On Mon, 4 May 2009 12:35:12 +0200 Toni Mueller
wrote:
> Hi,
If you read the boot.conf(8) man page, you'd already know it's a hack
to get around the limitations of the x86 BIOS.
> > The new installer (destined for 4.6) in snapshots *already* picks
nope.
> > On all archs, when you wish to boot to a different on-disk kernel
You need to inspect your own bias and hence, opinion. There is no solid
technical reason for one being OK and the other being disliked.
The only place where moving kernels around can be problematic is on
systems where there is a limitation of "bootable" disk space (i.e.
often the first 512MB of the disk). Again, this is caused by a bad
design of the x86 PC BIOS.
> > When you treat i386\amd64 differently with the boot.conf kernel
I think a more accurate way to phrase it is "less-flawed" ;-)
Doing some reading and testing with OpenFirmware (macppc, sparc,
alpha, ...) will show you some better ways to handle things. If you
understand the hardware itself, know how to program in Forth, and the
system has NVRAM, you can actually write scripts to patch the
OpenFirmware BIOS on the fly as it loads from FLASH/ROM.
> > Now, let's say you are using the /etc/boot.conf hack to boot to
Nope. It is better for everything to be consistent. When a new kernel
is built and installed, it should be written to "/bsd" rather than some
custom name.
Having a second stage boot loader like boot.conf that on *only* some
archs will change the decision of which kernel to load leads to
confusion and mistakes.
Just because you might like the non-portable utility offered by the "set
image" feature of boot.conf does not make it a good idea.