Hi Peter, On 05/31/2013 11:33 AM, Peter Levart wrote: > Implementation issues aside, the name of the tag specified in the > @Contended annotation could be evaluated in the context of the fully > qualified class name in which the annotation is used. For example:
I thought about implying the fully-qualified names for the tags; but that only makes sense if we allow tag inheritance, and we don't. > When the tag contained at least one dot, it would be interpreted as > fully qualified, otherwise as a simple name. Ugh. Messy... :) > Implementation could allow one of the contention groups to be used > in a "global" context on the boundary of superclass/subclass, but > which one is the one, would have to be anticipated and explicitly or > implicitly noted on the superclass, so that the layout logic could > group those fields at the end of object as soon as the superclass is > loaded. We do something like that for normal oops layout, so that we have the contiguous oops region on the superclass/subclass boundary. This is the corner case though, which does not scale beyond just two classes. I think this case is superficial corner case, and does not deserve the treatment. Anyhow, the contended tags are just the hints within the hint: we communicate that particular fields *can* be grouped together if implementation thinks that is a good idea. The way @C is spec'ed now, we might even ignore the tags at all and pad the fields distinctively, i.e. we say those fields are not required to be isolated pair-wise with any field in the group, but only with all other contended groups and fields. Having said that, I should probably add the sentence in the docs, like this: "A contention group defines a set of one or more fields that collectively must be isolated from all other contention groups. [The fields in the same contention group may not be pair-wise isolated]". > I don't know if this is useful anyway. Could a superclass correctly > anticipate the "global" group in some situations? I think what is being proposed for @C to work across the hierarchy is the overkill, handles no useful cases, but bears the significant development cost to do. That's the recipe for saying "nope!" :) Don't try to turn @Contended into the swiss army knife field layout hint. It is not. -Aleksey.
