Often the question comes up what to information do you want to log and when do you log it?

First rule of thumb, don't put log messages in anything that would be inside tight loop. This likely will impact performance and will spew lots making the log pretty useless.

There are now 7 additional logging level, INFO1-5 and DEBUG1-2. The old logging levels/priorities are retained SEVERE, WARN, INFO, DEBUGGING. The LogSink controls what log messages go into the log file. Set the filter to INFO1 then SEVERE, WARN, INFO, and INFO1 will go into the logfile, set it to DEBUGGING and everything will go into the log file. The log filter level can be controlled for the global logger in python using casalog.filter.

Paraphrasing Steve Myer's CAS-355 issue, here are some guidelines
SEVERE errors that cause the operation to fail 

WARN it ran, but there are issues 

INFO information that the user requested by running 
     the task/tool (e.g. imstat) 

INFO1 diagnostic information helpful to the user to see what 
      happened (e.g. total cleaned flux) 

INFO2 information to indicate what is going on, 
      progress through task (e.g. for timing) 

INFO3 a user supplied note (e.g. entered in logger, 
      or via a "addnote" task) 

INFO4,5 DEBUG1,2 DEBUGGING lowest priorities, any debugging info the 
                     developers feel they have to include, but are 
                     limited utility for users (most of current 
                     messages) 

The priority levels in the C++ source have different names from the levels displayed in the logger...

C++ (LogMessage.h) PrioritySort Logger Priority DESCRIPTIONSorted descending
NORMAL{4,5}, DEBUG{1,2}, DEBUGGING INFO{4,5}, DEBUG{1,2}, DEBUGGING lowest priorities, any debugging info the developers feel they have to include, but are limited utility for users (most of current messages)
WARN WARN it ran, but there are issues
NORMAL2 INFO2 information to indicate what is going on, progress through task (e.g. for timing)
NORMAL INFO information that the user requested by running the task/tool (e.g. imstat)
SEVERE SEVERE errors that cause the operation to fail
NORMAL1 INFO1 diagnostic information helpful to the user to see what happened (e.g. total cleaned flux)
NORMAL3 INFO3 a user supplied note (e.g. entered in logger, or via a "addnote" task)

Please note when ever you output a LogMessage to a LogSink, the message is constructed and the filter level determines whether it goes into the log file. The current filter level is set to INFO, this should probably modified to INFO2 or INFO3 per Steve's guidelines. Take your best guess as to what an appropriate level is for a log message, the users will likely complain if there is too little or too much information. As a rule of thumb, most folks want to know, did it start, is it running, did it finished.

Logging Facility

Logging Interface

This topic: Software > WebHome > LegacyCASA > Implementation > GuidelinesForLogging
Topic revision: 2010-01-21, RemyIndebetouw
This site is powered by FoswikiCopyright © by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding NRAO Public Wiki? Send feedback