NagyDonat wrote:

> This is a semantic change.

Nice euphemism for "bug" :sweat_smile: Old behavior restored in 
https://github.com/llvm/llvm-project/pull/210774/commits/fd86bcdef6119cc5f1736b903e061c9c46b4d4a3.
 I really should have added a test for this zero-size element case earlier.

I also added a FIXME to avoid reporting this case (accessing the 6th element in 
an array of 10 zero-sized elements), I intend to fix this in a non-NFC 
follow-up PR. @steakhal Do you agree that this is a false positive and 
shouldn't be reported -- or do you know about some quirk of the standard that 
justifies reporting it?

I also added 
https://github.com/llvm/llvm-project/pull/210774/commits/947d88db187f90dec04176fa391e3bc3ff3c9e37
 which should _finally_ fix the CI which was still broken after my earlier 
patch.

https://github.com/llvm/llvm-project/pull/210774
_______________________________________________
cfe-commits mailing list
[email protected]
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits

Reply via email to