On Wed, 24 May 2023 12:35:05 GMT, Daniel Fuchs <[email protected]> wrote:
>> test/jdk/java/net/httpclient/ManyRequests.java line 274:
>>
>>> 272: } catch (CompletionException e) {
>>> 273: if (!Platform.isWindows()) throw e;
>>> 274: if (LIMIT.get() < MAX_LIMIT) throw e;
>>
>> Previously, `LIMIT.get()` was being compared against `MAX_COUNT`, because
>> `REQUEST` was initialized (as `final`) to that value. Now it's being
>> compared to `MAX_LIMIT`. Is this intentional?
>
> Yes. I think the previous behaviour was buggy. The idea is to reduce
> concurrency if the underlying OS has trouble with too many concurrent
> connections.
What it says here is that if we already reduced the LIMIT once and it fails
again we should fail.
-------------
PR Review Comment: https://git.openjdk.org/jdk/pull/14038#discussion_r1204053958
