[concurrency-interest] Synchronization question

Holger Hoffstätte holger at wizards.de
Thu Jun 29 04:34:36 EDT 2006


Dhanji R. Prasanna wrote:
>> I'm mainly interested in unambiguous clarification of the apparently (?)
>> possible time window that may or may not exist after the
>> synchronized set method has returned. 
> 
> Afaics, there is no way to unambiguously clarify this without a
> machine-specific answer.

That would have been fine, I was merely curious. Too many years of lurking
in comp.arch, I guess..

> Im not clear how this is a race condition, if the reader acts after the
> happens-before edge, there won't be a competition for shared
> memory--unless I am missing something (quite possible)? Do you just mean
> that there is an (potential) undefined delay?

Yes (the latter), and as Brian has commented it not only exists but is
also irrelevant to know from an external point of view - since the spec
says it can be forever. I understand his reluctance to give concrete
examples since that would indeed only lead to even more misguided
folklore, and Java already has enough of that.

Holger


More information about the Concurrency-interest mailing list