From owner-cvs-sys Thu May 21 07:26:01 1998 Return-Path: Received: (from daemon@localhost) by hub.freebsd.org (8.8.8/8.8.8) id HAA14940 for cvs-sys-outgoing; Thu, 21 May 1998 07:26:01 -0700 (PDT) (envelope-from owner-cvs-sys) Received: from roma.coe.ufrj.br (jonny@roma.coe.ufrj.br [146.164.53.65]) by hub.freebsd.org (8.8.8/8.8.8) with ESMTP id HAA14920; Thu, 21 May 1998 07:25:33 -0700 (PDT) (envelope-from jonny@jonny.eng.br) Received: (from jonny@localhost) by roma.coe.ufrj.br (8.8.8/8.8.8) id LAA15001; Thu, 21 May 1998 11:24:52 -0300 (EST) (envelope-from jonny) From: Joao Carlos Mendes Luis Message-Id: <199805211424.LAA15001@roma.coe.ufrj.br> Subject: Re: cvs commit: src/sys/pci brooktree848.c In-Reply-To: <199805210258.TAA12797@rah.star-gate.com> from Amancio Hasty at "May 20, 98 07:58:47 pm" To: hasty@rah.star-gate.com (Amancio Hasty) Date: Thu, 21 May 1998 11:24:52 -0300 (EST) Cc: bde@zeta.org.au, jkh@time.cdrom.com, ahasty@FreeBSD.ORG, cvs-all@FreeBSD.ORG, cvs-committers@FreeBSD.ORG, cvs-sys@FreeBSD.ORG, luigi@labinfo.iet.unipi.it, roger@cs.strath.ac.uk X-Mailer: ELM [version 2.4ME+ PL40 (25)] MIME-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: owner-cvs-sys@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk #define quoting(Amancio Hasty) // Actually, it is more natural to standardized around a unified interface Yes... // ala snmp . /dev/hairydevicename with its godzilla ioctl and matching // hairy control program is not intuitive nor a centralized mechanism // to access both the kernel and its devices. // // sysctl type interface could easily be extended to support // multiple variables: // sysctl -w hw.bt848.1.mute=1 hw.bt848.2.tuner=0. // // If I want to find out all the bt848 devices on the system: // sysctl hw.bt848 // which does a snmplike walk on the node hw.bt848 This is not the SNMP way. All unit identifiers are the last part of the mib variable. If you want to detect or count the devices there must be another variable for that. // Incidentely, you can still maintain the device control programs as shell // wrappers which just use sysctl commands;however, the converse is not true. I have one doubt on the subject. Device control can always have permissions controlled on the file node. What about sysctls ? sysctls are good for monitoring and/or settings defaults at rc.local, but not for final user control. That's why I still have no position on this discussion. My position is that sysctl is the choice, then it must look like SNMP. Period. Jonny -- Joao Carlos Mendes Luis M.Sc. Student jonny@jonny.eng.br Universidade Federal do Rio de Janeiro