[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: transformation declares java.lang.Object as super of java.lang.Object
Just to follow up on this: Clark sent me a version of Coffi that writes
0 into the super_class field if you tell it to. Coffi reads unsootified
and sootified classfiles just fine, reporting 0 and some index into the
constant pool (e.g. 21) for the super_class field in java.lang.Object
respectively. SableVM also reads the super_class field properly.
Soot (with and without '-f jasmin') writes a non-zero super_class value,
and so it looks like the bug in Soot is after Coffi but before the phase
that produces jasmin input (I am not sure if this phase is skipped if
'-f jasmin' is not specified). If I remove the '.super
java.lang.Object' line from the produced jasmin input, jasmin refuses to
compile it. It is strange that D-Java has the same bug -- perhaps it is
a limitation of the jasmin grammar.
In any case, I am not affected by this problem any longer, but it still
needs to be fixed.
Cheers,
Chris