Re: Conflicting packages not of extra priority.
On Fri, 5 Feb 1999, Dale Scheetz wrote:
> On Fri, 5 Feb 1999, Santiago Vila wrote:
>
> > On Fri, 5 Feb 1999, Dale Scheetz wrote:
> >
> > > Section 2.2:
> > >
> > > `extra'
> > > This contains packages that conflict with others with higher
> > > priorities, or are only likely to be useful if you already know
> > > what they are or have specialised requirements.
> > >
> > > The paragraph clearly states "higher priorities", not "the same priority",
> > > [...]
> >
> > These paragraphs just explain what do we find in each of the different
> > priorities. They are not worded in a way that they explicitly tell us how
> > we have to "move" things from one priority to another one, this may be
> > derived easily from the definition.
> >
> Derived from what definition?
>From the definition of "extra", of course.
> It seems to me that when you say "This priority contains packages that
> conflict with higher priorities", you are _explicitly_ declaring that the
> only way a package may conflict with a package of higher priority than
> Extra, is to be given an Extra priority.
Exactly.
> This implies that, when assigning
> a priority to a package, the existance of any conflicts other than with
> Extra packages, requires that the priority Extra be assigned to this
> package.
Yes.
> This says that it is OK for a package of priority Optional to
> Conflict with a package of priority Extra, but not with one of any other
> priority.
Exactly.
> This seems a bit rediculous to me...
Mmm, why?
The rationale for this policy is to make the set of
required+important+standard+optional packages a self-consistent set.
Why would this have to be ridiculous? It actually allows the user
to install as many optional packages as he/she wants, without fearing
about conflicts.
> > So, the paragraph about extra packages says that if we look at the extra
> > packages, we should find packages that conflict with others with higher
> > priorities, i.e. we find packages that conflict with required, important,
> > standard or optional packages, or are only likely to be useful etc.
>
> So it is your suggestion that Extra packages may not, according to this
> statement, conflict with other packages of Extra priority? That is
> interesting...
No, I don't suggest that, because clearly this would be impossible
to achieve (think about all the mail-transport-agent packages, all of them
are (or should be) of extra priority but only one (currently exim)
is of important priority.
Also, please note that the conflicts thing is not the only reason why a
package should be extra.
> In addition, you suggest that the sited paragraph says nothing about
> packages with priority other than Extra, and makes no demands on how one
> priority may Conflict with another, unless that package has the priority
> Extra?
>
> Then it is ok for a Required package to conflict with an Optional one, or
> vice versa?
No, because both of them would conflict with a package of priority
higher than extra.
> There is nothing in this paragraph that says that Optional
> packages can't conflict with Optional packages.
> How does this support the professed rationale?
The paragraph says that packages that conflict with others of higher
priority than extra should have extra priority.
Since optional is higher than extra, at least one of the two priorities
would have to be downgraded to extra.
Thanks.
--
"bf42120d2c45a6c7757886531a17134d" (a truly random sig)
Reply to: