Re: Cleaner stacktraces and diagnostics

[Mark Wielaard <mark at klomp dot org>]

Hi, Andrew,

On Fri, 2005-02-11 at 18:08 +0000, Andrew Haley wrote:
> I think you've spoiled the diagnostics.
> 
> With your patch, a failure to find a class will return only the
> classpath of the final loader that tried to find the class.  But the
> user needs to see all the classpaths that were searched, not just the
> last one.

OK if you think that is needed. It is indeed a nice GNU extension that
our URLClassLoaders actually give the whole seearch path. I think I know
how to extend that to provide the parent classloaders (plus search
paths) without having to create and rechain extra
ClassNotFoundExceptions.

Is the rest of the patch (minus the java.lang.ClassLoader.loadClass()
part) OK? Then I commit that and provide a new patch for the chaining
confusion and efficiency problem in loadClass() later.

Cheers,

Mark

Attachment: signature.asc
Description: This is a digitally signed message part