[RFR] jdk9.0.4+12 + functionality

[Simon Nash]

On 17/04/2018 20:24, Bob Vandette wrote:
>> On Apr 17, 2018, at 10:00 AM, Andrew Dinn <adinn at redhat.com> wrote:
>>
>> On 17/04/18 13:26, Bob Vandette wrote:
>>>> On Apr 17, 2018, at 4:23 AM, Andrew Dinn <adinn at redhat.com> wrote:
>>>>
>>>> However, another reason to not go that route is that the AArch32 port
>>>> was derived from the AArch64 port which was, in turn, derived from the
>>>> x86 port. Whatever the provenance of the (32- or 64-bit) arm code it
>>>> differs in significant ways from the x86 code.
>>>
>>> The 32-bit ARM port has been around since the JDK 8 days, although
>>> the sources were maintained in the closed forest.  This port was not
>>> derived from the x86 port or the aarch64 port.  The aarch64 port was
>>> derived from the arm32 port in order to make it easier to maintain both ports.
>> I don't think you can be talking about he same thing as me here.
> 
> I was talking about the history of the Oracle developed 32/64 bit ARM port located in
> hotspot/src/cpu/arm/vm, which is now in open/src/hotspot/cpu/arm.
> 
>> Andrew Haley and I wrote almost all the existing AArch64 code. We based
>> it on the x86 code. We didn't have access to Oracle's 32 bit arm port
>> until it was essentially complete and integrated into JDK9.
>>
>> The code in the AArch32 project was also not derived from Oracle's code.
>> Development on that code started before Oracle's 32-bit arm code was opened.
>>
>>> The 32-bit ARM port has been well tested for many many release including
>>> leading up to the completion of JDK 9 and has been used in countless embedded
>>> products.  It was a last minute decision by Oracle to discontinue the delivery of
>>> binaries so the port is pretty up-to-date.
>> I don't disagree. However, I still stand by what I said: the changes
>> made in Oracle's open sourced 32-bit arm port are not going to help
>> those maintaining the AArch32 port as much as the changes made to x86
>> code will. You may have a different opinion over that but the history is
>> as I set out above.
> 
> I guess I didn’t think it was a foregone conclusion that someone would
> have to maintain the aarch32 when a completed project exists but
> perhaps you’re further along than I thought.  I hadn’t seen any contributions
> to the aarch32 port project on http://hg.openjdk.java.net/aarch32-port/jdk9 <http://hg.openjdk.java.net/aarch32-port/jdk9>
> in over 18 months and I thought there was a consensus that the community
> would adopt the one we provided.
> 
Speaking as a user of these ports, my requirement is for 32-bit and 64-bit
ARM ports that are supported by the community going forward.  The recent
discussion on "Supported platforms" on other OpenJDK lists has left me
unsure about which of these ports (if any) will be supported in JDK 10, 11
and beyond and will not suffer from "bit rot" (see [1]).

Simon

[1] http://mail.openjdk.java.net/pipermail/jdk-dev/2018-April/001051.html

> Bob.
> 
> 
> 
>> regards,
>>
>>
>> Andrew Dinn
>> -----------
>> Senior Principal Software Engineer
>> Red Hat UK Ltd
>> Registered in England and Wales under Company Registration No. 03798903
>> Directors: Michael Cunningham, Michael ("Mike") O'Neill, Eric Shander
> 
>