[Bug tree-optimization/13741] [3.3/3.4 Regression] [tree-ssa] Variable-length arrays broke in gcc 3.x

["mark at codesourcery dot com" <gcc-bugzilla at gcc dot gnu dot org>]

------- Additional Comments From mark at codesourcery dot com  2004-06-24 01:54 -------
Subject: Re:  [3.3/3.4 Regression] [tree-ssa]
 Variable-length arrays broke in gcc 3.x

Giovanni Bajo wrote:

>reichelt at gcc dot gnu dot org wrote:
>
>  
>
>>But there are two open questions:
>>
>>a) Will this be fixed on the 3.3 and/or 3.4 branch?
>>    
>>
>
>There is absolutely no way it will, in my opinion. Mark, do you confirm that a
>fix for this on any release branches would be very intrusive and inappropriate?
>  
>
I think it's unlikely that a sufficiently simple patch can be found.

>>b) Are we going to have a testcase in the testsuite?
>>    
>>
>
>Yes, why not. People tend to not test Ada, so it might help. Can you please
>take care of this?
>  
>
People have been discussing whether or not this should be valid GNU C.

Kenner things it should be; Nathan and RTH think it should not.  If you 
add a test case for this feature, you should (a) first get consensus 
that we really want the feature, and then (b) add documentation that 
explains that is legal.  Despite Kenner's claims, this does not follow 
from the current GNU C docs, which talk only about automatic arrays -- 
not arrays that are part of other structures.  Even though C99 supports 
VLAs, it does not support the kind of code in the test case.



-- 


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13741