[gclist] region inferencing
Thu, 6 Nov 1997 12:45:53 -0600 (CST)
On Thu, 6 Nov 1997, David Chase wrote:
> At 10:45 AM 11/6/97 +0100, Christian Fabre wrote:
> One possible system-level problem with region analysis is that if
> it is used in combination with a generational garbage collector,
> both the region analysis and the garbage collector are going to
> be chasing the same short-lived objects, so the performance
> of the system might not be as good as you might hope. Of course,
> not all systems will be built that way.
I was actually wondering if you could mix pseudo-region analysis with
garbage collection. The basic idea I had was, given a language with
functions as first-class objects (like Java), have each function frame be
it's own generation. When the function exits, it tells the garbage
collector to mark and sweep the whole frame- all objects either get
promoted to a older generation or collected.
Or is this already being done?
Inquiring newbies want to know...