[cfe-dev] Unneeded declarations invalidation

[Enea Zaffanella]

Eli Friedman wrote:
> On Thu, Jul 2, 2009 at 1:30 AM, Abramo Bagnara<abramobagnara at tin.it> wrote:
>> The following typescript show what happens when trying to compile a
>> simple C source not legal for C99 standard, but accepted and compiled by
>> gcc.

[...SNIP...]

>> OTOH in both cases, I'd like to avoid to invalidate the declaration and
>> thus to inhibit ast generation if this is not strictly needed.
> 
> Mmm... yeah, even if we don't fix this, we probably don't need to mark
> the declaration invalid; it's well-formed in any case.
> 
> -Eli

Hello.

Please find attached a small patch that fixes what is, to my 
understanding, another occurrence of an unneeded declaration 
invalidation (triggered by non-standard use of VM types).

The testcase is the following:
====================
void p(int a) {
   struct {
     char x[(int)(char*)2 - 4000];
     char y[a];
   } f;
}
====================

which currently clang treats as follows:

====================
# Debug/bin/clang-cc -ast-print bug.c
bug.c:4:10: error: array size is negative
     char x[(int)(char*)2 - 4000];
          ^
bug.c:5:10: error: fields must have a constant
       size: 'variable length array in structure' extension will never be
       supported
     char y[a];
          ^
[...SNIP...]
void p(int a) {
     struct {
         int x;
         int y;
     } f;
}


2 diagnostics generated.
====================

After the patch, it will become:

====================
# Debug/bin/clang-cc -ast-print bug.c
bug.c:4:10: error: array size is negative
     char x[(int)(char*)2 - 4000];
          ^
bug.c:5:10: error: fields must have a constant
       size: 'variable length array in structure' extension will never be
       supported
     char y[a];
          ^
[...SNIP...]
void p(int a) {
     struct {
         char x[(int)(char *)2 - 4000];
         char y[a];
     } f;
}


2 diagnostics generated.
====================

Hence, the error is still reported as it was,
but the AST generation is more faithful to the original.
The latter property is important for clients such as ours.

Cheers,
Enea Zaffanella.
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: VM_types_in_records.patch
URL: <http://lists.llvm.org/pipermail/cfe-dev/attachments/20090707/b0e6dce3/attachment.ksh>