Experimental Physics and
Industrial Control System

"Laznovsky, Michael" <[email protected]> · Wed, 3 Jan 2007 14:46:10 -0800

Hi also- it's not a problem to include the escaped C code 
in the same file as the sequence code; it just becomes part 
of the file written by snc.  I usually insert it before any 
state code to avoid the need for forward references.

Note that when()'s are evaluated only when something happens 
to wake up the corresponding thread, e.g. a monitored/sync'd 
PV update or a delay() expiring; if a state has only when()'s 
that wait on C functions, there's not going to be any way for 
it to continue (unless one of the when()'s triggers immediately 
on initial entry to the state).  There needs to be at least one 
reference to a sync'd PV or a delay() in some when() in each 
state for the seq() task loop to wake up.  It doesn't need to 
be the one with the C function, since ALL when()'s in a state 
are evaluated on ANY event "registered" for that state, but 
that also means your C function when() may not happen when you 
think it will.  In particular, checking a PV _only_ in the escaped 
code won't make the compiler notice that you want to wake up 
when that PV changes.

What sort of compiler errors were you getting?

Mike

-----Original Message-----

From: Al Honey [mailto:[email protected]] 
Sent: Wednesday, January 03, 2007 11:32 AM
To: Dennis Nicklaus; [email protected]
Subject: RE: C function in when() clause of sequencer

Hi

At the Keck observatory, most of our sequencers call 'C' functions 
(hmmm these are sequencers running on vxWorks with EPICS 3.13.10). 
Seems to me it is just a matter of linking and loading.

What you proposed should be functional except, I think, the static 
for the softCheck function may be a problem as it will exist in a C 
source file separate from the SNL file and may not be resolved correctly 
(you might have to experiment).

.....

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System