Jeremy Manson writes:
>The major problem with this is that we aren't really allowed (or we
>weren't, with JSR-133) to mess with the language or bytecode. The
>"powers that be" actually prefer to make changes in APIs.
I understand the (powers' that be) desire not to mess with the classfile
format, but it looks to me like the tide is swinging the other way with all
the new stuff introduced at that layer in 1.5.
I also thought this sort of distinction made sense to expose explicitly to
method callers, though I admit I haven't given it a ton of thought.
>The libraries would have to have a laundry list of explicit special
>cases anyway, because each individual call would have to be defined to
>require the proposed access flag.
Sorry, I was being too loose with my language. I meant *out-of-model* and,
by implication, non-user-specifiable special cases. I have no major
objection as long as there's a reasonable way for an end-user to make that
sort of declaration and proceed to write code in the obvious style (that
is, without explicitly writing reflection calls), which is what I was
trying to get at.
JavaMemoryModel mailing list - http://www.cs.umd.edu/~pugh/java/memoryModel
This archive was generated by hypermail 2b29 : Thu Oct 13 2005 - 07:01:09 EDT