Time goes by, and I have now sought sponsorship in Ubuntu for my proposed changes, including dropping the hunspell-gl binary.
Can't help wondering, though, if a better long-term solution would be possible. Rene, you called my attention to the fact that LO ships its own set of locale data, and that it does not include any el_CY locale. I then noticed that if I start LO with the el_CY.UTF-8 locale, el_GR.{aff,dic} and hyph_el_GR.dic are recognized. LO notices that it's a "Greek" locale without bothering about the country code. However, if I start LO with the it_CH.UTF-8 locale, it_IT.{aff,dic} and hyph_it_IT.dic are not recognized. In this case LO does care about the country code, and look for (the non-existing) it_CH.{aff,dic} and hyph_it_CH.dic. So, what's the difference? $ ls -1 i18npool/source/localedata/data/el_* i18npool/source/localedata/data/el_GR.xml $ ls -1 i18npool/source/localedata/data/it_* i18npool/source/localedata/data/it_CH.xml i18npool/source/localedata/data/it_IT.xml While there is only one Greek LO locale, there are two Italian LO locales. Is that the simple explanation? If it is, would it be possible to patch LO somehow to behave differently? -- Gunnar Hjalmarsson https://launchpad.net/~gunnarhj