Experimental Physics and
Industrial Control System

Benjamin Franksen <[email protected]> · Wed, 16 Nov 2005 14:03:56 +0100

On Tuesday 15 November 2005 20:18, Andrew Johnson wrote:
> It would be nice to get Banjamin's patch to the logClient library
> included, which we're expecting tomorrow.

Small problem, here. Some explanation(s) first: The current module named 
'logClient' contains two kinds of routines: generic (logClient*) and 
specific (iocLog*, for the standard IOC error logging i.e. via 
errlogAddListener). The latter stuff is only a very thin layer that 
actually uses only the public interface of the generic routines.

However, teh separatioon is not complete; there are still a few places 
where the generic routines do specific things:

(1) logClientSendMessage() checks the value of iocLogDisable and returns 
without doing anything if it is set.

(2) logClientInit() uses the EPICS_IOC_LOG_* environment vars to 
identify server address and port.

(3) logClientInit() also contains a call to errlogAddListener, so that 
for all calls to errlog* a message is sent via the created logClient 
instance.

These things do not belong to the generic functionality and instead 
should be done by the special instance, namely the standard ioc (error) 
logging client and its supporting routines.

OTOH, for strict compatibility, the behavior must remain exactly the 
same as before, which means that instead of changing the existing 
functions logClientSendMessage() and logClientInit(), I must add new 
additional versions of them.

What should I do?

Note: The generic routines are not yet documented anywhere.

Does anyone actually use them, relying on the above behavior?

Ben

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System