This change adjusts how ASSERT_NEAR and EXPECT_NEAR treats infinity, such that ASSERT_NEAR(inf, inf, 0) passes. This makes the behavior more consistent with ASSERT_EQ(inf, inf) which succeeds.

Some examples of asserts that now pass:
```
ASSERT_NEAR(inf, inf, 0)
ASSERT_NEAR(-inf, inf, inf)
ASSERT_NEAR(inf, x, inf)  // x is any finite floating point value
```

PiperOrigin-RevId: 685748133
Change-Id: I7b3af377773e8e0031e4c6b86830cbbf76bf20c6

docs/reference/assertions.md

@@ -276,7 +276,8 @@ Units in the Last Place (ULPs). To learn more about ULPs, see the article
 `ASSERT_FLOAT_EQ(`*`val1`*`,`*`val2`*`)`
 
 Verifies that the two `float` values *`val1`* and *`val2`* are approximately
-equal, to within 4 ULPs from each other.
+equal, to within 4 ULPs from each other. Infinity and the largest finite float
+value are considered to be one ULP apart.
 
 ### EXPECT_DOUBLE_EQ {#EXPECT_DOUBLE_EQ}
 
@@ -284,7 +285,8 @@ equal, to within 4 ULPs from each other.
 `ASSERT_DOUBLE_EQ(`*`val1`*`,`*`val2`*`)`
 
 Verifies that the two `double` values *`val1`* and *`val2`* are approximately
-equal, to within 4 ULPs from each other.
+equal, to within 4 ULPs from each other. Infinity and the largest finite double
+value are considered to be one ULP apart.
 
 ### EXPECT_NEAR {#EXPECT_NEAR}
 
@@ -294,6 +296,11 @@ equal, to within 4 ULPs from each other.
 Verifies that the difference between *`val1`* and *`val2`* does not exceed the
 absolute error bound *`abs_error`*.
 
+If *`val`* and *`val2`* are both infinity of the same sign, the difference is
+considered to be 0. Otherwise, if either value is infinity, the difference is
+considered to be infinity. All non-NaN values (including infinity) are
+considered to not exceed an *`abs_error`* of infinity.
+
 ## Exception Assertions {#exceptions}
 
 The following assertions verify that a piece of code throws, or does not throw,