Improve libm-test.inc comment

["Joseph S. Myers" <joseph at codesourcery dot com>]

libm-test.inc has a comment describing what is and isn't tested.  It turns 
out the list of tested functions was incorrect - some functions / macros 
listed as tested actually have no testcases, while others not mentioned 
are in fact tested.  I've committed this patch to correct the list and 
also note the lack of testing signs of NaNs.

(Regarding the lack of testing functions aliasing other functions, the 
comments mentioning being aliases are intended to indicate such testing is 
lower priority - but we do have gamma_test as well as lgamma_test, and 
similarly we should probably have some tests for the other aliasing 
functions, since they *are* public interfaces to glibc.)

diff --git a/ChangeLog b/ChangeLog
index 8da19b5..3d3fa6d 100644
--- a/ChangeLog
+++ b/ChangeLog
@@ -1,5 +1,13 @@
 2012-10-05  Joseph Myers  <joseph@codesourcery.com>
 
+	* math/libm-test.inc: List nexttoward, cimag, clog10, conf and
+	creal in comment listing functions tested.  List finite, isinf,
+	isnan, isless, islessequal, isgreater, isgreaterequal,
+	islessgreater, isunordered, lgamma_r and pow10 as functions and
+	macros not tested.  Mention which functions not tested are aliases
+	for other functions.  Fix typo.  Note that signs of NaNs are not
+	tested.
+
 	* scripts/config.guess: Update from config.git.
 	* scripts/config.sub: Likewise.
 
diff --git a/math/libm-test.inc b/math/libm-test.inc
index bed8fc6..6cc0407 100644
--- a/math/libm-test.inc
+++ b/math/libm-test.inc
@@ -42,11 +42,10 @@
    cbrt, ceil, copysign, cos, cosh, erf, erfc, exp, exp10, exp2, expm1,
    fabs, fdim, floor, fma, fmax, fmin, fmod, fpclassify,
    frexp, gamma, hypot,
-   ilogb, isfinite, isinf, isnan, isnormal,
-   isless, islessequal, isgreater, isgreaterequal, islessgreater, isunordered,
+   ilogb, isfinite, isnormal,
    j0, j1, jn,
    ldexp, lgamma, log, log10, log1p, log2, logb,
-   modf, nearbyint, nextafter,
+   modf, nearbyint, nextafter, nexttoward,
    pow, remainder, remquo, rint, lrint, llrint,
    round, lround, llround,
    scalb, scalbn, scalbln, signbit, sin, sincos, sinh, sqrt, tan, tanh, tgamma, trunc,
@@ -54,10 +53,17 @@
 
    and for the following complex math functions:
    cabs, cacos, cacosh, carg, casin, casinh, catan, catanh,
-   ccos, ccosh, cexp, clog, cpow, cproj, csin, csinh, csqrt, ctan, ctanh.
+   ccos, ccosh, cexp, cimag, clog, clog10, conj, cpow, cproj, creal,
+   csin, csinh, csqrt, ctan, ctanh.
 
-   At the moment the following functions aren't tested:
-   drem, nan
+   At the moment the following functions and macros aren't tested:
+   drem (alias for remainder),
+   finite (functions equivalent to isfinite macro),
+   isinf, isnan,
+   isless, islessequal, isgreater, isgreaterequal, islessgreater, isunordered,
+   lgamma_r,
+   nan,
+   pow10 (alias for exp10).
 
    Parameter handling is primitive in the moment:
    --verbose=[0..3] for different levels of output:
@@ -83,8 +89,9 @@
    aren't checked at the moment.
 
    NaN values: There exist signalling and quiet NaNs.  This implementation
-   only uses quiet NaN as parameter but does not differenciate
-   between the two kinds of NaNs as result.
+   only uses quiet NaN as parameter but does not differentiate
+   between the two kinds of NaNs as result.  Where the sign of a NaN is
+   significant, this is not tested.
 
    Inline functions: Inlining functions should give an improvement in
    speed - but not in precission.  The inlined functions return

-- 
Joseph S. Myers
joseph@codesourcery.com