Subject: Re: Another changer, another changer problem
To: der Mouse <mouse@rodents.montreal.qc.ca>
From: None <seebs@plethora.net>
List: current-users
Date: 10/01/1998 23:20:18
In message <199810020409.AAA18680@Twig.Rodents.Montreal.QC.CA>, der Mouse write
s:
>> I don't like the nailing-down thing; I don't mind that it exists, but
>> I don't want to have to do it to get nailed-down names.

>"I want nailed-down names, but I don't want to have to nail down names."

Yup.

>This sounds to me like little more than a request that they be there by
>default.

Essentially, yes.  I would ideally like to have both naming conventions.

Sometimes, I'd like to have /dev/d0[a-h] just be "disks", without regard
to whether they're SCSI or EIDE.  Sometimes, I'd rather be specifying
exactly what device I get.

I wouldn't entirely mind having a /dev/devtree such that I could say
	mount -u /dev/devtree/pci0/aic0/target0/lun0/a /
but that may be a bit excessive.

>> Basically, depending on the phase of the moon, I may find myself
>> wanting to refer to devices by *either* the nailed-down or the
>> non-nailed-down names, and I'd like to have both.

>I don't believe this is possible without either using a new major or
>using a minor bit to tell the difference.

Quite possible.

>Perhaps you want a scsifs that you can mount -o union on /dev, that
>hooks into the scsibus config machinery to present exactly and only the
>devices that exist - do a reprobe and more entries magically appear.
>(And disappear, once reprobe is fixed to unconfig removed devices.)

Perhaps; I could happily live with /dev/scsi being as transient as /dev/fd.
But I'd love to have it available sometimes.

>Is it worth it?  To me, it is not worth the effort it would take to
>implement it.  If you feel differently, go to it!  I think such a
>scsifs would be a Pretty Cool Hack, even if I probably wouldn't use it
>myself.

In general, I'd like a lot more run-time control over the device tree,
and a lot more ways to probe at it.

-s