4 ms·
That's the worrying thing to me. Oracle's business model seems to be to keep prices reasonable just as long as it takes to build up a customer base, and then st
by mumblemumble 6y ago
That's the worrying thing to me. Oracle's business model seems to be to keep prices reasonable just as long as it takes to build up a customer base, and then start squeezing.
I'm fascinated by GraalVM, but I'm hesitant to take even one step down a path that leads to Oracle using my actual software implementation platform to shake me down for money. I'd actually rather take the development effort and performance hit on using gRPC or similar to get my Java and Ruby modules talking to each other than to get locked into a potentially weaponized version of the Java platform.
- qorrect 6y ago> weaponized version of the Java platform That was really sad to read. Oracle is really hurting the java language and ecosystem.
- mumblemumble 6y agoRealistically, this isn't super new. After what happened to Dalvik 10 years back, and later to Adopt, one could argue that the Java Community Process itself, and therefore the very soul of Java, has been weaponized.
- thu2111 6y agoMost of GraalVM is open source. If anything, their biggest commercial error may be open sourcing too much. GraalVM EE is quite expensive for what it adds over the open source versions.
- dividedbyzero 6y agoIs there a summary anywhere of what parts/features are and aren't open source?
- thu2111 6y agoBasically more performance and managed Sulong (blocks memory management errors in C/C++ you run on the JVM). And for native AOT compiled images, G1 garbage collector (which is itself open source but the integration isn't).
- mumblemumble 6y agoThe profile-guided optimization is a pretty big deal if you're concerned about performance. Based on the benchmarks I've seen, GraalVM Community tends to outperform JIT-compiled Java for applications where warm-up time is a significant portion of total run time, but JIT-compiled Java tends to outperform Graal, sometimes by a significant margin, for anything long-lived. Profile-guided optimization would theoretically close that gap.