On Sun, 2018-08-12 at 09:52 +0200, Rene Engelhard wrote: > > I think you should use colibre *instead* of tango. My understanding > > is > > No, that looks too much Windows'ishy to me. Ah, OK. In this case it might make sense to do an upstream patch to copy these icons to tango, as we already did for a lot of other icons that were copied to tango from galaxy (see tdf#118123). > > that colibre is at least as complete as tango, and is actively > > maintained unlike tango. And tango isn't even part of the fallback > > That is not 100% true, see my paste. > > For the desktop detection icon detection it is. The desktop detection code you cited is about defining the *preferred* icon theme, not about "fallback" - which means what to do when the requested theme isn't found. The fallback for a complete icon theme (as opposed to an individual icon) is done like this: /*static*/ OUString IconThemeSelector::ReturnFallback(const std::vector<IconThemeInfo>& installedThemes) { if (!installedThemes.empty()) { return installedThemes.front().GetThemeId(); } else { return FALLBACK_ICON_THEME_ID; } } ... which means that it will return the first found theme, which might or might not be tango. Imagine a case of a user running gnome or kde (which have other defined themes in IconThemeSelector::GetIconThemeForDesktopEnvironment), while having only tango and colibre. It will get through the above code, and might pick colibre as well, which is not what you want. Maxim