Subject: use of IGNORE= for packages also incorporated into the system....
To: NetBSD Packages Technical Discussion List <firstname.lastname@example.org>
From: Greg A. Woods <email@example.com>
Date: 02/18/1999 14:21:25
I've noticed a number of packages have begun using "IGNORE=" to prevent
their use if they appear to have already been installed in the system.
Two recent examples are "top" and "gtexinfo".
I asked Simon about this and he suggested that perhaps it might be an
appropriate topic for tech-pkg.
So, I'm wondering if this is a good thing to do, and/or if it's even
necessary. It seems to me that that this is just a little too much
local policy to enforce by default, and for no really good reason.
My thoughts are that if say I'm running NetBSD-1.4 which has top-3.5
integrated, and top-4.0 comes out and is supported in pkgsrc, I might
want to install the new top-4.0 as a package, and of course allow my
users to run which ever version they please by selecting the appropriate
PATH or whatever. This is something that system administrators have
done to upgrade vendor tools from time immemorial, though rarely with
such sophisticated tools as pkg_*! ;-)
The only time I can see a real conflict between a (possibly newer)
package and an equivalent system tool is when the package is built and
installed with PREFIX=/usr. Even then, if I do that and obliterate my
original "top" binary then that's my business, isn't it?
I think the same sort of argument might apply to the NO_PKG_REGISTER=
and NO_PACKAGE= settings used by things like pkgtools/pkg_install. I
ended up commenting those out so that I could easily distributed the
package to a number of similar machines (though in the end it didn't
appear that the update was necessary just to install the pkg binaries
for the dependent packages).
Greg A. Woods
+1 416 218-0098 VE3TCP <firstname.lastname@example.org> <robohack!woods>
Planix, Inc. <email@example.com>; Secrets of the Weird <firstname.lastname@example.org>