People writing JVMs very aware of the issues raised by precise
exceptions in Java
including the fact that precise exceptions can prevent reordering of
operations.
This is one of the reason people are so eager in JVMs to prove that
runtime exceptions
will not occur, because doing so removes these barriers to reordering.
There is absolutely nothing about Exceptions and RuntimeErrors that
involves
the memory model. They are all completely standard issues that have to
be dealt with even in the single threaded case.
Nothing to see folks, time to move on.
Bill
On Apr 11, 2004, at 8:54 AM, Vijay Saraswat wrote:
>
> Bottom line is that I think the JMM needs to discuss these issues.
>
> Why is any of this any different from what 11.3.1 implies, and what
> must
> already be implemented by VMs? If we aren't really changing anything,
> and
> it isn't part of Chapter 17, why do we need to discuss it?
> Because it is not clear to me that anyone really understands these
> issues.
>
> Java design -- as far as concurrency and numerical computation are
> concerned -- was ad hoc (albeit still a vast improvement on other
> languages).
>
> We are now working on "getting right" a very fundamental building
> block of the whole language (the memory model). This might have hidden
> a whole bunch of performance problems in other parts of the language.
>
> Understanding whether this is the case is important.
>
> Best,
> Vijay
>
>
-------------------------------
JavaMemoryModel mailing list - http://www.cs.umd.edu/~pugh/java/memoryModel
This archive was generated by hypermail 2b29 : Thu Oct 13 2005 - 07:01:03 EDT