At the moment there is no (public) way to do a clean
shutdown of the logger system. If one uses a custom
logproc that opens files etc. it is not possible to
detect the correct time a logfile should be closed etc. 

So a mechanism to associate a destructor with the
logging service would be very nice to cleanup things,
without extra bookkeeping.

Either a new subcommand of ${log} or as an option to
logger::init, the procs should be called when a delete
operation is performed, walking the tree from the
children to the root of the tree.

Michael

Logged In: YES 
user_id=240

Closing this bug after CVS commits to add the requested feature.

Logged In: YES 
user_id=240

Ok, I updated it in CVS.  I'll leave the bug open until I
update the man page and you are happy with the new behavior.

Logged In: YES 
user_id=302287

I looked at the patch. Two comments:

1. In most extensions callbacks are executed in the global
scope via uplevel (think of bind and fileevents)
2. it would be nicer if delcmd could be a script, as
otherwise one needed interp alias to curry extra arguments
(like channel names of the logging channel to close etc.)

So perhaps change:
    variable delcallback

    logger::walk [namespace current] delete
    catch { $delcallback [namespace current] }
     namespace delete [namespace current]

to:
            variable delcallback

    logger::walk [namespace current] delete
    catch { uplevel #0 $delcallback [namespace current] }
     namespace delete [namespace current]

Otherwise it works fine.

Michael

File Added - 70679: latest.diff

Logged In: YES 
user_id=240

Try the following patch and let me know if it's what you need.

Thanks!

Logged In: YES 
user_id=302287

Something like it, yes. Preferably bevor the namespace is
finally deleted..

Logged In: YES 
user_id=240

Good idea.  Just to be clear, you need something like:

proc foobar ......

${loggerinstance}::setdeletecallback foobar

That is then called when that logger has 'delete' called?