interaction between isolated regions and logical clocks

[GoogleCodeExporter]

Consider this program:

  let(isolated (signal | Ltimer(1) >> println("ISOLATED")))

When the isolated region's parent group terminates, it eventually
terminates the root logical clock so that the Ltimer(1) never responds. It
seems that what needs to happen is that quiescence is tracked independently
of liveness, so that a "dead" region may actually be non-quiescent (if it
contains an isolated region). However this seems incompatible with the
desired behavior of group cells, that when closed the region is instantly
quiescent.

One easy solution would be to have "isolated" introduce a new clock. But
I'd like to avoid this.

A related question is what this program does:

  let(withLtimer(lambda () = isolated (signal | Ltimer(1) >>
println("ISOLATED"))))

Original issue reported on code.google.com by adrianqu...@gmail.com on 17 Apr 2009 at 5:05

[GoogleCodeExporter]

Original comment by adrianqu...@gmail.com on 17 Apr 2009 at 5:07

• Added labels: Research

[GoogleCodeExporter]

r1103 implements the easy solution of having "isolated" introduce a new clock. 
We
still need to understand this problem better.

Original comment by adrianqu...@gmail.com on 17 Apr 2009 at 2:07

[GoogleCodeExporter]

The isolated keyword will be removed in 1.1, and has been supplanted by the 
Site site. In as much as this is an 
issue with the Site site itself, we are addressing it separately in the theory 
of logical time.

Original comment by dkitc...@gmail.com on 19 Feb 2010 at 5:01

• Changed state: WontFix