Bart Smaalders wrote:
Dan Price wrote:
On Thu 09 Apr 2009 at 04:01PM, Bart Smaalders wrote:
Why do you want to have packages that don't install in zones, rather
than having some of the contents of the package not install in a zone?
fwiw, Bart and I don't see eye to eye on this 100%. Or maybe
it's better to admit: "I have vague concerns" :)
I think there are nasty things which can happen in either direction.
- If you allow packages to be installed, but empty, then
the presence or absence of such a package can not be used as
an indicator of availability of features. It gets to the
problem of "what does installed actually mean"? While I'm
not a huge advocate of using the presence of a package as
a feature test, I fear that said ship has sailed. Maybe not.
- If you disallow some packages from being installed, then packagers
have a new and exciting decision to make, which they will often
make incorrectly. And anything which comes to depend upon such
a package (even erroneously) will transitively inherit the
uninstallability, since its dependencies cannot be met.
In particular, it means that all group packages need to come in
multiple sets. Image a package that depends on all packages that
are needed for the liveCD image... that package could not be
installed in a zone if some of those packages were "global
zone only".
There are many packages that may choose to have some content
for all types of images, and additional content for zones
and global images...
I don't feel strongly about this either way, though I personally liked
the three different options in SVR4 (install in all zones, install only
in global, or install for real in global and "hollowly" in non-global).
Sometimes it's just easier to do it at the whole package level rather
than at the individual action level.
I do want to point out, though, that it's confusing to have the zone
variant lines at the top of every package if they aren't actually
instructions, especially when you see these two lines together:
set name=variant.arch value=sparc value=i386
set name=variant.zone value=global value=nonglobal
Although identical in format, they don't have the same semantics -- the
first is instructions, the second data only. I don't think it was an
unreasonable assumption that the second would actually be instructions
as well...
Thanks,
Nick
_______________________________________________
pkg-discuss mailing list
[email protected]
http://mail.opensolaris.org/mailman/listinfo/pkg-discuss
