Will Nashorn facilitate Gradle being used for more web application assembly and test tasks, without using Node?

[KARR, DAVID]

> -----Original Message-----
> From: Marcus Lagergren [mailto:lagergren at gmail.com]
> Sent: Friday, December 26, 2014 9:35 AM
> To: KARR, DAVID
> Cc: nashorn-dev at openjdk.java.net
> Subject: Re: Will Nashorn facilitate Gradle being used for more web
> application assembly and test tasks, without using Node?
> 
> Startup/warmup is one of our major things that we are concentrating on for
> 9. Results in the lab are really already quite impressive, I am happy to
> say.
> 
> In the meantime, with 8u40, you can use the code caching / optimistic type
> caching feature to serialize code to disk. This makes consecutive
> invocations of a script very fast.

Ok, now that we've determined that Nashorn startup is going to get better, can you try to address the original question?  Will Nashorn facilitate this?

> > On 25 Dec 2014, at 17:03, KARR, DAVID <dk068x at att.com> wrote:
> >
> > Gradle already mitigates the startup time problem with the Gradle
> Daemon, so typical client invocations will be faster than without it.
> >
> > From: Benjamin Sieffert [mailto:benjamin.sieffert at metrigo.de]
> > Sent: Thursday, December 25, 2014 7:44 AM
> > To: KARR, DAVID
> > Cc: nashorn-dev at openjdk.java.net
> > Subject: Re: Will Nashorn facilitate Gradle being used for more web
> application assembly and test tasks, without using Node?
> >
> > A problem with JVM-based components being used in a tool-like manner is
> the relatively huge cost associated with just starting the JVM.
> > E.g. if I start a JRuby-console ("irb"), it takes 2-3 seconds, whereas a
> MRI console basically opens instantly. Even with the work being put into
> reducing
> > nashorn's own warm-up time, the JVM's share will remain. An approach
> here seems to be to have one JVM running at all times and then relegating
> the
> > invocations of your tools to it. Nailgun is an implementation of this
> recommended by the JRuby team. But such an approach certainly brings its
> own
> > share of problems. In the end, I think that with the JIT, the JVM's very
> promise has always been that getting prime performance on it requires your
> > application to be a bit on the longer-lasting side. Otherwise, AOT
> optimized code will just be plain faster. And with most commandline tools,
> I feel,
> > invocation speed means a lot.
> >
> > On 24 December 2014 at 19:02, KARR, DAVID
> <dk068x at att.com<mailto:dk068x at att.com>> wrote:
> > Nashorn and Avatar are interesting to me, but not personally as much for
> the ability to write standalone or server-based JavaScript applications.
> >
> > I see Node.js being primarily used in two different ways.  It is used to
> write those standalone and server-based JavaScript applications, but it is
> also used entirely in the building and testing process of web
> applications, primarily with the Karma, Bower, and Grunt Node modules. I
> imagine there are other Node modules like that that are primarily used as
> a tool, not as a component in a custom Node.js application.
> >
> > You might consider this a "niche" application, but I'm primarily
> interested to see whether Nashorn will be able to make it easier to use
> Gradle to build web applications and run JavaScript unit tests, without
> involving Node.js.
> >
> >
> >
> > --
> > Benjamin Sieffert
> > metrigo GmbH
> > Sternstr. 106
> > 20357 Hamburg
> >
> > Geschäftsführer: Christian Müller, Tobias Schlottke, Philipp Westermeyer
> > Die Gesellschaft ist eingetragen beim Registergericht Hamburg
> > Nr. HRB 120447.