Re: Soot bugs when first Unit is a target

[Chris Pickett <chris.pickett@mail.mcgill.ca>]

Chris Pickett wrote:
> Ondrej LHOTAK wrote:
>
> > On Mon, Jul 19, 2004 at 08:22:29PM -0400, Chris Pickett wrote:
> >
> >
> > > So, if I understand correctly, you're saying each entry/exit is a 
> > > head/tail, but not vice versa.  Can you give an example?
> >
> >
> > A head/tail is a node with no predecessor/successor. These shouldn't
> > appear in verifiable bytecode, but they can appear temporarily in Jimple
> > in between optimizations (before the unreachable code eliminator runs).
>
>
> Okay.
>
>
> > That brings up an interesting question, though. In a brief unit graph,
> > the beginnings of exception handlers are heads. Are they, however entry
> > points (or successors of the single entry node, if we choose to make
> > one)?
>
>
> I think if you can get there from outside the method, then it's an entry 
> point.  Maybe you could look at the callgraph to stop some or all of 
> them from being entry points.

Sorry, nevermind.  I think that interprocedurally the start of an 
exception handler is more equivalent to the successor of an invoke than 
an entry point.  So I guess the question is whether it's useful to 
include them intraprocedurally as entry points ...

Chris