Re: Patch: FYI: More efficient ResourceBundle calls

[Andrew Haley]

Bryce McKinlay writes:
 > Dalibor Topic wrote:
 > 
 > >> I think the solution here is to pass the system class
 > >> loader. This should always be correct for these bootstrap
 > >> classes. It doesn't solve the performance issues for cases where
 > >> a security manager is present, since a check will be performed
 > >> by the getSystemClassLoader call - however its better than what
 > >> we had before and certainly better than my incorrect patch ;-)
 > >
 > > Ah, the curse of getting microoptimizations right ;)
 > 
 > I don't really consider this to be a micro-optimization. In libgcj
 > it causes pretty severe performance problems. You could argue that
 > this is because libgcj's calling-classloader check is too slow, and
 > you'd be right, but I would imagine that in most VMs, stack-walking
 > checks are relatively slow.

I agree with that: the more you optimize the VM, the greater the
relative cost to walk the stack chain.  

Andrew.