Re: [9fans] interesting potential targets for plan 9 and/or inferno

[Geoffrey Avila <avlg@sdsc.edu>]

>  2. The interfaces are baffling.  As Master Shake said, "It's too
>  advanced to be compatible". Not just rio; I mean adding a user to a system,
>  logging in under a different account by rebooting, mounting devices as
>  filesystems, making configuration changes, everything is different. I don't
>  think anything makes sense until you understand the entire system, top to
>  bottom. It's a difficult investment to make. I appreciate that this is not
>  your problem.

A nice readable softcover book, in the style of the Nemeth-Snyder-Seebass 
"Unix System Administration Handbook", but for plan9 would be a huge 
honking help; it would make the epiphany one gets from installing a 
fileserver and a terminal and reading the papers and manpages a much more 
accessible moment.


> I think that I understand the picture that the forefathers had in mind when
> developing the system, it's just that it's very difficult to implement an
> infrastructure that shows the benefits of the plan 9 approach.

True. Systems today are homogenous x86en that consume power and desk 
space-there is no usual need to have more than one in your den.


And when you
> finally do, It's not clear that you've got something better, because there's
> nothing you can do after building it that you couldn't do before.

I beg to differ. My interest is less in clusters, and more in providing a 
homogenous reliable computing environment comprising desktops and servers 
across boundaries of hardware and Unix vendor. This takes enormous amounts 
of effort, more than it should, because of assumptions about how the 
system is to be used havn't changed at the most basic level in over two 
decades. You can approximate the same functionality with 
*nix+LDAP+AFS+krb5 or Windows with AD, but the effort to support that 
approximation does not increase linearly with size.

-GBA