occam pi has/had mobile channels, where data pointed to by the pointer that's implicitly sent over the channel will fall out of scope in the sender side. I don't even think that data is converted to a const, the data won't exist anymore. The language takes care of this.
Sending a pointer across without such a mechanism is rather risky, isn't it - of one doesn't introduce a "taken" return message and all parties respect it. If not, is it avoidable to insert data state and eventual polling? Øyvind fredag 8. april 2011 22.13.44 UTC+2 skrev Ian Lance Taylor følgende: > > siddarth shankar <sith.dart...@gmail.com <javascript:>> writes: > > > The reason I asked if it is discouraged is because of the absence of a > > const-style type-constraint. > > > > In that case there won't be an error thrown if the code on the > > subscriber-side modifies the data passed as a pointer(library design > > for instance) > > That is true. > > > Could the compiler do this optimization if it knows that the data sent > > is not modified? > > This could help the GC a lot.. > > Or would that lead to too much automagic-code/over-engineering? > > The compiler could do that in principle in some cases, but I think the > context would be so limited that it's unlikely that this optimization > would ever be implemented. It could only be done if all the relevant > code were in the same package. The compiler would have to be able to > prove that it could see all possible sends and receives on the channel > in question. Without thinking about it too hard, I suspect that there > would be many cases where it would seem that the optimization would > apply but where the compiler could not quite prove that it would apply. > > Ian > > -- You received this message because you are subscribed to the Google Groups "golang-nuts" group. To unsubscribe from this group and stop receiving emails from it, send an email to golang-nuts+unsubscr...@googlegroups.com. For more options, visit https://groups.google.com/d/optout.
