On Thu, Jan 20, 2011 at 11:23:15PM +0100, Andreas Raab wrote:
> On 1/20/2011 6:53 PM, Levente Uzonyi wrote:
> >On Thu, 20 Jan 2011, Chris Muller wrote:
> >
> >>We're down to just the following failing tests (2202 VM):
> >>
> >>1 CompilerExceptionsTest>>#testUnusedVariable
> >>2 DecompilerTests>>#testDecompilerInClassesMAtoMM
> >>3 DecompilerTests>>#testDecompilerInClassesSNtoSZ
> >>4 ExceptionTests>>#testHandlerFromAction
> >>5 MirrorPrimitiveTests>>#testMirrorAt
> >>6 MirrorPrimitiveTests>>#testMirrorInstVarAt
> >>7 ProcessTest>>#testAtomicSuspend
> >>8 StandardSystemFontsTest>>#testRestoreDefaultFonts
> >>
> >>This is the list when run with the latest Cog VM:
> >>
> >>1 CompilerExceptionsTest>>#testUnusedVariable
> >>2 DecompilerTests>>#testDecompilerInClassesMAtoMM
> >>3 DecompilerTests>>#testDecompilerInClassesSNtoSZ
> >>4 ExceptionTests>>#testHandlerFromAction
> >>9 FileStreamTest>>#testPositionPastEndIsAtEnd
> >
> >This one is passing for me.
> >
> >>10 MorphicUIBugTest>>#testShowAllBinParts
> >>
> >>1, 4, 5, 6 and 9 are VM-related and cannot be fixed for 4.2. Levente
> >
> >1 is a Compiler bug, but it requires a lot of work to fix. 4 is also a
> >bug in the image, though it may require VM changes to fix it.
> 
> I think 4 can be fixed from the image but since my first attempt was too 
> naive we shouldn't wait for it. I vote for declaring it an expected 
> failure until we've figured out how to deal with it properly.
> 
> 5-7 are probably just missing in VMMaker. I would coordinate with David 
> to see if we can get these into the interpreter (#7 actually *is* in 
> VMMaker but no VMs have been built with it).

I took a look at the mirror primitives a couple of weeks ago, and I'm
afraid it is a bit beyond my abilities to integrate them. Eliot may have
some further updates in the works, but aside from that I would say that
we should call it an expected failure for the interpreter VM.

Dave

> 
> >7 is a VM bug IIRC. Do you know why is 8 failing?
> 
> See above for 7. 8, I suspect, may be a side effect of running tests in 
> a particular order.
> 
> Cheers,
>   - Andreas