Experimental Physics and
Industrial Control System

Babak Kalantari <[email protected]> · Mon, 25 Oct 2010 12:02:45 -0000

In general I like the idea of named events. I have faced situations where I was lost about the used soft events in an IOC and it was tedious to track down what HW event fired what SW event to avoild clashes.

My comments:
1) As far as I understood the ordring of adding/removing scan lists is not affected. I think it is important that the position of the event nodes in the lists remains the same in both cases (event / named_event).

2) The new method has introduced an additional lock for each priority list (event_list_lock) and for accessing a node in the event list first the lock of the corresponding priority has be acquired, the priority list be found, and the lock is then released. This will also introduce some performance penalty (in addition to strcmp). In the current method this is achieved by means of 2-dimentional array of [priority][event].

I know that normally number of soft events in an IOC is not big, nevertheless I think one has to study the performance effects of the new appraoch more deeply. 

-- 
https://code.launchpad.net/~dirk.zimoch/epics-base/named-soft-events/+merge/26190
Your team EPICS Core Developers is requested to review the proposed merge of lp:~dirk.zimoch/epics-base/named-soft-events into lp:epics-base.

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System