Re: [Sablevm-developer] Does SableVM actually work? (on XYZ arch)

[Prof. E. M. G. <eti...@uq...>]

Grzegorz B. Prokopski wrote:
> The problem is that I don't think running HelloWorld is enough to be
> sure that SableVM is actually properly ported to every single arch.

An extensive test suite would really be welcome. Such a test suite would cover 
all bytecode instructions (in fact, it should cover all SableVM internal 
instruction codes), and should eventually also test things like vm<->library 
interface, etc.  I am planning to get some of this work done through future 
internships by undergraduate students, but it will take some time to get it 
done.  Getting "mauve" to test SableVM would be a first step toward building 
this bigger suite (which would include mauve as a subcomponent).

But on the short term, I think HelloWorld is pretty good because (probably 
better than you think):

1- it triggers the loading and initialization of approximately 150 classes,

2- it causes the execution of a significant amount of code, using most bytecode 
instructions,

3- it uses the JNI interface (native->Java & Java->native) [good for making sure 
libffi works],

4- it even tests synchronization primitives, even though it does not spawn 
multiple threads, because many class library methods are either synchronized, or 
they contain explicit synchronized(obj){...} blocks.

> 1. Write simple java program especially targetted at testing SableVM's
> architecture specific pieces, I mean: ...
 > I don't have enough knowledge of
> Java to write such program myself...

It is probably a good time to go to the book store and buy one of the countless 
Java books. ;-)

> 2. ... I belive SableCC uses many threads

Unfortunately it doesn't.  Soot does, but it is quite a big piece of software, 
so I'm not sure you want to include it as a simple test in a build process.

I really think that HelloWorld ought to be enough to include into a:
   $ ./configure ; make ; make test ; make install
build precedure.


Have fun!

Etienne

-- 
Etienne M. Gagnon, Ph.D.             http://www.info.uqam.ca/~egagnon/
SableVM:                                       http://www.sablevm.org/
SableCC:                                       http://www.sablecc.org/