Hacks Versus Designs

I remember back in the day, when computers and programming were first becoming somewhat ‘cool’. Back then, the coolest thing you could be in the computing world was a ‘hacker’. Hackers were awesome renegades who could tear down opposing systems using nothing but their superior intellect. Being able to hack was one of the best skills you could have. Now, after studying and working in computer science for a while, the term ‘hack’ has taken on a very negative connotation.

When you’re writing code, there are several things that you’re aiming for. The two broadest and most important of these are:

  1. The code works – it does what you intend for it to do.
  2. The code is good – it’s efficient, understandable, and easy to use/improve.

It seems like these two things would go hand-in-hand, and for well-designed code, that is often the case. However, the road to that ‘well-designed’ code is often fraught with terrible, terrible code. So, what’s the intrinsic difference between these two goals, and how does it lead to bad code?

 

boolean

Hack First, Ask Questions Later

When you’re working on a large, intricate system, and you need to add something or make a change, these two goals lead to two different types of results – a hack, or a design:

  • A hack is a piece of code with only the first goal in mind – you’re just trying to ‘make it work’. You don’t want to put a lot of thought or time into the implementation, you just want it to work.
  • A design has both goals in mind – you’re spending time to come up with a good solution. You’re willing to work a little harder to end up with a more robust, long-loving solution.

Designing a solution to a task leaves you with good code. It’s easy to understand, easy to use, and easy to update. The algorithm makes sense, not just in terms of “does it do what I want”, but also in terms of “does it make sense with the theory behind it”.

Looking at these two descriptions, it’s pretty easy to see – designs are better than hacks. So why would anyone ever want to use a hack to get something done? There are a few reasons.

Designs take more time. You have to come up with a solution, consider its long-term viability, consider how it will interact with every part of your system, present and future, tweak it accordingly, and make sure that it still matches the theory of your application. A hack, on the other hand, involves simply coming up with a quick and dirty solution, and implementing it.

Designs require deeper understanding. In order to fully understand the impact of your newly-designed code, you have to completely understand the current state of your application, remember all of the assumptions you made when coding it, and ensure that your new stuff won’t interfere with any existing stuff (Note that this is much harder to do on a larger team, as there are areas of the code you may not be as familiar with).

Designs are often much larger in scope. When designing a solution, it will often involve creating a ‘system’ or ‘engine’ of sorts. Not only does this take longer to think through and implement, but it also opens the door to a lot of subtle interactions between systems. Hacks are (usually) much more localized – “I’m gonna make this hack here, but I won’t use it in other places”.

You don’t want to spend a lot of effort on code that will be replaced eventually. This is really just a combination of the above points, but it’s an important reason why hacks exist. If you have to update a small piece of code, but you know that you’re going to come in and change the whole thing next month anyways, why would you put a lot of time and effort into designing a solution when a quick, hacky fix will do the trick?

Looks about right Cropped

This is what happens when you leave hacks in your code!

 

Here’s An Example

Let’s say you’re you’re working on a pretty simple game in a pretty simple game engine, using a pretty simple programming language (hint: this means I’ll be using pseudo-code rather than real code). You’ve got your character on the screen, and you want to make him move back and forth along some flat ground whenever you hit an arrow key. You might start out with something like this:

if (keys.leftArrow) {
  dudeGuy.position.x -= 10;
}

if (keys.rightArrow) {
  dudeGuy.position.x += 10;
}

Pretty simple and straightforward – if you’re pressing the left arrow key, move your dudeGuy to the left, and if you’re pressing the right arrow key, move him to the right.

