[Soot-list] important: vote on raising Soot compilation level to Java 5

Eric Bodden eric.bodden at mail.mcgill.ca
Wed Feb 7 18:10:39 EST 2007


Hi everybody.

Recently there has been some discussion on when would be the
appropriate time to migrate the source code of Soot to Java 5 code
compliance. Moving to Java 5 would for us Soot maintainers obviously
have several benefits with respect to code maintenance, stronger
typing guarantees and so on. However, we also believe that for Soot
extenders, there would be benefits, more or less for the same reasons.
Generics not only allow for stronger typing guarantees but also are a
form of code documentation and allow for fault avoidance that is not
possible without them.

On the other hand, as it stands now, Soot uses Java 1.4 compliant
source code which is compiled down to Java 1.2 (!) bytecode,
originally for the benefit of some Soot users which I do not recall at
the moment. Java 5 source code however, can to my best knowledge only
be compiled intp Java 5 bytecode. (I know about "retroweaver" but it
has some limitations.)

Consequently, moving the codebase to Java 5 would severly impact all
projects depending on Soot. Hence, we would like to do a little survey
and see if there are any objections from our users. We believe that by
now basically everybody should be using Java 5 compiler and runtime or
at least should be able to use one. Note that even if you have
extensions of Soot, written in Java 5, they would still work. The only
change is that you would have to run this extended version of Soot on
a Java 5 JVM.

So if any of you have objections and would require/like us to maintain
a Java 1.4 compliant version of Soot, please let us know!

If I do not receive any objections via this mailing list within the
next month or so, I will consider making the next major Soot release
Java 5 compliant.

Cheers,
Eric (wondering what changes Java 6 actually brought along)

-- 
Eric Bodden
Sable Research Group
McGill University, Montréal, Canada


More information about the Soot-list mailing list