[Soot-list] [Fwd: Question about Soot]

[Ondrej Lhotak]

Cyrille,

I don't think I quite understand your questions. As far as I know,
jsr/ret instructions in bytecode presented to Soot are inlined
immediately when the bytecode is read in, in the Coffi module.
In the rest of Soot, there are no longer any jsr/ret instructions
to deal with.

> for completeness in a paper I am publishing, I have a question about Soot:
> I have found papers where the structure of bytecode subroutines 
> (jsr/ret) is analyzed for decompilation (Dava). Probably this also 
> applies to transformation into a different format such as Jimple.

I am not a Dava expert, and I am not aware of such an analysis. Could you
be more specific about where you saw it?

> However, I am not sure whether Soot actually performs inlining in a way 
> that generates again original Java bytecode without subroutines (or, 
> more precisely, jsr/ret instructions)? If so, which paper or tech. 
> report describes that?

I don't understand this question. The original bytecode *does* have
jsr/ret instructions. The code processed within Soot, and the bytecode
output by Soot, do not have jsr/ret instructions.

Ondrej