Opened 7 years ago
Closed 5 years ago
#11769 closed Bugs (fixed)
64-bit hash_combine has weakness against zero input
| Reported by: | Andrey Semashev | Owned by: | Daniel James |
|---|---|---|---|
| Milestone: | To Be Determined | Component: | hash |
| Version: | Boost Development Trunk | Severity: | Problem |
| Keywords: | Cc: |
Description (last modified by )
The hash_combine_impl function for 64-bit size_t (see here) does not have an additive component. As a result, hashing strings of zeros always results in zero hash value regardless of the length of the string.
Change History (3)
comment:1 by , 7 years ago
| Description: | modified (diff) |
|---|
comment:2 by , 7 years ago
| Status: | new → assigned |
|---|
comment:3 by , 5 years ago
| Resolution: | → fixed |
|---|---|
| Status: | assigned → closed |
Note:
See TracTickets
for help on using tickets.
You're right about the function but it turns out that there's no real problem for an embarrassing reason - that function has a spurious template argument so it's never used.
There are unit tests for this exact problem (in
hash_sequence_test.hpp), and I'm relieved to say that they would have been caught the problem if that function was actually used, so at least I got that right.I think I'll fix the function and put it to use. I had originally intended to do more rigorous testing of hash quality before releasing it, but ended up having little time. I think I'll be more cautious this time, so the new version might not make it into the next version of boost.