Hacker News Reader

Comment by weberc2 | Hacker News Reader

Good to know, but 1-2 seconds is still very slow compared to a native app. I have an 8MB Go CLI application that can startup and print its help in .05 seconds, and even that is probably slow compared to a C or Rust equivalent.

View on HN

Any savy Java developer knows how to AOT compile to native code, if that really matters.

There are plenty of JDKs that support it, and even Oracle is finally adding support for it with Java 9, initially only for Linux x64.

By Java 10 timeframe no one that only knows about OpenJDK and not the several other JDKs can state that Java starts slow vs C, because both will be AOT compiled to native code, loading native .so files.

Good point, but I wonder what size will the final binaries have, given that bytecode inflates significantly when translated to the machine code (there was a good discussion on HN recently about app sizes)

In any case smaller than Electron apps.

Electron apps aren't really my standard for "reasonably sized CLI applications"...

From what I've seen Java/JVM AOT won't produce binary executables but shared libraries that will be loaded on startup by the JVM. When I first read about Java AOT-compilation I expected it to produce real executables that's why I think this is important to mention. But please correct me if I am wrong. So if this is true even with AOT compilation you still need the JVM as dependency (although I think you can get a minimal JVM in under 10MB). It also seems that the AOT compiled code needs to be recompiled on Java updates. Although I quite like Java and the JVM, I still wouldn't choose Java for writing simple command line tools.

Nevertheless it is quite useful for improving startup for tools like Gradle or server applications. But sure, I also don't expect Java startup performance to be a problem in the mid- to long-run.

Correct, but that is no different than having to ship glibc.so or MSVCRT.dll alongside your application, specially if you want to be certain which version gets used.

This is only relevant to Oracle's implementation though.

Other JDKs do support static linking, as long as you restrain yourself from features that could require dynamic compilation.

There was a talk at JVM Languages Summit about the current state of AOT compilation regarding Java 10.

https://www.youtube.com/watch?v=n5DCg6M2MDM&index=11&list=PL...

I would rather ship glibc (8MB) or musl (.5MB) than the JVM (100s of MBs?), but it's not necessary given that nearly every target already has a libc. Also, Go programs don't depend on libc, and C and Rust can probably statically link against some libc in a pinch.

Rust can use musl, yes.

Well, printing stuff is easy. Just tried:

  $ time java HelloWorld 
  Hello, World

  real	0m0.088s
  user	0m0.088s
  sys	0m0.016s

That's presumably not loading any libraries at all though - the time to being interactive (or just doing something useful) will get worse as more bytecode has to be decompressed and loaded.

> That's presumably not loading any libraries at all though

Sure it is loading libraries, everything that System.out.println() depends on.

Reply on news.ycombinator.com