In message <199803131432.JAA00676@bach.mc.com>  "Gordon W. Ross" wrote:
> > From: "Chris G. Demetriou" <cgd@pa.dec.com>
> > Date: Thu, 12 Mar 98 16:15:03 -0800
> 
> > The notion of printf() as it is used now in the kernel (i.e. without a
> > priority) should go away completely, if you want to solve this problem
> > "correctly."
> 
> Do I smell "cmn_err()" lurking about?

And than everybody and his brother does cmn_err(CE_CONT,...) 
to make all the fancy printf tricks again? 

If the way kernel messages are handled is changed, there should be a
completly different approach to it:
	* time stamped log records
	* subsystem specific buffers ( some messages should be kept 
		for a long time others are only interesting for a
		short period of time )
	* async thread/process for printing (with a flush if the system crashes)
	* msg-catalog support (maybe)

Thats a majoe project, and I don't know if its worth the effort. 

An addition (like BSDI) to have a printf for the config messages with
varying verbosity is a quick addition which would help the original
poster.

Stefan
--
Stefan Grefen                                Tandem Computers Europe Inc.
grefen@hprc.tandem.com                       High Performance Research Center
If a group of N persons implements a COBOL compiler, there will be N-1
passes.  Someone in the group has to be the manager.
                -- T. Cheatham