[Vm-dev] Re: Different bytecode set per process

[tim Rowledge]

The problem I anticipate here is almost the same as that which came up many, many, years ago when Ted K  had a go at making projects act a bit like namespaces (by swapping in certain methods or something like that? I forget). In that case if you had any background processes (which we do by default) then they could get totally screwed by this change-over. 

Similarly here we are talking about multiple processes within the same object world, so the same panoply of methods are available. You start a new process, assign a restricted bytecode set for it, start it running… and send a message. Which finds a method that uses the now-forbidden bytecodes. What happens? 

How about a bug or other interruption? Now you want to open the debugger; what happens?

I’d be much more in favour of being able to have disjoint object spaces and decently effective communications between them. If nothing else it would help (a little) to decently enforce encapsulation, something I see constantly being eroded by poor programming and the insanely common use of ‘accessor methods’ that do nothing but rip open the kimono.

tim
--
tim Rowledge; tim at rowledge.org; http://www.rowledge.org/tim
"Oh bother" said Pooh, as he reached for the reset button