RFA: 8073139: PPC64: User-visible arch directory and os.arch value on ppc64le cause issues with Java tooling

[Andrew Hughes]

On Thu, 10 Jan 2019 at 09:40, Severin Gehwolf <sgehwolf at redhat.com> wrote:

snip...

> > Why was this commit credited as 'sgehwolf' rather than the original
> > author, 'asmundak'?
> >
> > https://hg.openjdk.java.net/jdk8u/jdk8u-dev/rev/a5b23c21a665
>
> Because it's a rewrite of the 9 changeset to make it relevant for 8u?
> It made sense to me at the time.
>
> Take build changes for example. The build system between 9+ and 8 is
> very different. I personally wouldn't expect attributions for the 9+
> changes to be present for 8 build changes for the same bug.
>
> Question is where to draw the line :)
>

Some things are moved around in 9, mainly due to the HotSpot build being
brought fully into the configure/make system used by the rest of the build.
I would not go as far as to say they are "very different", especially not when
compared with the gap between 7 and 8.

I would expect attributions to be carried across, because there is little to no
new work involved; it is a translation of existing work to a new context. If
you translated a novel to a different language, you would not take on the
role of its primary author. It is the original author who has investigated the
problem and come up with the solution. I can see a case for jcheck
supporting a 'Backported-by' credit, but if anyone is to lose credit in the
current situation, it should be the backporter.

I say this having backported hundreds of changesets and not taking sole
credit for any of them. Nearly every changeset in OpenJDK 7 for the last
few years would be credited to me if that was the case.

I find the case for changing the authorship of this particular patch
particularly
weak, because I developed the original solution on 8u and so my 8u version
has also been around for some time. So I fail to see what you contributed to
be worthy of claiming sole authorship.

> > It's also missing the original Contributed-by line:
> >
> > 'Contributed-by: Andrew Hughes <gnu.andrew at redhat.com>, Alexander
> > Smundak <asmundak at google.com>'
> >
> > https://hg.openjdk.java.net/jdk9/jdk9/jdk/rev/2ff471390a03
> >
> > Can we please make sure to retain the original authorship when making
> > such backports in future, please?
>
> Yes and no. If the patch were a clean backport from 9 then it would
> make sense. Otherwise it's debatable which attributions need to be
> present. Are there good guidelines for this sort of thing?

I'm not aware of any. The limited guidelines on the OpenJDK website
could certainly use some improvement. The same issue came up when
I was discussing a ppc backport with Gustavo recently.

I tend towards the addition of a 'Backported-by' credit in similar manner
to 'Reviewed-by'.  As stated above, I don't see how attributing the original
author is in any way debatable; to do so is to suggest that one has created
something from scratch when this is not the case.

>
> Thanks,
> Severin
>

Thanks,
--
Andrew :)

Senior Free Java Software Engineer
Red Hat, Inc. (http://www.redhat.com)

Web Site: http://fuseyism.com
Twitter: https://twitter.com/gnu_andrew_java
PGP Key: ed25519/0xCFDA0F9B35964222 (hkp://keys.gnupg.net)
Fingerprint = 5132 579D D154 0ED2 3E04  C5A0 CFDA 0F9B 3596 4222