Re: [infinispan-dev] Read flags from command or context?
On 3 Jan 2012, at 17:10, Sanne Grinovero wrote:
Hi Galder,
I agree with you, the command looks like a better place to place the
Flags.
+1. The InvocationContext is something local, so if we need remote
propagation the flags should be associated with the command.
The only mismatch is with the origin where flags are enabled,
but I consider this as a convenience rather than a source of
surprising mismatches as I don't see how this API could lead to
unexpected results.
So the idea is in the future to avoid checking for flags in the
context but rather look in the command directly? So that we can avoid
copying them in the invocation context.
can't we just drop the flags from the
invocation context entirely?