Re: [jdt-dev] Follow up on Bug 377850

[Gunnar Wagenknecht <gunnar@xxxxxxxxxxxxxxx>]

To close the loop on this one, I believe the errors we saw happen because of too early compilation, i.e. the classpath isn't complete yet because code isn't fully complied. 

I noticed that the batch compiler is disabling class file generation when "-proc:only" is specified. I wonder if this option wasn't available before. Anyway, disabling class file generation made the issues go away.

-Gunnar

-- 
Gunnar Wagenknecht
gunnar@xxxxxxxxxxxxxxx, http://guw.io/


> On Mar 5, 2020, at 21:00, Gunnar Wagenknecht <gunnar@xxxxxxxxxxxxxxx> wrote:
> 
> Hi Stephan,
> 
> I thought it might be better to continue the discussion on the mailing list. Thanks a lot for your help!
> 
> I started debugging the issue I'm experiencing described here:
> https://bugs.eclipse.org/bugs/show_bug.cgi?id=377850#c34
> 
> So far I found out that the problem is being reported because 
> org.eclipse.jdt.internal.compiler.lookup.PackageBinding.hasType0Any(char[])
> 
> returns true for the problematic name.
> 
> What I noticed during debugging is that:
> - the PackageBinding is an instance of PlainPackageBinding
> - the found type is an instance of ProblemReferenceBinding
> - the problemReason is type 1 (ProblemReasons.NotFound)
> 
> Can this be a valid case for reporting this error? If the problem reason is NotFound, shouldn't this be handled as "no collision"?
> 
> -Gunnar
> 
> -- 
> Gunnar Wagenknecht
> gunnar@xxxxxxxxxxxxxxx, http://guw.io/
> 
>