Java binaries

[Russel Winder]

Sorry to come to this thread late, and apologies if I am missing the
real point to this thread.

On Sun, 2013-02-17 at 10:13 +0100, js.mdnq wrote:
> On Sunday, 17 February 2013 at 07:53:15 UTC, Walter Bright wrote:
> > On 2/16/2013 7:26 PM, js.mdnq wrote:
> >> Would it ever be possible to compile D source directly to java 
> >> to take advantage
> >> of what java offers. (e.g., the ability to run d code inside a 
> >> browser)

Running Java in the browser is now unlikely to ever happen again.  If
anything the target should be JavaScript or Dart, and even Dart would be
a huge risk.

> >> I'm not talking about necessarily fully fledged 
> >> functionality(obviously stuff
> >> like asm isn't going to work) but basically the ability to use 
> >> D's syntax and
> >> some of it's compiler features(mixins, templates, etc).

A lot of work is going into creating Python executability in browsers.
Skulpt, Brython, etc., but traction is minimal. Browsers mean HTML5 and
ECMAScript (aka JavaScript)

> >
> > Java doesn't have pointers, so right off the bat there'd be big 
> > problems.
> 
> Java does have pointers... you just can't get to them easily. At 
> the very least, you could allocate and manage your own chunk of 
> memory and do all the pointer arithmetic yourself on that chunk. 
> This would require the chunk to be pinned in some way and the GC 
> to be turned off but it is an option.

There is the possibility of using NIO2 buffers in ways that were never
envisaged. This is effectively what happened in RTSJ, which was a
real-time systems framework using Java and JVM. RTSJ sadly was too far
ahead of its time and so died. The last vestige will die when the Mars
Rover does.

> Also, have you looked much at sun.misc.Unsafe?

You really don't want to go there ;-)

Obviously for garbage collection and concurrency and parallelism control
you have to go there so as to subvert the JVM object model. The question
is that unless you are working on the G1 garbage collector or
java.util.concurrent primitives, is there any point?

> Or maybe one could use the JNI to write an external memory 
> manager for each platform and all pointer arithmetic is passed 
> through that. Of course, if one goes this far then any "javaD" 
> would need to be able to work well with importing java libraries 
> to be of any use.

Sounds like a mountain rather than a molehill. We are looking at
something vaguely analogous to get CUDA and OpenCL working from
GPars/Groovy/Java. Good for users, very ugly for implementors.

Unless there is a mapping of the semantics of a language to the
intermediate code of the platform, then it is better not to bash head on
wall but to go put energies into something more productive. If the D
semantics cannot be mapped down to the JVM bytecodes, then D is not a
language you want to run on the JVM. Given
Java/Scala/Kotlin/Ceylon/Groovy as the static languages and
Groovy/JRuby/Jython/Closure as the dynamic languages (yes Groovy is
correctly in both categories!) is there really a market for a minor
player native code language trying to ease itself onto the JVM platform?

The world is split into native code, PVM, JVM, JavaScript/ECMAScript. D
only really has a play in one of these, and needs to get real traction
there first before looking for new lands to conquer. Else it risks being
seen as a solution looking for a problem to solve.

-- 
Russel.
=============================================================================
Dr Russel Winder      t: +44 20 7585 2200   voip: sip:russel.winder at ekiga.net
41 Buckmaster Road    m: +44 7770 465 077   xmpp: russel at winder.org.uk
London SW11 1EN, UK   w: www.russel.org.uk  skype: russel_winder
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 198 bytes
Desc: This is a digitally signed message part
URL: <http://lists.puremagic.com/pipermail/digitalmars-d/attachments/20130217/ee8cd64c/attachment.pgp>