[kaffe] Re: darwin6 failures (stack size?)

[Riccardo]

I increased the stack size to 1000 K and it still fails

-R

In <20040926173329347+0200 at news.gmane.org> Riccardo  wrote:
> Hello,
> 
> when I compile on darwin6 the compilation ends with some zip errors. 
> In  reality, some stuff was not compiled at all, look here:
>
> [ parsed org/xml/sax/helpers/XMLReaderAdapter.java in 3,166 ms ]
> [ parsed org/xml/sax/helpers/XMLReaderFactory.java in 1,580 ms ]
> [ checked modifiers in 12,572 ms ]
> <GC: heap 150532K, total before 142169K, after 113056K (2561797/
> 2186696  objs)  24.9% free, allocated 36226K (#532936), marked 110091K, 
> swept 29113K (# 375101)  150 objs (4K) awaiting finalization> <GC: 
> heap 152580K, total before 144371K, after 116309K (2732226/2259591 
> objs)  23.8% free, allocated 36545K (#692100), marked 113300K, swept 
> 28062K (# 472635)  0 objs (0K) awaiting finalization> <GC: heap 
> 154628K, total before 145054K, after 120216K (2845499/2343299  objs)  
> 22.3% free, a  0 objs (0K) awaiting finalization> [ checked interfaces 
> in 569,086 ms ] java.lang.StackOverflowError    at at.dms.kjc.
> JBinaryExpression.dependsOnField (JBinaryExpression. java:74)    at at.
> dms.kjc.JBinaryExpression.dependsOnField (JBinaryExpression. java:74)
> <repeated a lot of times>
> (JBinaryExpression.java:74)
>    at at.dms.kjc.JFieldDeclaration.analyse (JFieldDeclaration.java:191)
>    at at.dms.kjc.JClassFieldDeclarator.analyse (JClassFieldDeclarator.
> java:57)
>    at at.dms.kjc.JBlock.analyse (JBlock.java:89)
>    at at.dms.kjc.JInitializerDeclaration.checkInitializer (
> JInitializerDeclaration.java:222)
>    at at.dms.kjc.JInterfaceDeclaration.checkInitializers (
> JInterfaceDeclaration.java:166)
>    at at.dms.kjc.JCompilationUnit.checkInitializers (JCompilationUnit.
> java:239)
>    at at.dms.kjc.Main.checkInitializers (Main.java:432)
>    at at.dms.kjc.Main.run (Main.java:193)
>    at at.dms.kjc.Main.compile (Main.java:70)
>    at at.dms.kjc.Main.main (Main.java:61)
> echo timestamp > lib/stamp
> rm -f rt.jar
> 
> so either our stack is too small.. or we have indeed a bug.
> 
> -Riccardo
>