I think he dismisses JRuby was too quickly. It seems to me that JRuby
is ruby’s best hope for a non-sucky runtime. I’m looking forward to
Java 7, my hope is that it results in Ruby running so much better than
any other implementation that it becomes the obvious choice. Am I
being too optimistic with my expecations of invoke dynamic?
The current approach of writing C code to get speed out of Ruby is IMO
retarded. C code is brittle, many times platform specific, and much
harder to write than Java. And I say this as someone who finds
writing Java code irritating…
Am I being too optimistic with my expecations of invoke dynamic?
Aw no… you just need to hibernate for 2-3 years until Java 7 RC1 will
be released, but then it’ll all be just dandy.
(BTW Java 7, complete with JSR 277, invokedynamic and Closures was
supposed to be out last summer at least according
to Sun release mathematics, ie. a Java release every 18 months.
Considering there isn’t even an idea what’ll be in Java 7,
and vital bits keep on rotting, eg http://alblue.blogspot.com/2008/11/rise-and-fall-of-osgi-and-jsr277.html
… oh well…).
Oh: and even if we see an official Java 7 (with invokedynamic,
lightweight methods and everything else necessary to make JRuby really
fast) this decade (Prototypes and Research VMs don’t count), you’ll
spend another
2-3 years waiting for Apple to follow, Enterprises to even consider
considering, etc.
But to end on a positive note: by 2013, there’ll be 13 million Ruby
devs, so everything’ll be fine
Oh: and even if we see an official Java 7 (with invokedynamic, lightweight
methods and everything else necessary to make JRuby really fast) this decade
No reason it has to be “official”, or from sun. Just like IcedTea has
taken the OpenJDK and redistributed it, there is no reason there
couldn’t be an opensource release of some snapshot of Java 7, that is
just for running JRuby and Jython.
i.e. Treat the OpenJDK 7 JVM as analogous to the Ruby VM, or the
Rubinius VM - just don’t care about legacy java applications, or
enterprises that don’t want to move off java 1.3. Take a snapshot,
make sure it is stable, make it redistributable, and you have your
very own high performance, JITing, GCing, dynamic language VM without
having to write it from scratch.
Sure, building the OpenJDK is not as simple as pulling the source from
mercurial and doing ./configure && make && make install, but its not
impossible either, and a group who really wanted it done could get it
done.
Not that JRuby needed a bigger opening than it already has, but heh …
If JRuby continues on its current trajectory, most server deployments of
it will be running on the JVM it I’d imagine. But I’m severely biased.
Keep up the great work JRuby.
Thomas E Enebo wrote:
retarded. C code is brittle, many times platform specific, and much
He complains about the lack of momentum in Ruby which is super unfair
for us (including reducing memory use) and then just pulls the memory
card on us. We have made so much continual progress the last two
years that I still get excited daily thinking about the future of
JRuby. In fact, I think most people should be excited about where
JRuby is today. We keep getting faster and we REALLY do represent
what he wants: A fast continually improving Ruby implementation built
on top of the most sophisticated VM in the world. -server indeed.
-Tom
On Thu, Nov 20, 2008 at 10:44 PM, phil swenson [email protected]
wrote:
harder to write than Java. And I say this as someone who finds
writing Java code irritating…
Oh: and even if we see an official Java 7 (with invokedynamic, lightweight
methods and everything else necessary to make JRuby really fast) this decade
No reason it has to be “official”, or from sun. Just like IcedTea has
taken the OpenJDK and redistributed it, there is no reason there
couldn’t be an opensource release of some snapshot of Java 7, that is
just for running JRuby and Jython.
This is an extremely important point…the fact that you can build and
run this stuff yourself now means that the latest and greatest Ruby VM
(OpenJDK) is just a build away.
harder to write than Java. And I say this as someone who finds
writing Java code irritating…
Yes, he dismissed JRuby much too quickly. Honestly, I doubt he ever even
tried it. And he obviously doesn’t get why it represents a much better
path to “next-gen Ruby” than Rubinius or IronRuby. Plus, he doesn’t
allow comments, so he’s obviously just a talker, not a listener
Yes, he dismissed JRuby much too quickly. Honestly, I doubt he ever even
tried it. And he obviously doesn’t get why it represents a much better path
to “next-gen Ruby” than Rubinius or IronRuby. Plus, he doesn’t allow
comments, so he’s obviously just a talker, not a listener
True!
Eng Khaled al Habache
On Fri, Nov 21, 2008 at 5:05 PM, Charles Oliver N. <
He complains about the lack of momentum in Ruby which is super unfair
for us (including reducing memory use) and then just pulls the memory
card on us. We have made so much continual progress the last two
years that I still get excited daily thinking about the future of
JRuby.
In the last 8 months JRuby has changed from being 41% slower than MRI
1.8.6 to 54% faster running my simple XML benchmark in REXML.
REXML is all pure Ruby code so I’d say that’s a pretty impressive
rate of change for JRuby – an very impressive effort on the part of
the core developers!
Regarding the original blog post: any blog post which doesn’t allow
comments of some sort is much less interesting than it would be with
comments.
What is the deal with Java 7? Has sun started it? I listen to the
JavaPosse podcast and they interviewed a guy from sun who indicated it
would be out next year. But then I keep hearing Sun still hasn’t
decided on what’s in it? It seems to me that Sun is getting bogged
down in all these feature debates. perhaps they should re-think their
feature strategy and just add features one at a time in point
releases. It’s a shame that invoke dynamic (one feature) is being
held up by the dozens of other features on the list. Of course I’m
saying this as someone completely on the outside who doesn’t
understand how the JVM is built, so maybe I’m being simplistic.
There’s a process involved in defining “Java 7” that I don’t really
follow myself, but it’s been under heavy development since before Java 6
was even finalized. What I do know is that invokedynamic is officially
supposed to be in Java 7, so we’ve got a huge opportunity there.
What is the state of OpenJDK anyway? Does it have hotspot?
OpenJDK is HotSpot, plus Sun’s implementation of all the Java class
libraries. It’s the best, and it’s totally open.
What is the deal with Java 7? Has sun started it? I listen to the
JavaPosse podcast and they interviewed a guy from sun who indicated it
would be out next year. But then I keep hearing Sun still hasn’t
decided on what’s in it? It seems to me that Sun is getting bogged
down in all these feature debates. perhaps they should re-think their
feature strategy and just add features one at a time in point
releases. It’s a shame that invoke dynamic (one feature) is being
held up by the dozens of other features on the list. Of course I’m
saying this as someone completely on the outside who doesn’t
understand how the JVM is built, so maybe I’m being simplistic.
What is the state of OpenJDK anyway? Does it have hotspot?
that’s good news. So have they started invoke dynamic? Does this
mean invoke dynamic will be available via Open JDK before Java 7 is
out (sounds like it)?
OpenJDK is HotSpot, plus Sun’s implementation of all the Java class
libraries. It’s the best, and it’s totally open.
very good to keep