RFR: 8013117: Thread-local trace_buffer has wrong type and name

[Rickard Bäckman]

Nils,

no it doesn't matter. Rather intended. By initializing it to NULL we forced implementors to use a pointer that would have to be initialized at some point. Now it can be a class / struct
that is instead initialized by a default constructor.

/R

On Apr 24, 2013, at 10:45 AM, Nils Loodin wrote:

> Does it matter that the pointer gets initialized to NULL before, but not now? There isn't any null checks anywhere that depends on that?
> 
> Regards,
> Nils
> 
> On 04/24/2013 09:51 AM, Rickard Bäckman wrote:
>> Hi all,
>> 
>> can I have a couple of reviews for this small change. The short story is that the current way the thread-local _trace_buffer is somewhat inflexible.
>> By changing the type of the getter this structure gets more flexible for different implementations. I also think that the name is misused. Just naming it
>> to _trace_data is more generic and less implementation-specific.
>> 
>> The webrev: http://cr.openjdk.java.net/~rbackman/8013117/
>> 
>> Thanks
>> /R
>> 
>