[PATCH] prepare streams based messaging

[Volker Lendecke]

On Wed, Jun 22, 2016 at 04:45:50PM +0200, Stefan Metzmacher wrote:
> Hi Volker,
> 
> > On Wed, Jun 22, 2016 at 05:56:22PM +1000, Amitay Isaacs wrote:
> >> Since one of the motivation of tfdd is to avoid file system access, I agree
> > 
> > It seems there is some confusion around the name tfdd. Likely that's an
> > unfortunate name not descriptive enough. I'm open for suggestions.
> 
> Maybe 'tfdmsgd' as 'tfdd' is a 'dementia detection test' in a german
> google search.
> 
> Why was the name changed away from 'tmond' (just being curious, I might
> missed the relevant discussion)

Initially I had planned to add central process monitoring (watch for a
process to exit) to the daemon. To keep it simple and eventually
release it, I decided to go with the quick way to monitor process
death with explicit socketpairs, watching for readability. So the
daemon does not monitor anything anymore at this point, hence tMONd
was not appropriate anymore.

Eventually I might re-add the monitoring capability to the daemon,
because with the dbwrap_nolock work I've talked about sambaxp the
explicit socket pairs might be too heavy weight if used a lot. So we
might go with tmond again?

Volker