I hold that the behaviour is not incorrect. Ideally there would be an error message to say that you have two rules for the same thing. That is hard to do at present, but will be the case in cfengine 3. That way it becomes the responsibility of the maintainer to resolve conflicts.
M On Sun, 2005-11-13 at 20:22 -0600, Brendan Strejcek wrote: > Mark Burgess wrote: > > > Brendan you have > > > > > > > bash > > > > > pkgmgr=rpm > > > > > define=bash_pkg_installed > > > > > > > > > bash > > > > > pkgmgr=rpm > > > > > define=bash_package_installed > > > > i.e. the "do part" is the same in both cases, so both defines would > > always be true, except that cfengine thinks (rightly I believe) > > that it is unnecessary to do the same thing twice, so only one gets > > defined. There are no other classes that I can see... > > One says: if bash is installed, define the class "foo". The other says: > if bash is installed, define the class "bar" (class names changed from > the example above for clarity). Since foo != bar, they are not the same > thing, and should not be treated as such by locks. > > The practical downside to the current behavior is that, as an admin > interacting with a collection of cfengine policy files, to be able to > use one of the above class definitions with certainty, I have to examine > every other imported file and make sure someone else is not defining a > different class based on the same package. > > Best, > Brendan > > -- > Senior System Administrator > The University of Chicago > Department of Computer Science > > http://www.cs.uchicago.edu/people/brendan > > http://people.cs.uchicago.edu/~brendan _______________________________________________ Help-cfengine mailing list [email protected] http://lists.gnu.org/mailman/listinfo/help-cfengine
