Re: PR java/14070: gij and -jar argument should set the manifest Class-path recursively

[Andrew Haley <aph at redhat dot com>]

Tom Tromey writes:
 > >>>>> "Andrew" == Andrew Haley <aph@redhat.com> writes:
 > 
 > Andrew> Okay, I think I see.  So, at entry to the URLClassLoader you create a
 > Andrew> hash table and then when you're looping around all the sub-loaders you
 > Andrew> check the hash table to see where you've been.  I'll try it.
 > 
 > I just meant while reading the Class-Path entries of the jar files,
 > keep track of where you've been and don't revisit.  Then at the end
 > you will have a unique-ified list that you keep.  Is this what you
 > meant? 

Could be.  From the point of view of correctness I doubt that it makes
any difference.

 > I couldn't tell from what you wrote if you meant you would do
 > the check when looking up a class or resource -- I would expect that
 > to be too expensive.

I guess so, but given the overhead of class loading I wouldn't expect
a little hash table to make any discernible difference.  But given the
choice I guess I'll try to do it the efficient way...

Andrew.