So, you use this code for your movement, and it works, and you continue working on your game. Then, suddenly, you have an epiphany – what if your dudeGuy could jump? You add a variable and hook it up:

int jumpingTimer = 0;

...

if (keys.spaceBar && jumpingTimer == 0) {
  dudeGuy.position.y += 30;
  jumpingTimer = 3;
}

if (jumpingTimer > 0) {
  dudeGuy.position.y -= 10;
  jumpingTimer--;
}

As you continue making your game, you design some levels where you realize that you want the gravity to be less strong, so you have to account for that:

float gravity = 10;

...

if (keys.spaceBar && jumpingTImer == 0) {
  dudeGuy.position.y += 30;
  jumpingTimer = 30 / gravity;
}

if (jumpingTimer > 0) {
  dudeGuy.position.y -= gravity;
  jumpingTimer--;
}

Then you realize that your back-and-forth movement looks pretty choppy, so you decide to add some ‘smoothing’, so your dudeGuy speeds up and slows down:

int movingLeftTimer = 0;
int movingRightTimer = 0;
int jumpingUpTimer = 0;
int jumpingTimer = 0;
float gravity = 10;

...

if (keys.leftArrow) {
  if (movingLeftTimer < 3) {
    movingLeftTimer++;
  }
} else if (movingLeftTimer > 0) {
  movingLeftTimer--;
}

if (movingLeftTimer > 0) {
  dudeGuy.position.x -= 10 / (4 - movingLeftTimer);
}

if (keys.rightArrow) {
  if (movingRightTimer < 3) {
    movingRightTimer++;
  }
} else if (movingRightTimer > 0) {
  movingRightTimer--;
}

if (movingRightTimer > 0) {
  dudeGuy.position.x += 10 / (4 - movingRightTimer);
}

if (keys.spaceBar && jumpingTimer == 0) {
 dudeGuy.position.y += 30;
 jumpingTimer = 30 / gravity;
}

if (jumpingTimer > 0) {
 dudeGuy.position.y -= gravity;
 jumpingTimer--;
}

And,  before you know it, with only a few changes to what we were trying to do, we end up with a piece of code that’s incredibly messy, almost impossible to understand, and prone to bugs and off-by-one errors. Honestly, I just wrote this thing, and I have no idea what it’s supposed to be doing.

Now, this example is a bit of an esoteric one, just to prove a point. However, it is definitely not the worst code I’ve ever seen (or written), and that’s saying something. What should we have written instead? Well, if you couldn’t guess, the above code is an example of a hack (or a number of hacks put together). Rather than examining what it was we needed in the long run, we repeatedly implemented something that did the job in the short term. So, let’s make a design for this use-case, and think about what we need overall.

We want to be able to move left/right, jump, have different values for gravity, and have smoothing on our movement. This sounds a bit like actual physics, so lets steal some important concepts from them – acceleration and deceleration. We’ll determine some rules that match our design, modify the dudeGuy’s acceleration in each direction based on those rules, and then move his position all at once:

float maxSpeed = 10;
float acceleration = 3;
float jumpAcceleration = 10;
float gravity = 3;
float friction = 5;
float minY = 0;

float vx = 0;
float vy = 0;

...

// If the left arrow key is down, accelerate to the left
if (keys.leftArrow) {
  vx -= acceleration;
}

// If the right arrow key is down, accelerate to the right
if (keys.rightArrow) {
  vx += acceleration;
}

// If the spacebar is down and the dudeGuy is on the ground, accelerate upwards
if (keys.spaceBar && dudeGuy.position.y == minY) {
  vy += jumpAcceleration;
}

// Accelerate downwards for gravity
vy -= gravity;

// Decelerate for friction
if (vx > 0) {
  vx -= friction;
} else if (vx < 0) {
  vx += friction;
}

// If we're going to fast to the right, slow us down to the max speed
if (vx > maxSpeed) {
  vx = maxSpeed;
}

// If we're going to fast to the left, slow us down to the max speed
if (vx < -maxSpeed) {
  vx = -maxSpeed;
}

// Update the dudeGuy's position based on our current velocity in each direction
dudeGuy.position.x += vx;
dudeGuy.position.y += vy;

// If the dudeGuy is below the ground, move him up to ground level
if (dudeGuy.position.y < minY) {
  dudeGuy.position.y = minY;
  vy = 0;
}

While we have a similar number of lines of code here, it’s much clearer what’s happening on each line. Every block serves an easy-to-understand purpose, and making changes to the ‘rules’ of movement is very easy. There are a lot of different ways to improve this code, depending on your game’s overall design, but this is a decent, and most importantly simple, place to start.

