# HG changeset patch
# User drewp@bigasterisk.com
# Date 1559271068 0
# Node ID bd0e24f6254aecd09cc5fe1c417314fcfa619afd
# Parent  9b46968b8881d8aa311509412f750255c1592421
doc update on SyncedGraph
Ignore-this: cec485f067fbde38bd99766e1c57eb68

diff -r 9b46968b8881 -r bd0e24f6254a rdfdb/syncedgraph.py
--- a/rdfdb/syncedgraph.py	Thu May 30 08:38:13 2019 +0000
+++ b/rdfdb/syncedgraph.py	Fri May 31 02:51:08 2019 +0000
@@ -42,10 +42,16 @@
 class SyncedGraph(CurrentStateGraphApi, AutoDepGraphApi, GraphEditApi):
     """
     graph for clients to use. Changes are synced with the master graph
-    in the rdfdb process.
+    in the rdfdb process. 
+
+    self.patch(p: Patch) is the only way to write to the graph.
 
-    This api is like rdflib.Graph but it can also call you back when
-    there are graph changes to the parts you previously read.
+    Reading can be done with the AutoDepGraphApi methods which set up
+    watchers to call you back when the results of the read have
+    changed (like knockoutjs). Or you can read with
+    CurrentStateGraphApi which doesn't have watchers, but you have to
+    opt into using it so it's clear you aren't in an auto-dep context
+    and meant to set up watchers.
 
     You may want to attach to self.initiallySynced deferred so you
     don't attempt patches before we've heard the initial contents of
@@ -54,8 +60,7 @@
     statements unless we have the correct graph.
 
     If we get out of sync, we abandon our local graph (even any
-    pending local changes) and get the data again from the
-    server.
+    pending local changes) and get the data again from the server.
     """
 
     def __init__(self,