request for review: 6436034: Instance filter doesn't filter event if it occurs in native method

[Alan Bateman]

Keith McGuigan wrote:
> :
> I like the flexibility of being able to swap in/out different JVMs -- 
> be it earlier versions of Hotspot or something else that maybe doesn't 
> yet support JVMTI 1.2.  Having the code written this way also makes it 
> easier to port this code verbatim into other JDK codebases (jdk6?) if 
> it's required.  I understand this is different than things were done 
> in the past, but I don't think HSX was around the last time the JVMTI 
> version was updated either.  Decoupling the JDK from the JVM as much 
> as possible is a good thing in my opinion.  We can discuss this 
> further (offline?) if I'm not convincing enough yet :)
I like the idea of flexibility too but just concerned that things could 
get messy once additional functions or events are added. Also I'm not 
sure how practical it will be to drop-in older VMs into jdk7. With the 
HSX model it's about dropping in newer HotSpots into older JDKs and I'm 
not sure if we've thought much about the reverse. I'm just thinking 
about the HPI for example. We hope to remove that from jdk7 once 
hs20-b04 is in a promoted build. The implication is that it just won't 
be possible to drop in older HotSpot builds that require the HPI. I'll 
go crawl back under my rock now :-)

-Alan.