#62 "Invalid opcode 196" exception

[Eugene Kuleshov]

Attached test throws the following exception. Please
note that code being compiled is using ASM toolkit from
CVS HEAD. http://asm.objectweb.org/

java.lang.RuntimeException: dump(): Invalid opcode 196
at offset 36720

at net.janino.CodeContext.flowAnalysis\(Unknown Source\)

at net.janino.CodeContext.flowAnalysis\(Unknown Source\)

at net.janino.Java$FunctionDeclarator.compile\(Unknown

Source)

at net.janino.Java$ClassDeclaration.compile\(Unknown

Source)

at net.janino.Java$CompilationUnit.compile\(Unknown Source\)

at net.janino.Java$CompilationUnit.compile\(Unknown Source\)

at

org.objectweb.asm.attrs.JaninoTest.main(JaninoTest.java:5217)

Exception in thread "main"

[Eugene Kuleshov]

[Eugene Kuleshov]

Logged In: YES
user_id=158797

I guess the problem is related to wide offset (196 opcode)...

[Arno Unkrig]

Logged In: YES
user_id=865893

I can't reproduce this one... On JRE 1.4, I get stack overflow
errors due to the 5000+ string concatenations, and on JRE
1.2.2 I get UTFDataFormatException due to the string
constant longer than 64K.

What exactly is your environment? (JANINO version, JDK
version, etc.)
I renamed your class to "Bug_1032563" and ran:

java net.janino.Compiler -verbose -g -rebuild
Bug_1032563.java

[Eugene Kuleshov]

Logged In: YES
user_id=158797

I've picked up Janino 2.0.15 and still can reproduce this bug.

[Arno Unkrig]

Logged In: YES
user_id=865893

Hi Eugene,

again, what is your environment (JDK, OS, etc.)?

[Eugene Kuleshov]

Logged In: YES
user_id=158797

It is windows, JRE 1.4, but issue is totally generic.

Janino just doesn not handle large offsets properly in large
methods. est case is already attached to this bug.

[Arno Unkrig]

Logged In: YES
user_id=865893

JANINO now unrolls the string concatenation properly and
reduces it to one string constant. That constant, however,
is larger than 64K and is thus split up into several 64K
string constants which are concatenated at run-time.

[Eugene Kuleshov]

Logged In: YES
user_id=158797

It seems that problem is still in 2.0.16. I'll try 2.0.17
shortly. Let me know if you need more information.

[Eugene Kuleshov]

Logged In: YES
user_id=158797

Just checked. Bug is in 2.0.17 too. Here is

Parser p = new Parser(new Scanner(name, new
StringReader(source)));
return p.parseCompilationUnit().compile(CL,
DebuggingInformation.ALL)[0].toByteArray();

See an attachement for testcase. javac from JDK 1.4 can
compile it without problem. Source has dependency on ASM
toolkit ver. 2.0 (built from CVS HEAD).

[Eugene Kuleshov]

Source file that Janino failing to compile

[Arno Unkrig]

Logged In: YES
user_id=865893

Migrated this ticket to

http://jira.codehaus.org/browse/JANINO-2