bin/88119: [patch] partial matching for ifconfig

[Brooks Davis]

The following reply was made to PR bin/88119; it has been noted by GNATS.

From: Brooks Davis <brooks at one-eyed-alien.net>
To: bw at desync.com
Cc: Brooks Davis <brooks at one-eyed-alien.net>, freebsd-gnats-submit at freebsd.org
Subject: Re: bin/88119: [patch] partial matching for ifconfig
Date: Fri, 28 Oct 2005 15:32:02 -0700

 On Fri, Oct 28, 2005 at 05:42:58PM -0400, bw at desync.com wrote:
 > Comments inline.
 > 
 > On Fri, Oct 28, 2005 at 11:39:57AM -0700, Brooks Davis wrote:
 > > Nice idea.  I'm not convinved of the implementation though.  First, the
 > > isdigit() check is bogus because of interface renaming.  You might well
 > > want to match all the wan interfaces on a system with a naming scheme
 > > like lax-wan, sfo-wan, sea-wan.
 > 
 > That's true.  This functionality might be less useful if your interface
 > naming scheme doesn't have some sort of sensible prefix.
 >
 > The isdigit() call wasn't well thought out with regard to interface
 > renaming.  I have revised the patch to match partially only if
 > the interface specified doesn't exist.
 > 
 > http://desync.com/~bw/ifconfig.partialmatch.revised.diff
 
 The manpage entry is problematic in that "match" is poorly defined.
 Also, "if (ifindex == 0)" should probably be "if (ifindex == 0 && argc
 == 0)" so the result of "ifconfig l <stuff>" isn't:
 
 lo0: flags=8049<UP,LOOPBACK,RUNNING,MULTICAST> mtu 16384
 	inet6 ::1 prefixlen 128
 	inet6 fe80::1%lo0 prefixlen 64 scopeid 0x3
 	inet 127.0.0.1 netmask 0xff000000
 
 > > Second, this turns something that was previously an error
 > > into something that works which may be confusing.  My suggestion would
 > > be to implement a new flag, probably -e, which implements egrep style
 > > matching using regex(3) (usr.bin/pkill looks like a good example.)  This
 > > would be both more powerful and avoid the whole issue of POLA
 > > violations.
 > 
 > This is a very simple modification, intended to add convenience for those
 > of us that use FreeBSD with lots of interfaces.  I don't think the
 > addition of a flag or regex matching adds convenience.
 
 Thinking about it more I find this to be a significant POLA violation.
 It means you can no longer reliably test for interface existence with:
 
 if ifconfig $ifn > /dev/null 2>&1; then
 ...
 
 As such I believe a new flag is required.  If you used strstr to
 implement grep-like semantics, -g would be appropriate.  I'm not sure
 what flag would be best for prefix matching.  IMO, -e with egrep regexps
 would be best.  Ideally, -l would be enhanced to support this filtering
 as well.
 
 -- Brooks