From: Albert C. <sap...@gm...> - 2019-03-11 17:36:57
|
Thank you Frank for touching on #3! This was clarifying. Albert Missatge de fwi...@gm... <fwi...@gm...> del dia dl., 11 de març 2019 a les 17:03: > > Thanks very much for the offer to help! Others have given great > answers to (1) and (2), I can give a more detailed answer for (3): > > On Thu, Mar 7, 2019 at 6:40 AM Albert Cardona <sap...@gm...> wrote: > > Question 3: rpython (from pypy) used to have a jvm backend in its git > > tree. I assume jython developers have considered pursuing that route, > > and if not, why not? > I discussed this with the PyPy folks some years back. It's a very > interesting prospect but it has a major stumbling block: a big > advantage that Jython has is the ability to leverage the advances in > the underlying JVM. It would be very difficult to utilize these > advances in a JVM oriented PyPy because PyPy has its own technologies > in these areas deeply baked in. To take one major example, Jython uses > the garbage collection technologies provided by the JVM, where a JVM > PyPy implementation would use the PyPy developed garbage collection > mechanisms which would not be as effective as the JVM native version. > The expense of trying to route around PyPy's technology in these areas > was deemed to high (for these reasons I think work on the JVM PyPy has > been stalled for a very long time). > > -Frank |