On windows when an octet string ("Binary Data" as displayed by Directory Studio) is being edited as text (double click on attribute then click "Edit as Text..." in the opened hex editor) wrong encoding/code page is used and there is no way to set the encoding to the one that should be used.
The Text Editor that is opened uses most likely CP1250 no matter what the string is encoded in. So an UTF-8 string is mostly uneditable.
I assume the editor uses the default system encoding.
I think this because on Linux the text is displayed/edited correctly, and nowadays Linux's use UTF-8 as default (albeit that would create the same artifacts if the strings were stored with some other encoding like eg. Windows-1252).
The correct behavior for the Text Editor windows would be to honor the global Eclipse workspace setting regarding the encoding of text files.
As about the concrete circumstances to reproduce the error:
I was editing a vacationInfo attribute (OID: 18.104.22.168.4.1.11048.1.1.1.3 type: binary)