Z-level update: lots of overloads

[Coolthulhu]

Quite a bit of overloads. No in-game effect yet.

Includes #11472

Doesn't yet implement everything in #6818 but is close.

[Lain-]

Just stating the obvious here, but I hope this get merged soon, because 1) It's awesome and 2) A big PR like this tend to get unmergeable very quick.

[Zireael07]

Agreed!

[BevapDin]

You could avoid the overriding of lp below by using this neat reference trick:

tripoint lp( u.posx() + u.view_offset_x, ... );
int &lx = lp.x;
int &ly = lp.y;
int &lz = lp.z;

If you write to lx, it will automatically update lp because lx and lp.x are the same place in memory.

[BevapDin]

"cache cache"? Is that something like a cache of a cache, it caches caches? (-:

[Coolthulhu]

So you can lookup your lookups

[BevapDin]

Why not get_submap_at( p, lx, ly )?

[Coolthulhu]

Had a bit of "branching hell" there for a while. Some of the changes are from earlier branches (before I created some of the useful overloads).

[BevapDin]

This looks fine.

---

Does not really belong in this PR, just some things I noticed (should probably addressed, if at all, after this has been merged to prevent conflicts):

Have you noticed that map::has_flag_ter_and_furn is implemented as ter_at.has_flag && furn_at.has_flag while map::has_flag_ter_or_furn is implemented quite differently (look up of the submap only once)? That makes no sense, both function should be the same except for the single || vs && when combining the flag status.

Having the look up of the submap and the calculation of the submap local coordinates lx, ly (and checks for inbounds only once) should be a bit faster than doing it twice.

---

map::tername, map::furnname and map::name should probably return a const reference.

Similar some functions that take a furniture / terrain / trap id string (e.g. map::set) should take that string as const reference.

---

This might reduce the code size a bit: make get_submap_at check inbounds itself and return a nullptr if the input is not inbounds. That way the call can avoid that inbounds check and one should probably check the return value get_submap_at anyway. Instead of:

    if( !inbounds( p ) ) {
        return false; // or whatever
    }
    int lx, ly;
    submap *const current_submap = get_submap_at( p, lx, ly );

One can write:

    int lx, ly;
    submap *const current_submap = get_submap_at( p, lx, ly );
    if( current_submap == nullptr ) {
        return false; // or whatever
    }

Nah, probably not a good idea.

[Zireael07]

Merge please before it gets un-mergeable...