Problem with Cyrillic in messages Openfire 4.8.0

Good afternoon!

We have updated Openfire from version 4.0.4 to the latest version 4.8.0.

Unfortunately, after the update we found out a problem. In any place in the message, one Cyrillic character is replaced by two question marks.

We did not find any pattern. When sending the same message several times in a row, the recipient’s question marks appear in different places.

This is very similar to the problem appeared in 2007 (Problem with cyrillic in messages)

There is no way to implement the recommendations for downgrading a seventeen-year-old library.

We have achieved a downgrade of the Openfire server version to 4.7.5 while maintaining the same settings, users and messages.

And it doesn’t have this problem!

All characters in messages arrive correctly!

Unfortunately the client refuses to stay on version 4.7.5 and wants to work only on version 4.8.0.

We kindly ask you for support in solving the problem on version 4.8.0

Thank you for reporting this issue. I have created a new ticket for it: [OF-2814] - Ignite Realtime Jira

The characteristics suggest that Openfire 4.8.0 does not properly parse multi-byte characters. Data is processed in chunks. When the boundary between two such chunks happens to be in the middle of two bytes that make up one character, an issue like the one you’re describing can occur.