[e-lang] Vat figure

Mark Miller markm at cs.jhu.edu
Mon Jun 20 11:39:49 EDT 2005


Karp, Alan H wrote:
> After seeing Ping's figures and reading the comments about them,  I
> think truth is getting in the way of clarity.  Why not describe the data
> structure as a dequeue?  If you do, then you can say that immediate
> calls put items at the head of the dequeue and eventual sends put them
> at the tail.  I think both the picture and the explanation get simpler.

When I gave the Promise Pipelining talk at HP, Bob Tarjan made a suggestion
along these lines. I've toyed with the idea repeatedly since then, but I 
decided not to, because
* activation frames vs pending deliveries really are very different
* I want to highlight event (turn) boundaries, rather than obscuring them,
   since interleavings happen at event granularity.

-- 
Text by me above is hereby placed in the public domain

     Cheers,
     --MarkM



More information about the e-lang mailing list