Subject: Re: Share common code/data across ports?
To: Curt Sampson <email@example.com>
From: Scott Reynolds <scottr@Plexus.COM>
Date: 01/10/1997 17:20:22
On Fri, 10 Jan 1997, Curt Sampson wrote:
> As I said before, since these are static entities, you build a
> program that does a static dump of a procfs into a directory tree.
So what, exactly, is the benefit to this approach? Problems with this
include requiring the user to manually create a `snapshot' faked-up procfs
tree before using any debugging tools, and the fact that it takes up real
disk space. Don't underestimate the former; the last thing you want to do
when you're debugging is to have it take even longer than it does.
Also, it's not clear to me that the overall level of complexity of the
system has decreased even one iota; it's just that it's been pushed
somewhere else. The gain from this doesn't appear to outweigh the pain.
PS - since this problem only occurs for people who track -current, or who
mistakenly install a mismatched kernel/userland set, I'm not sure why it's
such a big issue. There are far more useful things to take care of. I've
always been under the impression that one tracks -current at their own