On Tue, 16 Aug 2022 23:15:38 GMT, SWinxy <[email protected]> wrote:

>> ScientificWare has updated the pull request incrementally with one 
>> additional commit since the last revision:
>> 
>>   Removes whitespace.
>>   
>>   Remove a whitespace at the end of the file
>
> That's a bit sus for the enhanced switch statement to be significantly slower 
> (7.5x more time than the map). Your methodology still shows that you aren't 
> creating new objects for each invocation and returning the same objects each 
> time. Is that accurate to your performance table?

Yes @SWinxy, you're right. Map and TreeMap implementations return the same 
object. Therefore comparisons with the current implemention or switch case 
solution are not relevant.  I going to add a test result with Map + your 
workaround.

Presently, the switch case solution is the best in respect of the current 
behaviors you want to preserve.

I integrated all your comments but yesterday, I was busy with the test case and 
other propositions that could content you. I take time to found the best 
solution because there are other places in swing html package where such 
optimisation would be interresting, especially in view factories.

-------------

PR: https://git.openjdk.org/jdk/pull/9825

Reply via email to