update README for new acascii

README.md

@@ -2,13 +2,14 @@
 
 [![GoDoc](https://godoc.org/github.com/signalsciences/ac?status.svg)](https://godoc.org/github.com/signalsciences/ac) [![Build Status](https://travis-ci.org/signalsciences/ac.svg?branch=master)](https://travis-ci.org/signalsciences/ac)
 
-Golang implementation of Aho-Corasick for rapid substring matching on byte
-strings.
+Golang implementation of Aho-Corasick for rapid substring matching on either byte
+strings or ASCII strings.
 
 This is based on the excellent library
 [cloudflare/ahocorasick](https://github.com/cloudflare/ahocorasick) (BSD
 License).  The fork/changes were needed for a specific application usages
-that are incomptabile with the original library.
+that are incomptabile with the original library.  Some other minor optimizations 
+around memory and setup were also done.
 
 ## Examples
 
@@ -40,16 +41,17 @@ Output:
 true
 ```
 
-## NOTES
+## ac/acascii for pure ASCII matching
+
+The `ac/acascii` package assumes the dictionary is all ASCII characters (1-127) without NULL bytes.  This results in during setup:
+
+* 50% less memory allocations
+* 50% less memory users
+* 50% less CPU time
+
+as compared to the plain `ac` package.
 
-* This is designed for ASCII pattern matching at the byte level.
-  There is no rune support, no UTF-8 support (other than the ASCII subset).
-* Similar API to `regexp` package.
-* Byte and String-based API work the same.  Again, there is no UTF-8 support.
 
 ## IN PROGRESS
 
-* Current API allows for *overlapping* matches.  This is slow
-  and potentially exponential and is different than how golang's
-  regular expressions work.  This will be changed.
 * Support for ASCII case-insensitive matching.