Re: Static Garbage Collection

On Mon, May 25, 2009 at 09:28:01PM +0200, Matthias-Christian Ott wrote:
> [...]
> [This feels like it's equivalent to the halting problem which is known to
> be insoluble in the general case, albeit often soluble in specific
> instances. -John]


So you claim that the life-time of some objects cannot be determined
statically at compile-time.


If this claim is true (that seems very likely [1,2]), there are two
possible ways to do deallocate the objects:


1) Figure out all conditions which have to be satisfied, check at
      runtime whether they are met and then de-allocate the object.


      I'm not sure if the life-time of an object can "reduced" to conditions
      which have to be true a runtime.


2) Simply do garbage collection for those objects [1,2]. This seems to
      possible by reference-counting [2].


Moreover, I'm not sure what's faster in practice "condition evaluation"
or simple, non-periodic garbage collection.


Regards,
Matthias-Christian


[1] Arthur Veen. The JOSES Project: Compiling Java for Embedded
        Systems. In Proceeding of JOSES: Java Optimization Strategies for
        Embedded Systems, Genova, Italy, April 2001.
[2] Tobias Ritzau. Memory Efficient Hard Real-Time Garbage Collection.
        In Linkvping Studies in Science and Technology. Dissertations.
        Linkvping, Sweden, June 2003.

Return to the comp.compilers page.
Search the comp.compilers archives again.