OpenJDK11u: Backward incompatible behavior
Xuelei Fan
xuelei.fan at oracle.com
Wed Mar 11 15:53:33 UTC 2020
Hi Alexey,
I had run the testing for you, no surprise. Please commit to JDK 15,
and backport accordingly.
Thanks,
Xuelei
On 3/11/2020 7:16 AM, Alexey Bakhtin wrote:
> Hello Xuelei,
>
> Thank you for review.
> Can I commit it to JDK15 and create backports to JDK 14, 13 and 11 ?
>
> Thank you
> Alexey
>
>> On 10 Mar 2020, at 20:59, Xuelei Fan <xuelei.fan at oracle.com> wrote:
>>
>> Looks fine to me.
>>
>> Thanks,
>> Xuelei
>>
>> On 3/5/2020 8:50 AM, Alexey Bakhtin wrote:
>>> Hello Xuelei,
>>> I have renamed inputBuffer to recordBody.
>>> Also, as you suggested, recordBody is not removed but used for multiple records. So, it should be better for performance.
>>> JDK15 webrev: http://cr.openjdk.java.net/~dcherepanov/8239788/webrev.v5/
>>> Regards
>>> Alexey
>>>> On 4 Mar 2020, at 21:23, Xuelei Fan <xuelei.fan at oracle.com> wrote:
>>>>
>>>>> http://cr.openjdk.java.net/~bae/8239788/webrev.v4/
>>>>
>>>> SSLSocketInputRecord:
>>>> 54 // Cache for incomplete input record.
>>>> 55 private ByteBuffer inputBuffer = null;
>>>> This variable is used for record body, I may use a instinctive name, for example recordBody.
>>>>
>>>> Otherwise, looks good to me.
>>>>
>>>> I think, for performance, it may be possible to reuse this buffer for multiple records. I'd appreciate if you want to make an improvement in this update as well.
>>>>
>>>> Thanks,
>>>> Xuelei
>>>>
>
More information about the security-dev
mailing list