I know, that is what I was planning on doing. (and might be what I end up doing anyway since I need to truncate the utf-8 string if it is > 1024 bytes...) Still it is nice to have other options - this code is run in some tight loops so I will be profiling all the options.