[aarch64-port-dev ] OpenJDK extension to AArch64 and Windows

Mario Torre neugens at redhat.com
Thu Jun 25 16:18:51 UTC 2020

On Thu, Jun 25, 2020 at 6:14 PM Andrew Dinn <adinn at redhat.com> wrote:
> On 25/06/2020 16:05, Andrew Haley wrote:
> > On 25/06/2020 15:49, Dalibor Topic wrote:
> >> Typically such efforts would become with a porting Project of their own,
> >> just as the aarch64-linux port did back in 2013 [1], as that allows a
> >> community of developers across organizations to start to collaborate
> >> around a shared repository, mailing list, etc. on a specific porting task.
> >>
> >> It also makes it easier to do so, by allowing the initial set of
> >> developers on the port to become committers on the new project from the
> >> start, without having to work their way up to 8 or more significant
> >> sponsored changes first [0]. So that's what I (with my Porting Group
> >> Lead hat on) would usually recommend that course for new ports.
> >
> > That would be madness.
> >
> > The Windows port to AArch64 is vastly simpler than the aarch64-linux
> > port by more than an order of magnitude, because the AArch64 back end
> > barely changes at all and the Windows (x86) part doesn't change very
> > much because it's mostly C++ code and the Windows APIs are pretty
> > stable.
> >
> > The code as submitted by Microsoft is almost ready to go. Introducing
> > a whole lot of process at this stage would be entirely inappropriate
> > and greatly increase the work for both HotSpot developers and the
> > Microsoft porters.
> I agree entirely with Andrew. If you look at the webrevs provided by
> Microsoft they are smaller than a great number of patches that have only
> been covered by a JIRA issue. The amount of code touched and the
> significance of the changes is far too small to merit a whole project
> (or, I would say, a JEP).
> To me this contribution is a relatively small extension of the work
> covered by the existing AArch64 port. I cannot see any reason to smother
> it in a bureaucratic process that is very unlikely to improve the
> outcome and, indeed, much more likely to achieve the opposite.

Yeah, I don't have a strong opinion on the JEP, but regarding the
project home also the AArch64 port webpage strongly hints that this is
the natural home for such ports (and I may say, if Apple really wants
to contribute an Arm port this is where they should be looking at too
as first option!):

"It is hoped that this project will eventually be able to support
operating systems other than GNU/Linux, and welcomes contributors with
the necessary expertise."

Mario Torre
Associate Manager, Software Engineering
Red Hat GmbH <https://www.redhat.com>
9704 A60C B4BE A8B8 0F30  9205 5D7E 4952 3F65 7898

More information about the hotspot-runtime-dev mailing list