fine grain logging cotrol

[Eric S. Johansson]

Dennis Lee Bieber wrote:

> 	I've never resorted to the debugger -- it's always been faster for
> me to just wolf-fence* code with print statements...

depends on the situation for me.  normally I use log statements that 
turn on or off based on predicates (now I need to figure out how to 
handle a SIGUSR1 to reload the who-logs info)  hence this quest.  when 
done it's going into my rcsoc (random cross section of code) collection 
which I'll publish one of these days.  Also, I need to see if I can make 
this lower overhead by pre computing as much of the predicate term as 
possible and perform the test before calling a log statement.   I put in 
*lots* of log statements but leave most of them off until needed. 
different form of wolf fence.

> 	Could you possibly be confusing the execution of the "def"
> statement... Remember, in Python "class" and "def" are executable
> statements themselves. That is, what you are seeing during __new__ is
> the execution of the statement that defines the get_logger method, NOT a
> call of get_logger itself.

I know def executes.  forth plays similar games and I spent a *lot* of 
time in the guts of forth.   With python, I'm just not familiar enough 
with what goes on beneath the sheets.

> 
> 	Ah.. Okay... the wolf howled... Looking deeper at the code, it is
> not a direct call to get_logger, but an indirect one via the dictionary
> entry.

right.  still not comfortable the property call.  if I'm reading the 
right docs, the first arg is a get method and I'm not sure how a logger 
fills that need.

> 
> 	classdict["_Felis.alpha__logger"]
> 
> 	None the less, your posted code did have too many _ on the init,
> meaning the initialization never took place, AND the references to
> "test_names" needs to be prefixed with "self."

my most common brain fades.  I also miss tt/ttt, ss/sss, 1l and 
probably a few others.  I sometimes think IDE's should have a "did you 
mean" function to alert you to stuff that's "almost" right.  kind of 
like the lisp programmer's assistant.

--- eric