> If I read it correctly, a statsobj is used to get a snapshot, > represented as a string like "key=value ...", of the counters > registered by AddCounter(). The status string is fetched by > getStatsLine(). In this respect, AddCounter() should be a read-only > function. Should I found that the statsobj is useless, it could be > unplugged simply by comment out AddCounter() instead of twisting > additional lines to guarantee initialization. So in my opinion, the > side effect of AddCounter() is a bad idea and should be avoided. Then > we fall back to STATSCOUNTER_INIT() explicitly. The statsobj may be > extended to retrieve internal variables which are not essentially a > dedicated counter, iQueueSize be an example? > > The counter intialization even could be put together with, say, queue > initialization code, so that no need of mutex is required. > > How do you think about it?
I just made a couple of commits to implement this: http://git.adiscon.com/?p=rsyslog.git;a=shortlog;h=refs/heads/v5-stable-newst ats Thanks! Rainer _______________________________________________ rsyslog mailing list http://lists.adiscon.net/mailman/listinfo/rsyslog http://www.rsyslog.com/professional-services/
