RFR (XS): 8017578 : Hotspot compilation error with latest Studio compiler

Christian Thalinger christian.thalinger at oracle.com
Mon Jul 8 14:29:23 PDT 2013


On Jul 8, 2013, at 12:44 PM, Vladimir Kozlov <vladimir.kozlov at oracle.com> wrote:

> Stupid C++ :(
> 
> Fine, we go with adding 'virtual'.

Yeah, let's do it.  -- Chris

> 
> Vladimir
> 
> On 7/8/13 11:10 AM, David Chase wrote:
>> So where do we stand on this?  I'm in favor of my change because it works and we don't care about the last scrap of adlc performance anyway.  I tried alternatives but they did not work.
>> 
>> David
>> 
>> On 2013-07-04, at 3:13 PM, David Chase <david.r.chase at oracle.com> wrote:
>> 
>>> 
>>> On 2013-07-02, at 5:31 PM, Vladimir Kozlov <vladimir.kozlov at oracle.com> wrote:
>>> 
>>>> I am fine to make base destructor virtual. My question is why we need these destructors at all if they are all empty?
>>> 
>>> Removing the destructors does not help; I tried that, and I still got the "polymorphic needs virtual destructor".
>>> I think the reasoning is a subclass with an interesting destructor could always be created,
>>> and if that happened then delete of a base class (with "no" destructor) would fail to run the subclass destructor
>>> and that could lead to leaks.
>>> 
>>> So I did try getting rid of destructors, (including Form, MatchNode, and MatchRule, as well as others in that hierarchy)
>>> and the message did not go away.
>>> 
>>> David
>>> 
>> 



More information about the hotspot-compiler-dev mailing list