[cfe-dev] isVariablyModifiedType() method on Type not consistent with the standard

[Chris Lattner]

On Feb 14, 2008, at 7:58 AM, Eli Friedman wrote:

> Per the standard, "int (*)[1.0]" is a variably modified type; however,
> isVariablyModifiedType() returns false because it only checks for
> vlas.  I don't want to just fix this, though, because that will make
> isVariablyModifiedType() and getAsVariablyModifiedType() inconsistent.
> Should I rename the existing isVariablyModifiedType and
> getAsVariablyModifiedType() to something different?
>
> Ultimately, I think we should split incomplete array types into a
> separate class from VariableArrayType... they are conceptually
> separate and splitting them will make these predicates less confusing.
> Any thoughts on that?

Both sound very reasonable to me!

-chris