R devel - Mar 2010 - Suggestion: Not having to export .conflicts.OK in name spaces

Currently library() and attach() fail to locate an existing
'.conflicts.OK' in a package wit name space, unless it is exported.
Since there should be little interest in exporting '.conflicts.OK'
otherwise, one may argue that those methods should look for
'.conflicts.OK' even if it is not exported.  If so, a patch for
library() is:
>svn diff library.R
Index: library.R
==================================================================--- library.R 
(revision 51296)
+++ library.R   (working copy)
@@ -312,7 +312,7 @@
                     nogenerics <-
                         !.isMethodsDispatchOn() || checkNoGenerics(env, package
)
                     if(warn.conflicts &&
-                       !exists(".conflicts.OK", envir = env, inherits
= FALSE))

+                       !exists(".conflicts.OK", envir = ns, inherits
= FALSE))
                         checkConflicts(package, pkgname, pkgpath,
                                        nogenerics, ns)
                     runUserHook(package, pkgpath)


In order to be consistent, a patch for attach() should also be
provided, which requires some more changes, but I won't spend time on
that unless the above is a wanted feature.

/Henrik

On 3/17/10 9:11 AM, Henrik Bengtsson wrote:
> Currently library() and attach() fail to locate an existing
> '.conflicts.OK' in a package wit name space, unless it is exported.
> Since there should be little interest in exporting '.conflicts.OK'
> otherwise, one may argue that those methods should look for
> '.conflicts.OK' even if it is not exported.
I guess I agree that there is no real value in forcing .conflicts.OK to 
be exported.  OTOH, this seems like a dubious feature to begin.  When is 
it a good idea to use it?

+ seth

-- 
Seth Falcon | @sfalcon | http://userprimary.net/