[olsr-dev] Re: Die "%m", die

[Bernd Petrovitsch]

On Mon, 2006-01-09 at 11:58 +0100, Andreas Tønnesen wrote:
> > Andreas - did I miss anything?
> 
> Nope, I agree. Now just some tiny words of history. There used to be a

I feared that there is only history.

> olsr_print function, but this was replaced with the macro a couple of
> revisions ago for the reasons Thomas put up. Printouts for high
> debuglevels tend to be called very intensive and did cost a lot of CPU as
> shown by profiling. We discussed having debug options fixed at build time,
> but it makes debugging harder so we went with the define solution(which
> also allows us to remove the debug statements completly at buildtime).
> Also, there is really no big philosophy behind the stderr printout vs.
> syslog.  I've mainly just taken the approach that you syslog what you want
> an admin to know and possibly report to us(but not create gigantic logs)
> and you print what you want people debugging/developing to know. The
> syslog set is pretty much always a subset of the print set, but I think
> they should be regarded as two different entities.
> Now I'm sure this can be handled better but there has not been any effort
> put into cleaning this up. Suggestions for changes are always welcome :-)

Features:
- several debug levels as specified now e.g. in the config file for
  run-time selection.
  Which number range BTW? The numbers should somewhat correlate with
  the LOG_ERR levels of syslog() if we use that (and I don't know about
  the Win view on that).
- a #define for compile-time "selection" of said levels.
- indepently selection of "print to stderr" and "syslog" it.
OK?

	Bernd
-- 
Firmix Software GmbH                   http://www.firmix.at/
mobil: +43 664 4416156                 fax: +43 1 7890849-55
          Embedded Linux Development and Services