On Wed, Oct 26, 2016 at 05:38:51PM +0000, Komяpa wrote: > I think, back-patch. > > This is definitely a bug, so if someone was relying on this behavior, it's > better to get them fix their data than to say later "we can't go to next > major PostGIS, we rely on this bug being not fixed". You convinced me. Let's call it a bug. --strk;