RFR: 8339538: Wrong timeout computations in DnsClient [v5]

[Daniel Fuchs]

On Tue, 10 Sep 2024 09:49:08 GMT, Mark Sheppard <msheppar at openjdk.org> wrote:

>> Aleksei Efimov has updated the pull request incrementally with one additional commit since the last revision:
>> 
>>   Measure time the caller spent waiting. Simplify timeoutLeft computation
>
> test/jdk/com/sun/jndi/dns/ConfigTests/Timeout.java line 112:
> 
>> 110:             // Check that elapsed time is as long as expected, and
>> 111:             // not more than 67% greater. Given the min DNS timeout
>> 112:             // correction above the threshold value is equal to 61%.
> 
> this is a bit arcane, why not have a simple measure that elapsed time shouldn't be more than twice the expected timeout ... this is not that different to the  multipliedBy(2) and multipliedBy(3) --  elaspedTime.compareTo(expectedTime.multipliedBy(2) <= 0
> 
> Additionally based on the internal minimum timeout allowance of 50 secs and this upper bound calculation, it would suggest that an @implNote might be useful, or required, to alert developers  to potential timeout variability, and not to rely on timeout to be absolutely (real time) precise

I don't think we want to go down the rabbit hole of documenting too much. Agreed that using a simple factor 2 would make the code simpler, but do we want to go that high?

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

PR Review Comment: https://git.openjdk.org/jdk/pull/20892#discussion_r1752155232