Sybren A. Stüvel
63cf0d0890
`BKE_nlastrip_find_active()` and `update_active_strip()` now do a more thorough search for the active strip, by also recursing into meta-strips. There were some assumptions in the NLA code that the active strips is contained in the active track. This assumption turned out to be false: when there is a meta-strip, the active strip could be inside that, and then it's not contained in `active_track.strips` directly. Apart from the above, there are other situations in which the track pointed to by `AnimData::act_track` does *not* contain the active strip (even indirectly). Both these cases can happen when the transform system is moving a strip that's in tweak mode. Entering tweak mode doesn't just search for the active track/strip, but also falls back to the last-selected ones. This means that the track/strip pointers & flags can be out of sync with what's being tweaked/transformed. Because of this, the assumption that there is any relation between "active strip" and "active track" has been removed from the `update_active_strip()` function. All this searching could be prevented by passing the `AnimData` to the code that duplicates tracks & strips. That way, when the active track/strip is dup'ed, the `AnimData` pointers can be updated as well. That would require more refactoring and thus could introduce other bugs, so that's left for later.