[Soot-list] j2j inefficiency?

[Oege de Moor]

While running the jhotdraw benchmark for the abc OOPSLA
paper, I noticed that the pure Java version (without any
aspects) runs about 25% slower when compiled with abc
compared to other Java compilers. It appears to be a Soot
problem rather than an abc one...

The benchmark is accessible to abc-ers on musketeer at
Oxford under the tmbenches module. If any other Soot
maintainers are willing to take a look, let me know and
I'll send you a tarball.

Thanks,

-Oege