This short and well-written paper is a good example of the usefulness of Software--Practice and Experience. No other existing journal would have published it, since it does not present an important advance in any area. Rather, it describes and discusses a small experiment, and draws from it some interesting conclusions.

The experiment is the implementation in C++ of a concordance program for texts that cannot be encoded in any standard one-byte character code. An additional constraint is that the program should run on both Linux and Windows platforms.

I had the opportunity to look at concordance programs many years ago, when they were large applications running for long periods of time and using a lot of magnetic tape. Now, the problem is much simpler, since the entire database and concordance can fit in current memories.

Conversely, dealing with various character sets is now a real concern, since it is no longer considered acceptable to use only capital Roman letters in order to represent languages with additional letters and diacritics. The introduction of localization in languages and systems is relatively recent, and this paper shows that it does not solve all problems.

However, the C++ standard definition considers this problem seriously, and the paper shows that by writing a specific code conversion facet (given as an appendix), it is possible to have exactly the same program working on both the platforms intended.