0.8.11.1:

	Fixed bugs in ATANH
	... (atanh #c(1 <anything>)) now computes the right answer;
	... (atanh <large>) now does the right thing too.

diff --git a/BUGS b/BUGS
index f75671b..e41f9e5 100644 (file)
--- a/BUGS
+++ b/BUGS
@@ -1495,19 +1495,6 @@ WORKAROUND:
     when called with arguments
       (NIL).
    
-335: "ATANH completely broken"
-  a. (reported by Peter Graves sbcl-devel 2004-06-01)
-    (atanh #c(1 2)), and more generally atanh of any complex with real
-    part 1, computes entirely the wrong answer.
-  b. (discovered by CSR when investigating a.)
-    (atanh most-positive-double-float), and more generally atanh of any
-    number with magnitude larger than
-    sqrt(most-positive-double-float), computes a number whose real
-    part is the imaginary part of the correct answer, and whose
-    imaginary part is the real part of the correct answer.
-  (fixes for both of these were sent CSR sbcl-devel 2004-06-02, to be merged
-   post-0.8.11)
-
 336: "slot-definitions must retain the generic functions of accessors"
   reported by Tony Martinez:
     (defclass foo () ((bar :reader foo-bar)))

diff --git a/NEWS b/NEWS
index 8bebd46..56a74b6 100644 (file)
--- a/NEWS
+++ b/NEWS
@@ -2518,6 +2518,10 @@ changes in sbcl-0.8.11 relative to sbcl-0.8.10:
     ** ensured that pathnames, where they have a namestring, always
        print using #P"..." syntax.
 
+changes in sbcl-0.8.12 relative to sbcl-0.8.11:
+  * fixed bug 335: ATANH now computes the inverse hyperbolic tangent
+    even for difficult arguments.  (reported by Peter Graves)
+
 planned incompatible changes in 0.8.x:
   * (not done yet, but planned:) When the profiling interface settles
     down, it might impact TRACE. They both encapsulate functions, and

diff --git a/src/code/irrat.lisp b/src/code/irrat.lisp
index e308197..e7131b0 100644 (file)
--- a/src/code/irrat.lisp
+++ b/src/code/irrat.lisp
@@ -879,25 +879,25 @@
        (declare (optimize (speed 3)))
     (cond ((or (> x theta)
               (> (abs y) theta))
-            ;; To avoid overflow...
-          (setf eta (float-sign y half-pi))
-          ;; nu is real part of 1/(x + iy).  This is x/(x^2+y^2),
+          ;; To avoid overflow...
+          (setf nu (float-sign y half-pi))
+          ;; ETA is real part of 1/(x + iy).  This is x/(x^2+y^2),
           ;; which can cause overflow.  Arrange this computation so
           ;; that it won't overflow.
-          (setf nu (let* ((x-bigger (> x (abs y)))
-                          (r (if x-bigger (/ y x) (/ x y)))
-                          (d (+ 1.0d0 (* r r))))
-                     (if x-bigger
-                         (/ (/ x) d)
-                         (/ (/ r y) d)))))
+          (setf eta (let* ((x-bigger (> x (abs y)))
+                           (r (if x-bigger (/ y x) (/ x y)))
+                           (d (+ 1.0d0 (* r r))))
+                      (if x-bigger
+                          (/ (/ x) d)
+                          (/ (/ r y) d)))))
          ((= x 1.0d0)
           ;; Should this be changed so that if y is zero, eta is set
           ;; to +infinity instead of approx 176?  In any case
           ;; tanh(176) is 1.0d0 within working precision.
           (let ((t1 (+ 4d0 (square y)))
                 (t2 (+ (abs y) rho)))
-            (setf eta (log (/ (sqrt (sqrt t1)))
-                           (sqrt t2)))
+            (setf eta (log (/ (sqrt (sqrt t1))
+                              (sqrt t2))))
             (setf nu (* 0.5d0
                         (float-sign y
                                     (+ half-pi (atan (* 0.5d0 t2))))))))

diff --git a/version.lisp-expr b/version.lisp-expr
index 714e7e0..e1f2b9f 100644 (file)
--- a/version.lisp-expr
+++ b/version.lisp-expr
@@ -17,4 +17,4 @@
 ;;; checkins which aren't released. (And occasionally for internal
 ;;; versions, especially for internal versions off the main CVS
 ;;; branch, it gets hairier, e.g. "0.pre7.14.flaky4.13".)
-"0.8.11"
+"0.8.11.1"