RFR (round2) JDK-8038654: Separate SymbolTable and StringTable code

[Coleen Phillimore]

On 5/6/14, 4:27 AM, Stefan Karlsson wrote:
>
> On 2014-05-05 23:07, Coleen Phillimore wrote:
>>
>> Hi,
>>
>> I don't see how this compiles without precompiled headers without 
>> including symbolTable.hpp - there are files with 
>> SymbolTable::unlink() so it must be included transitively (but I 
>> haven't found which file indirectly includes it yet).
>
> I took a look at the include tree for 
> concurrentMarkSweepGeneration.cpp (by passing -H to gcc), and 
> apparently classFileParser.hpp includes symbolTable.hpp.
Thanks for figuring it out Stefan.  I should remember the gcc -H option.

It doesn't look like classFileParser.hpp _should_ include symbolTable.hpp.
>
>> I think it's better not to include symbolTable.hpp transitively but 
>> to include it directly in the files that use functions in 
>> SymbolTable, ie all the GC files use SymbolTable::unlink.
>
> I agree with Coleen. Thanks for catching this.

Sorry Gerard for not bringing this up sooner.

Thanks,
Coleen
>
> StefanK
>
>> Actually, I think the set of files that use stringTable and not 
>> symbolTable are very small (or none).
>>
>> Coleen
>>
>> On 5/2/14, 12:57 PM, Gerard Ziemski wrote:
>>> hi all,
>>>
>>> Please review this simple enhancement, which refactors symbolTable 
>>> into symbolTable and stringTable
>>>
>>> The changes since rev0:
>>>
>>> Re: Stefan Karlsson
>>>
>>> - sorted includes
>>> - replaced "#include symbolTable.hpp" with just "#include 
>>> stringTable.hpp" as needed
>>> - stringTable.cpp no longer includes symbolTable.hpp
>>> - stringTable.hpp forward declares Symbol instead of including 
>>> symbolTable.hpp
>>>
>>>
>>> Webrev: http://cr.openjdk.java.net/~gziemski/8038654_rev1
>>> Bug: https://bugs.openjdk.java.net/browse/JDK-8038654
>>>
>>> Tested with hotspot jtreg, vm.quicklooks and JPRT. Locally builds on 
>>> Mac OS X and Linux.
>>>
>>> Thank you.
>>>
>>>
>>>
>>>
>>
>