Hi, I see gnulib use gettext(), but it does not setup its own domain, which means it'll use the main program's domain. I don't see how this works because then every program that uses gnulib must run xgettext on gnulib source code, then translators may have to translat the same strings over again for each program.
Should gnulib generate .pot and maintain its own translations? If gnulib does not have a separate gettext domain, fine, each program's makefile can merge gnulib's .po files into their own. -- Duy
