[Info-ingres] What does XA_END mean in the SC930 log?

[Karl and Betty Schendel]

> On Jan 4, 2018, at 10:43 AM, Roy Hann  wrote:
> 
> This seems a killer argument for unambiguously marking, using an EQY,
> where a transaction actually ends (i.e. COMMITs or ROLLBACKs). 
> 
> Is there any hope that could be done? I mean, is there any way the
> information could be communicated from DMF (or whichever facility
> _really_ knows what's going on)?

It's certainly possible to indicate the XA state as well as the qef/sequencer
transaction state in the EQY.  It just needs the appropriate code change.
I can't imagine that it's a big deal, although to avoid a bunch of extra trace
calls into DMF, it might be worth a wee bit of extra effort to try to keep
the reliable XA state in the sequencer.

Karl