Update README regarding invalid UTF-8

README.md

@@ -55,6 +55,12 @@ In my local benchmarking (Mac laptop), `uax29/words` processes around 16MM token
 
 - There is [discussion](https://groups.google.com/d/msg/golang-nuts/_79vJ65KuXc/B_QgeU6rAgAJ) of implementing the above in Go’s [`x/text`](https://godoc.org/golang.org/x/text) package
 
+### Invalid inputs
+
+Invalid UTF-8 input is considered undefined behavior. That said, we’ve worked to ensure that such inputs will not cause pathological outcomes, such as a panic or infinite loop. Callers should expect “garbage-in, garbage-out”.
+
+There are two basic tests in each package, called `TestInvalidUTF8` and `TestRandomBytes`. Those tests pass, returning the invalid bytes verbatim, without a guarantee as to how they will be segmented.
+
 ### See also
 
 [jargon](https://github.com/clipperhouse/jargon), a text pipelines package for CLI and Go, which consumes this package.

graphemes/README.md

@@ -23,14 +23,20 @@ if err := scanner.Err(); err != nil {
 
 [GoDoc](https://godoc.org/github.com/clipperhouse/uax29/graphemes)
 
+### Conformance
+
+We use the official [test suite](https://unicode.org/reports/tr41/tr41-26.html#Tests29), thanks to [bleve](https://github.com/blevesearch/segment/blob/master/tables_test.go). Status:
+
+![Go](https://github.com/clipperhouse/uax29/workflows/Go/badge.svg)
+
 ### Performance
 
 `uax29` is designed to work in constant memory, regardless of input size. It buffers input and streams tokens. (For example, I am showing a maximum resident size of 8MB when processing a 300MB file, on my laptop.)
 
 Execution time is `O(n)` on input size. It can be I/O bound; you can control I/O and performance implications by the `io.Reader` you pass to `NewScanner`.
 
-### Conformance
+### Invalid inputs
 
-We use the official [test suite](https://unicode.org/reports/tr41/tr41-26.html#Tests29), thanks to [bleve](https://github.com/blevesearch/segment/blob/master/tables_test.go). Status:
+Invalid UTF-8 input is considered undefined behavior. That said, we’ve worked to ensure that such inputs will not cause pathological outcomes, such as a panic or infinite loop. Callers should expect “garbage-in, garbage-out”.
 
-![Go](https://github.com/clipperhouse/uax29/workflows/Go/badge.svg)
+There are two basic tests in each package, called `TestInvalidUTF8` and `TestRandomBytes`. Those tests pass, returning the invalid bytes verbatim, without a guarantee as to how they will be segmented.

sentences/README.md

@@ -23,14 +23,20 @@ if err := scanner.Err(); err != nil {
 
 [GoDoc](https://godoc.org/github.com/clipperhouse/uax29/sentences)
 
+### Conformance
+
+We use the official [test suite](https://unicode.org/reports/tr41/tr41-26.html#Tests29), thanks to [bleve](https://github.com/blevesearch/segment/blob/master/tables_test.go). Status:
+
+![Go](https://github.com/clipperhouse/uax29/workflows/Go/badge.svg)
+
 ### Performance
 
 `uax29` is designed to work in constant memory, regardless of input size. It buffers input and streams tokens.
 
 Execution time is `O(n)` on input size. It can be I/O bound; you can control I/O and performance implications by the `io.Reader` you pass to `NewScanner`.
 
-### Conformance
+### Invalid inputs
 
-We use the official [test suite](https://unicode.org/reports/tr41/tr41-26.html#Tests29), thanks to [bleve](https://github.com/blevesearch/segment/blob/master/tables_test.go). Status:
+Invalid UTF-8 input is considered undefined behavior. That said, we’ve worked to ensure that such inputs will not cause pathological outcomes, such as a panic or infinite loop. Callers should expect “garbage-in, garbage-out”.
 
-![Go](https://github.com/clipperhouse/uax29/workflows/Go/badge.svg)
+There are two basic tests in each package, called `TestInvalidUTF8` and `TestRandomBytes`. Those tests pass, returning the invalid bytes verbatim, without a guarantee as to how they will be segmented.

words/README.md

@@ -36,3 +36,9 @@ We use the official [test suite](https://unicode.org/reports/tr41/tr41-26.html#T
 Execution time is `O(n)` on input size. It can be I/O bound; I/O performance is determined by the `io.Reader` you pass to `NewScanner`.
 
 In my local benchmarking (Mac laptop), `uax29/words` processes around 16MM tokens per second, or 40MB/s, of [wiki text](https://en.wikipedia.org/w/index.php?title=New_York_City&action=edit).
+
+### Invalid inputs
+
+Invalid UTF-8 input is considered undefined behavior. That said, we’ve worked to ensure that such inputs will not cause pathological outcomes, such as a panic or infinite loop. Callers should expect “garbage-in, garbage-out”.
+
+There are two basic tests in each package, called `TestInvalidUTF8` and `TestRandomBytes`. Those tests pass, returning the invalid bytes verbatim, without a guarantee as to how they will be segmented.