Add expm1 test case to cover the k==1024 case #4508
Conversation
|
What is 0x1p1023 and why does it need extra coverage? Furthermore, why is the negated value not tested? |
Refer to jerryscript/jerry-math/expm1.c Line 268 in 8e83638 |
There was a problem hiding this comment.
In addition to the change requested in the code, please, update commit title and message as well.
Current commit title is misleading, as it is hard to cover a literal/number, unless it is the input value to the function. But it is not the case here. The test case aims at covering the k==1024 case. So, the title should read "Add expm1 test case to cover the k==1024 case".
As for the commit message, don't include the discussion in there. Use comments to answer questions. The commit message should describe what has been done, and why. But rarely something like "refer to ..."
tools/unit-tests/gen-test-math.c
Outdated
| @@ -404,6 +404,7 @@ main (int argc, char **args) | |||
| GEN_DBL_TEST (expm1 (INFINITY)); | |||
| GEN_DBL_TEST (expm1 (-INFINITY)); | |||
| GEN_DBL_TEST (expm1 (NAN)); | |||
| GEN_DBL_TEST (expm1 (7.095e+02)); | |||
There was a problem hiding this comment.
move this one line down, between 708 and 710 (and re-generate the .inc.h)
akosthekiss
added
jerry-libm
lygstate
changed the title
|
BTW, kind of answer-to-self: 0x1p1023 is a hexadecimal floating point literal with the value of |
-7.095e+02 won't trigger the k==1024 case, so we didn't add it into the test list JerryScript-DCO-1.0-Signed-off-by: Yonggang Luo [email protected]
Refer to
jerryscript/jerry-math/expm1.c
Line 268 in 8e83638
JerryScript-DCO-1.0-Signed-off-by: Yonggang Luo [email protected]