RFR(S): 8165439: Convert Test_TempNewSymbol to GTest

[Kirill Zhaldybin]

David,

Thank you for reviewing the test!

On 06.09.2016 02:55, David Holmes wrote:
> Hi Kirill,
>
> On 6/09/2016 3:37 AM, Kirill Zhaldybin wrote:
>> Dear all,
>>
>> Could you please review this fix for 8165439?
>>
>> WebRev: 
>> http://cr.openjdk.java.net/~kzhaldyb/webrevs/JDK-8165439/webrev.00/
>> CR: https://bugs.openjdk.java.net/browse/JDK-8165439
>
> Conversion seems fine but I have to admit I'm baffled by the final test:
>
>  70   Symbol* xyz = SymbolTable::new_symbol("xyz", CATCH);
>   71   int xyzcount = xyz->refcount();
>   72   { // inner scope
>   73     TempNewSymbol s_inner = xyz;
>   74   }
>   75   ASSERT_EQ(xyz->refcount(), xyzcount - 1)
>   76           << "Should have been decremented by dtor in inner scope";
>
> this appears to expect a refcount of zero even though we still have 
> the original reference. I guess I have no idea how this TempNewSymbol 
> ref counting works :)
I do not understand why it was implemented this way too but:
1. It was in VM internal test I did not change this part
2. The test works according the TempNewSymbol code which does not 
increase reference count if TempNewSymbol was created using default 
c-tor and then assigned with Symbol.

I filed https://bugs.openjdk.java.net/browse/JDK-8165535 on this.
It would be great if someone who is familiar with this code evaluates 
this issue.

Thank you.

Regards, Kirill
>
> Cheers,
> David
>
>
>> Thank you.
>>
>> Regards, Kirill