In message <20021219234721.A18457@vaasje.org>Frank van der Linden writes
>On Thu, Dec 19, 2002 at 01:47:17PM -0800, Jonathan Stone wrote:
>> 
>> The specific features we're missing which I think should be fix:
>>     1. Mounting non-NetBSD filesystems  (from NetBSD).
>
>Hm? ext2fs, msdosfs, ntfs, etc? 

Yes, that's right. We don't support that, insofaras *we dont support
finding where the damn filesysems are*. Once you've found them (and
taken the bass-ackwards step of hardcoding their locations in the
NetBSD disklabel), yes we can mount them.

Try mounting all the UFS filesystems in a FreeBSD partition sometime.
(no, mbrlabel does n't find them all).  Bill's proposal makes that
even harder than it is now.
 
Or try mounting ext2fs filesystems from a multi-partition native-Linux
setup, where some (possibly all but one, in a default RH setup?)
are in nested extended partitions.
 

>The only inconvenience there is
>that you have to 'find' them, i.e. NetBSD doesn't automatically
>find the partitions, you have to know the offsets and add them
>to a NetBSD disklabel, 

Frank, that's not an "incovenience":we flat dont support it.
Sure, we dont *prohibit* it, but we don't actually support it.
The tools we have don't do an even halfway-decent job of it.

Even if we did, its not an ``solution'' that meets the standareds of
NetBSD engineering. (Think: its not *our* space, it belongs to
Windows/Linux/BSDI/FreeBSD/whoever. If the user boots into that OS and
repartitions that OS, the NetBSD label holds stale info; attempting to
mount, or fsck, based on that stale info could lead to loss of data).

Bill's proposal is on the right track in that regard.