Subject: Re: kern/1781: 'magic' symbolic link expansion
To: None <netbsd-bugs@NetBSD.ORG>
From: Ty Sarna <tsarna@endicor.com>
List: netbsd-bugs
Date: 11/25/1995 19:20:11
In article <199511250251.UAA13757@solutions.solon.com>,
Peter Seebach  <seebs@solon.com> wrote:
> 
> Yes, by "follow" symlinks, I mean, try to read them (readlink) to see
> what they "should" point at, and possibly look at that file.

This is a problem with the links implementation, but does not exists
with the namei implementation... I'm interested in examining a
implementation based on nullfs-style stacked vnodes with the magic in
the namei part.

> I don't think the magic symlinks really solve the problem; we still need
> machine-specific floppies.

Perhaps for some machines they allways will be needed, but not all.
Also, there isn't *a* problem, there are a whole varienty of solutions
which magic names can solve: issues in diskless environments,
heterogenous environments, etc. If install-time multi-arch bootable
media were the only application, I'd be against magic names as the
solution. However, it's a imple solution to a whole variety of similar
problems, which is why I prefer it over implementing a different
solution for each of the little problems. Magic names provide one
unified approach.