I ran a test build of the latest Clojure 1.5 master (372f03e) on Windows and found a number of failures in the "clojure.test-clojure.pprint" tests. All of these seem to be caused by incorrect assumptions about line endings. Example:

[java] {:clojure.test/vars (ns-macro-test),
[java] :thread/name "main",
[java] :pid 1528,
[java] :thread 1,
[java] :type :assert/fail,
[java] :level :warn,
[java] :test/actual
[java] (not
[java] (clojure.core/=
[java] "(ns slam.hound.stitch\r\n (:use [slam.hound.prettify :only [prettify]]))\r\n"
[java] "(ns slam.hound.stitch\n (:use [slam.hound.prettify :only [prettify]]))\n")),
[java] :test/expected
[java] (clojure.core/=
[java] (clojure.core/with-out-str
[java] (clojure.pprint/with-pprint-dispatch
[java] clojure.pprint/code-dispatch
[java] (clojure.pprint/pprint
[java] (clojure.core/read-string
[java] "(ns slam.hound.stitch\n (:use [slam.hound.prettify :only [prettify]]))"))))
[java] (clojure.core/str
[java] "(ns slam.hound.stitch\n (:use [slam.hound.prettify :only [prettify]]))"
[java] "\n")),
[java] :line 173,
[java] :tstamp 1355113319212,
[java] :file "test_pretty.clj"}

It isn't totally clear what the right behaviour should be: should pprint be producing platform specific line endings or not? Either way, the test should confirm the expected behaviour.