Should back-ports cc the mailing lists for the area? (was Re: [7u8] Request for approval: 7113017: Use POSIX compliant include file headers in sun/awt/medialib/mlib_types.h)

[Andrew Hughes]

----- Original Message -----
> 
> Phil's comment about the back-port only being discussed on jdk7u-dev
> is
> probably worth further discussion.
> 
> One thing about jdk7u is that the barrier to get changes in is very
> low.
> On one hand this is is wonderful as it means we can get fixes into
> jdk7u
> in a timely manner and also keep the code mostly in sync with jdk8.
> On
> the other hand it allows changes to go into jdk7u before they have
> been
> tested or had any real usage in jdk8. I can point to several fixes in
> jdk8 that had side effects that only came to light after a few weeks.
> One would hope that if something creeps into jdk7u that it would be
> found in a timely manner, it's just that sometimes (not always) it is
> better to wait a bit before rushing a change into jdk7u.
> 
> I think Phil's suggestion is that back-port requests should cc the
> mailing list for the area. That would at least allow the original
> author
> and reviewers a chance to jump in if they wish. I think this
> suggestion
> make sense as it improves the chances of the "domain experts" seeing
> it.
> 
> -Alan.
> 

When doing the same with OpenJDK 6 & 7, we had an informal policy that changes
had to have "soaked" in 7 for some time before backporting.  I have myself
been amazed by how quickly fairly major changes have been backported, especially
when we've then later found flaws in them.  For example, the changes in building
HotSpot with debugging support were a fairly major change that was backported to 7
before we even saw it in 8.  We've since found multiple problems (loss of debug
data altogether, as the build started stripping it itself, then it being in the
wrong format) that affect our builds.

I don't know if things would have been different if it had taken longer for these
things to enter 7, but I do think that less obvious changes should take a little time.
Otherwise, what's the point in having two reviews in the first place?
-- 
Andrew :)

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

PGP Key: 248BDC07 (https://keys.indymedia.org/)
Fingerprint = EC5A 1F5E C0AD 1D15 8F1F  8F91 3B96 A578 248B DC07