What to do about sun.misc.Unsafe (docs.google.com)

You know what is unsafe? Installing java. Whether it is applet attacks on the client, timezone attacks on the server, it is just a gaping big glory hole on the computer ready for other people to insert their bits (of code) to exploit for fun and profit. Friends don't let friends install java. The biggest shame is that the apache project keeps pushing so much crappy java code out there, that this dinosaur persists.

Is this doc editable ? someone deleted the contents :-( Better to pdf it.

I'm very disappointed with Oracle's decision to remove Unsafe. What are they trying to accomplish? They aren't serving anyone by removing it. Who benefits from its removal? What actual user of the JVM or customer of Oracle has ever said "I really like this 10x performance increase from this library that uses Unsafe, but I can't sleep at night thinking my program could seg-fault at any time if something went wrong! Please get rid of Unsafe Oracle! I promise to buy 10x more hardware from you to make up for the difference!"

Java's "write once run anywhere at all costs" philosophy, combined with "be safe at all costs", have kept the JVM locked into server-side Linux applications for a long time, while C# has continued to expand in areas like the mobile space. If it weren't for Java being propped up by Google, Java would hardly exist in the mobile world at all.

C# + the CLR has world class native code integration and escape hatches to drop to unsafe code when performance matters. Java's native integration is absolutely terrible, but at least it had Unsafe, which provided a pragmatic tool for achieving excellent performance on the JVM in a number of scenarios.

Now, look at this comment from an Oracle employee: http://mail.openjdk.java.net/pipermail/openjfx-dev/2015-Apri...

"Let me be blunt -- sun.misc.Unsafe must die in a fire. It is -- wait for it -- Unsafe. It must go..."

That's some very strong language backed by a simple, ridiculous reason; it's "unsafe". I honestly hope this Oracle employee's ideas die in a fire because they are out of touch with the needs of actual Java programmers, but my hope for that is slim at this point.

This would be a terrible move and will retard the uptake of Java 9. Why not leave it as it is or just repackage it until replacements snake their way through the JSR process and then users can migrate their source.

Java will never bridge the performance gap with C/C++/etc. until it plays nice with modern architectures where being memory cache friendly is vital for performance. Unsafe has been shown to allow orders of magnitude speed improvements for certain tasks.

While I am on the subject could someone explain the point of Jigsaw? It will deliver nothing to current users and why bother with a cutdown Java on small devices these days when the cheapest phone has more power than desktops of 10 years ago? It seems to have consumed enormous resources and caused delays while Java is slipping further behind C# and other modern languages.