HONEYCOMB-359 - Wildcarded writers

Adds option to specify subtree writer that can handle whole subtree
of nodes without having whole subtree specified. Its checking
if node is children at runtime, rather than having pre-computed tree

Change-Id: Ic46f2bd6de84f0dd14865825399f5a90a1f80859
Signed-off-by: Jan Srnicek <jsrnicek@cisco.com>

1 files changed, 0 insertions, 8 deletions

diff --git a/infra/northbound/bgp/src/main/java/io/fd/honeycomb/infra/bgp/ApplicationRibWriterFactory.java b/infra/northbound/bgp/src/main/java/io/fd/honeycomb/infra/bgp/ApplicationRibWriterFactory.java
index 02f1788f3..00a12e7ac 100644
--- a/infra/northbound/bgp/src/main/java/io/fd/honeycomb/infra/bgp/ApplicationRibWriterFactory.java
+++ b/infra/northbound/bgp/src/main/java/io/fd/honeycomb/infra/bgp/ApplicationRibWriterFactory.java
@@ -45,14 +45,6 @@ final class ApplicationRibWriterFactory implements WriterFactory {
             InstanceIdentifier.create(ApplicationRib.class);
     private static final InstanceIdentifier<Tables> TABLES_IID = AR_IID.child(Tables.class);
 
-
-    // TODO (HONEYCOMB-359):
-    // BGP models are huge, we need some kind of wildcarded subtree writer, that works for whole subtree.
-    // 1) we can either move checking handledTypes to writers (getHandledTypes, isAffected, writer.getHandedTypes, ...)
-    // but then precondition check in flatWriterRegistry might be slower (we need to check if we have all writers
-    // in order to avoid unnecessary reverts).
-    //
-    // 2) alternative is to compute all child nodes during initialization (might introduce some footprint penalty).
     @Override
     public void init(final ModifiableWriterRegistryBuilder registry) {
         registry.subtreeAdd(ImmutableSet.of(TABLES_IID), new BindingBrokerWriter<>(InstanceIdentifier.create(ApplicationRib.class), dataBroker)