Even more problems with latest jdt

[cdietrich]

/home/jenkins/agent/workspace/xtext_main/org.eclipse.xtext.xbase.testing/src/org/eclipse/xtext/xbase/testing/InMemoryJavaCompiler.java:[207]
05:21:12 [ERROR] this.compilerOptions.inlineJsrBytecode = true;
05:21:12 [ERROR] ^^^^^^^^^^^^^^^^^
05:21:12 [ERROR] inlineJsrBytecode cannot be resolved or is not a field
05:21:12 [ERROR] 1 problem (1

[cdietrich]

@iloveeclipse what is the proper replacement?

[iloveeclipse]

Simply remove it, the option is enabled by default, it isn't option anymore.

[cdietrich]

of course now i see a ton of test errors i cannot reproduce locally.

[cdietrich]

Caused by: java.lang.Error: Unresolved compilation problem:
SANITY CHECK: Invalid opcode -88 at pc 162 for stackmap table attribute for method foo()

at foo.Test.foo(Test.java:7)
... 59 more

[cdietrich]

=> i assume we need to use another ton of reflection :(

[iloveeclipse]

Not sure why Xtext used inlineJsrBytecode at all, but that option is always on for JDT compilation with 4.33+ SDK, and before it was possible to switch it off for target < 1.8. I believe with very old targets (<1.4?) it shouldn't be used either.

See eclipse-jdt/eclipse.jdt.core#2759

[cdietrich]

xtext java 11 uses jdt 3.33

[iloveeclipse]

xtext java 11 uses jdt 3.33

You mean 4.33? The option there is always true independently what is set. In 4.34 the option is gone.

[cdietrich]

jdt core 3.33

[iloveeclipse]

OMG