GCC 4.7 OpenJDK Build Failure

David Holmes david.holmes at oracle.com
Thu Nov 15 02:25:10 PST 2012


On 15/11/2012 7:26 PM, Florian Weimer wrote:
> On 11/15/2012 08:13 AM, Mikael Vidstedt wrote:
>
>> Agreed, and it would obviously be interesting to know out why the
>> original fix was lost.
>
> The using declarations were moved in this commit:
>
> changeset: 685df3c6f84b
> user: jmasa
> date: Tue Sep 18 23:35:42 2012 -0700
> summary: 7045397: NPG: Add freelists to class loader arenas.

Yes thanks - that was covered in the original thread:

http://mail.openjdk.java.net/pipermail/hotspot-runtime-dev/2012-November/004737.html

> This may have been due to a misunderstanding how these declarations work
> in this case. The effect on dependent name lookup does not propagate to
> templated derived classes if the base class is type-dependent. The
> this-> marker is more explicit, that's why I prefer it.

The reason I dislike it is that to me it look quite bizarre to have a 
series of invocations only some of which require this->  To the reader 
there is no obvious reason as to why it is needed.

BTW there are numerous using directives in the binaryTreeDictionary.hpp 
file, so now we have a mixed solution which isn't good for anyone's 
ability to understand the code. :(

Cheers,
David


More information about the hotspot-dev mailing list