[ZOOKEEPER-3915] Digest hash - use UTF_8 explicitly when converting path String to bytes #1435
Conversation
…y - do not rely on the default locale.
|
|
||
| CRC32 crc = new CRC32(); | ||
| crc.update(path.getBytes()); | ||
| crc.update(path.getBytes(StandardCharsets.UTF_8)); |
There was a problem hiding this comment.
FWIW, I find it convenient to use static imports for UTF_8, as it makes the code a bit nicer to read.
|
Shall we make this value configurable? What will happen to existing clusters that are using a different encoding? |
eolivelli
left a comment
eolivelli
left a comment
There was a problem hiding this comment.
Thanks for contributing your fix and pointing out this issue.
I left one question.
We should also add a minimal unit test in order to prevent regressions on the future
The use of the old method is already configurable... via the user's locale settings. I'm not sure how this method is used, but if the result is serialized anywhere within ZooKeeper for reuse later, ZK itself should define the encoding explicitly. That's good practice for serializing... but I don't know enough about the code to know how it is used. Just throwing in my 2 cents for consideration. 😺 |
3 participants
Fixes a false Digest mismatch when participants are running in environments with different default locales.