[Soot-list] LOC in Dacapo-2006 benchmark
Richard Xiao
richardxx at cse.ust.hk
Wed Jan 30 03:50:45 EST 2013
Yes, as pointed out by Eric, if you add "-app" option, the classes that are
estimated to be the dependences for your given entry class are all counted
as application classes. However, this estimation is not equivalent to the
reachable methods computed by call graph constructor.
Regars,
Xiao
On Wed, Jan 30, 2013 at 4:25 PM, Zhoulai <zell08v at orange.fr> wrote:
> Thank you a lot, Xiao. Among others, I was doing a small benchmark with
> geometric encoding. Luckily the bad result is not yet submitted!
>
> But, the problem is now, why I counted 182 109 LOC for 'fop' benchmark
> using the same Soot options? I am running soot from Eclipse, but without
> that plugin. I run soot.Main.main with className dacapo.fop.Main. This
> file by itself should not contain 182109 lines, shouldn't it? I used -app
> option. The document of soot command line says,
>
> -app
> Run in application mode, processing all classes referenced by argument
> classes.
>
> How does Soot get all class referenced by argument class? By a call-graph
> I suppose? It is not sufficient to read Soot's document. One needs to
> guess:( Sorry I feel totally confused.
>
> So how can I count so few LOC for xalan, yet so many for fop, using the
> same soot options!!??
>
> Zell.
>
> p.s.
>
> Here is how I specify the class file name.
>
> ...
> String dacapoName=config.getProperty(Configuration.DACAPONAME);
> String
> JRE14="/System/Library/Frameworks/JavaVM.framework/Versions/1.4/Classes/jce.jar:/System/Library/Frameworks/JavaVM.framework/Versions/1.4/Classes/classes.jar";
>
> String sootClassPath=JRE14+":"+prePath+"/"+dacapoName+".jar"+":"
> + prePath+"/"+dacapoName+"-deps.jar";
>
> className="dacapo."+dacapoName+".Main";
>
>
> On Wed, Jan 30, 2013 at 9:02 AM, Richard Xiao <richardxx at cse.ust.hk>wrote:
>
>> Hi,
>>
>> Your code "SootClass c: Scene.v().getApplicationClasses()", that only
>> counts the methods manually provided in your soot command line.
>>
>> Please obtain the reachable methods from CallGraph.ReachableMethods()
>> function:
>> http://www.sable.mcgill.ca/soot/doc/soot/jimple/toolkits/callgraph/ReachableMethods.html
>>
>>
>> Regards,
>> Xiao
>>
>>
>> On Wed, Jan 30, 2013 at 3:51 PM, Zhoulai <zell08v at orange.fr> wrote:
>>
>>> Thank you for your quick reply.
>>>
>>> The codes for LOC counting is mixed in the project. But basically, it
>>> is computed iteratively (accumulated in TotalLoc below) the LOC of all the
>>> reachable methods from Main.
>>> //*** The computation of LOC is here//
>>> TotalLOC = 0;
>>> for (SootClass sc: class2profiles.keySet()){
>>>
>>> for (Profile pf: class2profiles.get(sc)){
>>> totalLOC+=pf.getBody().getUnits().size();
>>> }
>>> }
>>>
>>> //*** my analysis is basically here, which collects reachable method
>>> from Main,
>>> public MySceneTransformer(Map<SootClass, ArrayList<Profile>>
>>> class2profiles) {
>>> this.class2profiles=class2profiles;
>>>
>>> @Override
>>> protected void internalTransform(String phaseName, Map options) {
>>> public Map<SootClass, ArrayList<Profile>> class2profiles;
>>> for (SootClass c: Scene.v().getApplicationClasses()){
>>> ArrayList<Profile> profiles = new ArrayList<Profile>();
>>> for (SootMethod sm: c.getMethods()){
>>> ...
>>> if (!sm.isConcrete() || sm.getName().equals("<init>"))
>>> continue;
>>> ...
>>> profiles.add(sm);
>>> }
>>> class2profiles.put(c,profiles)
>>> }
>>>
>>> Any idea?
>>> Zell.
>>>
>>>
>>>
>>> On Wed, Jan 30, 2013 at 8:25 AM, Tony Yan <yan at cse.ohio-state.edu>wrote:
>>>
>>>> Hi,
>>>>
>>>> Can you show the code to count LOC?
>>>>
>>>> Thanks,
>>>> Tony
>>>>
>>>> On Wed, Jan 30, 2013 at 2:19 AM, Zhoulai <zell08v at orange.fr> wrote:
>>>> > Hello,
>>>> >
>>>> > Recently, I analyzed Dacapo-2006 using soot. I find the LOC that I
>>>> got for
>>>> > the benchmark is much less than the one in your 2006's oopsla paper.
>>>> >
>>>> > In particular, I find xalan has only 4786 LOC, hsqldb has 5634 LOC,
>>>> fop has
>>>> > 182109 LO . I am aware of issues related to whether to count LOC of
>>>> library
>>>> > class. But I am using the options that maximal LOC could have been
>>>> collected
>>>> > started from main class.
>>>> >
>>>> > The options I used include: -pp, -w, -app, and the following
>>>> >
>>>> > sootArgs.add("-include");
>>>> > sootArgs.add("org.apache.");
>>>> > sootArgs.add("-include");
>>>> > sootArgs.add("org.w3c.");
>>>> > sootArgs.add("-p");
>>>> > sootArgs.add("cg.spark");
>>>> > sootArgs.add("enabled:true");
>>>> > sootArgs.add("-p");
>>>> > sootArgs.add("cg.spark");
>>>> > sootArgs.add("on-fly-cg:true");
>>>> >
>>>> > Do you think it is normal to have 4786 LOC for xalan?
>>>> >
>>>> > Zell.
>>>> >
>>>> > _______________________________________________
>>>> > Soot-list mailing list
>>>> > Soot-list at sable.mcgill.ca
>>>> > http://mailman.cs.mcgill.ca/mailman/listinfo/soot-list
>>>> >
>>>>
>>>
>>>
>>> _______________________________________________
>>> Soot-list mailing list
>>> Soot-list at sable.mcgill.ca
>>> http://mailman.cs.mcgill.ca/mailman/listinfo/soot-list
>>>
>>>
>>
>>
>> --
>> Richard Xiao Xiao
>> PhD Student @ CSE @ Hong Kong University of Science and Technology
>> www.cse.ust.hk/~richardxx
>>
>> _______________________________________________
>> Soot-list mailing list
>> Soot-list at sable.mcgill.ca
>> http://mailman.cs.mcgill.ca/mailman/listinfo/soot-list
>>
>>
>
> _______________________________________________
> Soot-list mailing list
> Soot-list at sable.mcgill.ca
> http://mailman.cs.mcgill.ca/mailman/listinfo/soot-list
>
>
--
Richard Xiao Xiao
PhD Student @ CSE @ Hong Kong University of Science and Technology
www.cse.ust.hk/~richardxx
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mailman.cs.mcgill.ca/pipermail/soot-list/attachments/20130130/c8bc6709/attachment.html
More information about the Soot-list
mailing list