Sounds doable.  We'll look into it.  I just wanted to a gauge on whether this was even remotely possible before attempting ;-)

Derek


On Fri, Apr 26, 2013 at 10:08 PM, Roy Stogner <roystgnr@ices.utexas.edu> wrote:

On Fri, 26 Apr 2013, Derek Gaston wrote:

What's involved in enabling the ability to add a System to an ES
after init() has been called?   I can't see why it wouldn't be ok...
each System is fairly independent?

It's sort of like the "adding a vector after init() has been called"
case - we didn't originally support it because we didn't have use
cases and it was extra work, but the post-init vector case wasn't too
much extra work and I don't think the post-init system case would be
either.

Probably roughly the same process to fix it, too: when something new
is added, test to see if init has been called.  If it has, then we do
the initialization for that new addition right away instead of
waiting.
---
Roy