Does Java 8 delay mean less app developer feedback?

[Danno Ferrin]

On Thu, May 2, 2013 at 12:51 PM, Richard Bair <richard.bair at oracle.com>wrote:

>
> On May 2, 2013, at 10:05 AM, Mario Torre <neugens at redhat.com> wrote:
>
> > On Thu, 2013-05-02 at 08:56 -0700, Richard Bair wrote:
> >> I don't think Danno was making light of it at all, but rather was
> looking at "how can we solve this problem without waiting for Oracle"?
> which is pretty dang reasonable IMO.
> >>
> >> 1) Once OpenJFX is fully opensourced, could the v8 targeted branch be
> compiled to run on JDK 7?
> >>
> >> So this question is basically asking, when OpenJFX is fully open, can
> it be built and targeted at 7? In other words, maybe for your produce you
> could use OpenJDK and OpenJFX instead of Oracle's provided JavaFX 2.x? If
> this worked, then you could use the latest FX with an older JRE.
> >
> > I'm of course very interested in 1) myself.
>
> I'm interested, why is that? OpenJDK 8 is available under the normal
> friendly terms (I think?) so what is wrong with OpenJFX 8 on OpenJDK 8?
>

There are some organizations where getting a commercially produced JVM
deployed is easier than a community built one.  Stuff like support
contracts and security paranoia (who do I pay to fix a jvm halt for
OpenJDK8?)  There are less gates to arbitrary libraries, so OpenJFX8 on
Oracle Java 7 is something that can be pitched that won't be dismissed.  If
I pitched shipping OpenJDK8 at my day job there would be quite a pause
while my manager waits for the punchline, because surely I wouldn't be
serious.  In my case there are features we could ship 6 months earlier than
waiting for JDK 8 to be blessed.

On the hobbyist front there is also the issue of alternative VMs (GCJ,
Avian, IKVM, Dalvik, etc).  Very few of those support InvokeDynamic, which
is how Lambdas work.  Stripping out stuff like SwingNode and Printing is
fine, but I would like to put the idea out that "lambdas because we can"
would hurt on that front because the needed patches to get it compiling on
a version 7 based VM would become very large.  When need for bug fixes or
sensible APIs that's great, but if you really want a community iOS port
then we need to proceed with caution on the lambda front.