Subject: Re: NetBSD master CVS tree commits
To: Christoph Badura <bad@flatlin.ka.sub.org>
From: Chris G Demetriou <Chris_G_Demetriou@UX2.SP.CS.CMU.EDU>
List: current-users
Date: 04/14/1996 16:55:19
Please don't reply to the 'source' list...  The correct forum is
current-users.

> In hanse-ml.netbsd.source-changes you write:
> >if no DDB, define Debugger() to be a panic.
> 
> Shouldn't a driver not have to bother with this?

Yes, the driver _should_ have to bother with this.

remember the recent thread about "panics that were written to avoid
having to write real error handling code?"  This is one of them.

If something should be a panic, it should be coded as a panic.
If it's debugging, or otherwise informational, it should be:
	(1) printfs, and recovery code, or
	(2) a panic if it's an "unrecoverable error."

I think i'd argue that there should never be Debugger() calls in code
at all, except when actually debugging that code.  They certainly
shouldn't be in 'production' code.


cgd