summaryrefslogtreecommitdiff
diff options
context:
space:
mode:
authorThomas Haller <thaller@redhat.com>2019-10-21 18:37:42 +0200
committerThomas Haller <thaller@redhat.com>2019-10-21 18:37:42 +0200
commit4c5795596e392b54ac7d7e4b51bf2d4e8d8f4c75 (patch)
tree971e3714bc3aaba7c257b719c43725f017db7974
parentd188c61cab48250c01f8d0f7bf22e60c46133990 (diff)
downloadNetworkManager-th/libnm-hide-gobject-structs.tar.gz
fixup! libnm: hide GObject structs from public API and embed private datath/libnm-hide-gobject-structs
-rw-r--r--NEWS12
1 files changed, 2 insertions, 10 deletions
diff --git a/NEWS b/NEWS
index c865d5ce94..bb185c87dc 100644
--- a/NEWS
+++ b/NEWS
@@ -18,16 +18,8 @@ USE AT YOUR OWN RISK. NOT RECOMMENDED FOR PRODUCTION USE!
which delays reaching "connected" state for the address family accordingly.
* Various bug fixes and improvements.
* libnm: hide NMObject and NMClient typedefs from header files. This prevents
- the user from subclassing these types, but that was never supported nor
- intended anyway. For NMObject and subclasses that is because these types
- are instantiated by NMClient and not directly by the user. Instantiating
- those types aside NMClient does not give something useful and neither does
- subclassing the types make sense. Regarding NMClient, the user instantiates
- the type and maybe could have subclassed it. However, it doesn't make much
- sense either, because the type does not expect the oser to override internal
- workings. As so often, it's better to delegate instead of subclassing, and
- libnm no longer allows you to subclass these types. We don't expect anybody
- to be hit by this change.
+ the user from subclassing these types and is an ABI change (in case somebody
+ was doing so).
=============================================
NetworkManager-1.20