When writer saves a file as RTF, some non-ascii characters in metadata (e.g.
author name) are written literally without hex-escaping, in conflict with the
RTF specification and other software expecting all RTF content to be ascii.

For example, if you (having a new empty writer document):
1. Set first/last name in Tools->Options->User Data to Ján Ľos;
2. Update document metadata using File->Properties->General->Reset (with 'Apply
user data' checked);
3. Save the file as RTF;

you'll end up with this declaration inside the RTF:
{\info{\upr{\author Ján ?os}{\*\ud{\author J\'e1n \u317\'3fos}}} ...

Note that the U+00E1 A WITH ACUTE ACCENT in \upr is encoded as literal byte
0xe1, not as the proper (and later used) escape sequence \'e1, as required for
all characters beyond 7-bit ascii by the RTF specification.

Same issue applies to other (but not all) characters in the codepage used in the
file. 

For another example, using the name Ján Hraško, in locale sk_SK.UTF-8 and all
language settings set to Slovak, results in a file encoded in codepage 1250,
with relevant code:
{\info{\author Ján Hraško} (converted, the bytes are again 0xe1 and 0x9e, above
127 but not escaped).