# HG changeset patch
# User Steve Losh <steve@stevelosh.com>
# Date 1468246594 0
# Node ID 422cf6c3909f5225de8e89211cdc32502952b466
# Parent  54b199ec2b11316647bce74e832883337404c6fa
Update

diff -r 54b199ec2b11 -r 422cf6c3909f README.markdown
--- a/README.markdown	Sun Jul 10 15:31:23 2016 +0000
+++ b/README.markdown	Mon Jul 11 14:16:34 2016 +0000
@@ -562,6 +562,8 @@
 * Poked a bit and pushing the constant optimization down into the compilation
   phase, but ran into a bit of a wall so I threw away the diff.  Might take
   another stab at it later once I let it roll around in my head a bit.
+* Mostly refactored the Bones code store to be a `simple-array`, after profiling
+  indicates that most of the runtime is spent in data vector refs.
 
 #### Splitting the Bones Store
 
@@ -697,3 +699,8 @@
 modern implementations support hash tables with weak values, so if we use that
 for the index I think it would fix the problem.  We'd be able to map objects to
 fixnums without interfering with GC and leaking memory.
+
+### 2016-07-11
+
+* Finished up the [Bones][] code store refactor and a couple of other minor
+  things.