cvs commit: src/sys/pci brooktree848.c

[Joao Carlos Mendes Luis]

#define quoting(Amancio Hasty)
// When either you or Roger can verify the new sysctl hw.bt848.format,
// I will commit the fix.
//
// hw.bt848.format = 0 | 1
//
// 0 Denotes PAL
// 1 Denotes NTSC

Shouldn't there be also the sub-formats, like NTSM/M, NTSC/J, PAL/I,
PAL/M, etc ? I'm thinking of the BT848SFMT ioctl, and not the old
compatibility METEORSFMT one.

BTW: Somebody has messed up PAL/M again. PAL/M is NTSC-like, not
PAL-like, despite it's name. It's indeed a 60/525 system with
support for Phase Alternated Lines (PAL). So the ioctl BT848SFMT has
to be changed, and, please, add a comment to avoid future changes.
Curiously, the struct format_params seems unchanged. Unfortunatly,
I don't have a bt848 anymore to test these.

While I'm here, I'd like to note that sysctl is somewhat like SNMP,
and SNMP variables have a unit name as this small extract from a
snmpwalk shows.

interfaces.ifTable.ifEntry.ifInOctets.1 = 204219604
interfaces.ifTable.ifEntry.ifInOctets.2 = 0
interfaces.ifTable.ifEntry.ifInOctets.3 = 0
interfaces.ifTable.ifEntry.ifInOctets.4 = 32787782
interfaces.ifTable.ifEntry.ifInOctets.5 = 0

Things that have the notion of units should reflect this in its
sysctl, even if 99.999% of people only have one of those interfaces.

Of course, this is just my very humble opinion.

Jonny

--
Joao Carlos Mendes Luis M.Sc. Student
jo...@jonny.eng.br Universidade Federal do Rio de Janeiro

To Unsubscribe: send mail to majo...@FreeBSD.org
with "unsubscribe freebsd-multimedia" in the body of the message

[Mike Smith]

> > While I'm here, I'd like to note that sysctl is somewhat like SNMP,
> > and SNMP variables have a unit name as this small extract from a
> > snmpwalk shows.
>

> > Things that have the notion of units should reflect this in its
> > sysctl, even if 99.999% of people only have one of those interfaces.
>

> I strongly disagree. Devices that have a notion of units should export
> a unit-based interface in the filesystem (i.e., /dev/foo999). sysctl
> is very much unsuited to this. (I'm still unhappy about the way I had
> to kluge things for the ifmib. At least it works...)

It's debatable as to whether the MIB should exist in the filespace, but
note that it is actually quite trivial to export stuff in this fashion
using sysctl, as it supports handing traversal off to a client-supplied
traversal routine at any level.

--
\\ Sometimes you're ahead, \\ Mike Smith
\\ sometimes you're behind. \\ mi...@smith.net.au
\\ The race is long, and in the \\ msm...@freebsd.org
\\ end it's only with yourself. \\ msm...@cdrom.com