unit_tests: fix strtoul unit test #9214
Conversation
That's exactly why the test sets errno. C functions which interact with errno change it when an error occurs (per the man page), and otherwise leave it alone; it would be ridiculous to expect functions to document what they don't do. |
| EXPECT_EQ(0, ul); | ||
| EXPECT_EQ(p + 1, endp); | ||
|
|
||
| p = " \t 0"; | ||
| endp = nullptr; | ||
| ul = std::strtoul(p, const_cast<char**>(&endp), 10); | ||
| EXPECT_EQ(0, errno); | ||
| EXPECT_EQ(0, ul); | ||
| EXPECT_EQ(p + 9, endp); | ||
|
|
||
| p = "q"; |
There was a problem hiding this comment.
This part of the test would be the most likely to break across systems: Per the man page, some systems report EINVAL here: "EINVAL (not in C99) The given base contains an unsupported value." The c++ function makes no mention of EINVAL, so perhaps whoever wrote this test suite wanted a way to determine the actual behavior in a given build.
There was a problem hiding this comment.
I think you meant to put this comment on the test immediately below this one, yeah? According to the C standard, strtoul should skip over whitespace. But yes, I agree with the general point, so I updated the commit to just remove the EXPECT_EQ(0) from the invalid character test.
jeffro256
force-pushed
the
fix_strtoul_test
branch
from
155c924
to
399b884
Compare
errnonot guaranteed to be anything according to C standard, except on range error