On Wed, 8 Mar 2023 11:29:42 GMT, Daniel Fuchs <[email protected]> wrote:

>> The HttpClient internal DebugLogger could be simplified if its configuration 
>> was held in a record. Some of the methods in Utils that return a debug 
>> logger could also be simplified/removed. The system property that configures 
>> the debug logging could be extended to choose between System.err, 
>> System.out, System.Logger, or any of these configurations.
>> Passing -Djdk.internal.httpclient.debug=true still works as before.
>
> Daniel Fuchs has updated the pull request incrementally with two additional 
> commits since the last revision:
> 
>  - Make sure julLogger is not GC'ed before the end of the test.
>  - Integrated review feedback

test/jdk/java/net/httpclient/DebugLoggerTest.java line 65:

> 63:  * @run main/othervm -Djdk.internal.httpclient.debug=true,log 
> DebugLoggerTest ERR LOG
> 64:  * @run main/othervm -Djdk.internal.httpclient.debug=true,out 
> DebugLoggerTest ERR OUT LOG
> 65:  * @run main/othervm -Djdk.internal.httpclient.debug=err,OUT,foo  
> DebugLoggerTest ERR OUT

Given that we launch 11 `@run`s in this test, I was curious how long the entire 
test run would take to complete. Timeout wasn't a concern (I didn't think there 
would be one), but I was curious if marking them as multiple `@test` runs to 
run them parallely would gain any performance improvements. But after having 
run this test locally in the current form, I see that the whole test completes 
in slightly over 2 seconds. So there's no point in thinking of any 
optimizations.

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

PR: https://git.openjdk.org/jdk/pull/12900

Reply via email to