darcs patch: Support an interrupt flag, and primitive... (and 4 more)

[Tony Garnock-Jones]

Brian Rice wrote:
> Er, I meant to address this to Tony. Sorry, Nick. :)

Heh :)

> On Oct 26, 2005, at 5:30 PM, Brian Rice wrote:
> 
>> After using the first major patch in this set to produce a new VM  and 
>> image, starting up with those gives me the following error:

The only changes from the first patch were to:

  - introduce a new boolean VM global, InterruptFlag
  - change VM's Interpreter>>interpret to check the flag and
    return (after cleaning up the GC root) if it is set
  - add two primitives for manipulating the flag, and fresh
    Interpreters respectively

Nothing modifies the InterruptFlag (which reminds me - it needs 
initialising - C compilers don't always initialise things to false), and 
nothing calls the primitives, so in the absence of any other changes, 
the new system should work identically to the old.

I've no idea what could be causing the problem. It doesn't happen for 
me. Does a fresh bootstrap *without* any of these patches have the same 
problem?

The other thing is, I do have a BE environment I can test on, but I 
haven't been using it - I was working on a Little-Endian machine yesterday.

Tony