How is this supposed to work? The underscore is a convenience macro for the gettext localization function. But gettext works by looking up its string argument in the localization database, and here the string varies from one platform to another, so how can you be sure that the localization will be found?

I guess the answer is that you have to redo the localization for each platform that has a different expansion of PRIuMAX. But that's horrible: the localization will be identical except for the format specifiers, which will just be copied from the English version of the string. Is there support for automating this in the translation toolchain?