authored
-
- 26 Apr, 2010 1 commit
-
-
Gerald Pfeifer authored
-
- 05 Apr, 2010 2 commits
-
-
Alexandre Julliard authored
-
Alexandre Julliard authored
widl: Add a type_memsize_and_alignment function to make it explicit where the alignment is truly needed.
-
- 30 Mar, 2010 1 commit
-
-
Alexandre Julliard authored
-
- 26 Mar, 2010 1 commit
-
-
Alexandre Julliard authored
-
- 23 Mar, 2010 12 commits
-
-
Rob Shearman authored
widl: Check whether an expression is a constant one before assuming it is a toplevel conformance when writing conformance or variance descriptors.
-
Rob Shearman authored
This allows the code to not have to track the operator type twice.
-
Rob Shearman authored
It's only set, never read from.
-
Rob Shearman authored
-
Rob Shearman authored
-
Rob Shearman authored
It was just a simple indirection to get to a var_t, so just replace all uses of it with the latter.
-
Rob Shearman authored
-
Rob Shearman authored
-
Rob Shearman authored
In this case type_array_get_conformance will return NULL. The code here is actually replacing an existing array with an EXPR_VOID conformance with the specified conformance, so issue an error if type_array_get_conformance returns NULL or the conformance has already been defined.
-
Rob Shearman authored
-
Rob Shearman authored
widl: Convert assign_stub_out_args to using typegen_detect_type and fix the non-conformant array case.
-
Rob Shearman authored
-
- 16 Mar, 2010 16 commits
-
-
Alexandre Julliard authored
-
Alexandre Julliard authored
-
Rob Shearman authored
Clean valgrind runs of tested types and comparison with MIDL-generated code suggests that the existing code written in the finally block for client code is sufficient.
-
Rob Shearman authored
Remove the difficult manual walking of the alias tree and make the code easier to read and less likely to break if a new type is added.
-
Rob Shearman authored
Use a switch statement instead of multiple if statements and explicitly write out the fixed array cases.
-
Rob Shearman authored
widl: Remove some FIXME comments and replace them with comments explaining why the current behaviour is correct.
-
Rob Shearman authored
widl: Fix the writing of expressions in un-typedef'd structures by using write_type_left to write out the type, which adds the "struct ..." if necessary. Also improve the type comparison code to not crash on unnamed structures (although further changes need to be made to fix the output in this case) and make it more efficient by first doing a comparison on the values of the type pointers.
-
Rob Shearman authored
The FC code requires more than 4 bits to represent it and the correlation descriptor and union formats only allow 4 bits to represent the type referred to.
-
Rob Shearman authored
This allows write_conf_or_var_desc to have the context enabling it to write out an appropriate correlation descriptor when iid_is is used from within a structure.
-
Rob Shearman authored
widl: Use typegen_detect_type to determine which types should be written for the pointer description in write_struct_tfs. Otherwise extra pointers for user types could be written.
-
Rob Shearman authored
-
Rob Shearman authored
widl: Don't use the return value of write_embedded_types to determine whether a type contains pointers or not. It isn't correct in all situations, so type_has_pointer has to be used anyway so using the return value of write_embedded_types just clutters the code.
-
Rob Shearman authored
Avoid duplicating the pointer writing code.
-
Rob Shearman authored
The check for whether a pointer is the last one is redundant since typegen_detect_type does a stricter check on the ref type. Replace one other instance of last_ptr with a more readable equivalent.
-
Rob Shearman authored
Change write_pointer_tfs to rely on the detected type of the type to determine when to write a simple pointer rather than relying on typestring_offset not being set anywhere else.
-
Rob Shearman authored
-
- 01 Mar, 2010 1 commit
-
-
Gerald Pfeifer authored
-
- 08 Feb, 2010 1 commit
-
-
Alexandre Julliard authored
makefiles: Generate a general catch-all dependency on .tab.h files to replace the manually maintained ones.
-
- 22 Jan, 2010 1 commit
-
-
Rob Shearman authored
-
- 19 Jan, 2010 1 commit
-
-
Rob Shearman authored
-
- 11 Jan, 2010 1 commit
-
-
Paul Vriens authored
-
- 29 Nov, 2009 1 commit
-
-
Rob Shearman authored
-