Another important feature of this piece of code is that it is well documented. Every block is pretty small, but it still has a comment describing the purpose of the block. This is an extremely important part of programming in the context of larger systems – you want to make sure that you (or anyone else) can quickly understand what your code is doing, especially in complex cases. Even though some complex logic might seem simple to you, it’ll definitely seem more difficult when you come back to it in 6 months!

 

A Necessary Evil

Unfortunately, hacks are a necessary evil. While I would love to only ever have to deal with and implement beautifully-designed code, that world doesn’t exist. There’s always a timeline, there are always changing assumptions and new features, and there’s always someone who wants it to be finished yesterday. Inevitably, you’re going to have to write some code quickly, implement a feature that’s likely to change, or come up with a simple ‘solution’ to a difficult problem. In cases like this, you’re forced to use a hack.

Hack

I mean, it works… technically…

It’s not all bad, though. While hacks in general are pretty bad, they can be manageable if you make sure to use them correctly. In fact, I would be willing to bet that any system currently in production (of a certain size) contains quite a few hacks. There are certain qualities that hacks can have which make them a little bit more manageable, and you should try to aim for them whenever you find yourself implementing a hack:

  • Understandable – It’s important that, whatever your hack is, anyone else looking at the code can understand what you were trying to do, and how your hack works. This means leaving a lot of comments around your hack, as well as simplifying the logic as much as possible.
  • Localized – If you have to hack something in, you want it to only be in one place. Every time that code path is used, there’s a chance that something will go wrong. If your hack only touches a small part of your system, then its negative effects will be much less noticeable. This means that frequently-used code paths should never really have hacks in them, while hacks in rarely-used code paths are more acceptable.
  • Known – This is, to me, the most important part of making a hack. If you hack something in and then forget about it, when your system starts failing, you won’t know where to look. If you make sure you remember it (by writing it down somewhere and then telling every person you know), then you’ll know where to look if something goes wrong. On top of that, you’ll always have that hack in the back of your mind, so you’ll be more likely to think of a good design to replace it.

If you follow these guidelines and make sure to try to go back and fix them, then putting hacks into your code won’t end up destroying you.

I hope this was helpful to those of you just starting out in game development – or anything which involves designing complex systems! For those of you who already know a little something about computer science, I hope this at least reinforced your burning hatred of hacks!

 

= = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =

If you want to know more about how to deal with hacky code, or what kind of hacks are in Where Shadows Slumber so that you can exploit them, feel free to contact us! You can always find out more about our game at WhereShadowsSlumber.com, find us on Twitter (@GameRevenant), Facebookitch.io, or Twitch, and feel free to email us directly with any questions or feedback at contact@GameRevenant.com.

Jack Kelly is the head developer and designer for Where Shadows Slumber.

Advertisements

Fake It Til You Make It

Alright people. There’s something that’s been bothering me for a while, and I think it’s time we come clean. For the last two years, ever since the first prototype of Where Shadows Slumber, Frank and I have been lying to you.

“What?!” you ask incredulously, affronted dignity ablaze. “How can this be? What have you lied to us about? I must know!”

Well, never fear – we’ve never lied about the game. All of our screenshots are from the actual game, we’ve represented our progress pretty accurately, and we love you as much as we always have. Rather, inside the game, within Obe’s world, basically everything is a lie.

 

2

The video game Hydrophobia was criticized for focusing too much on its water physics and not enough on other forms of gameplay.

Faker!

This phenomenon is not unique to Where Shadows Slumber – in fact, it’s one of the defining features of video games. If you have experience with video game development, you know exactly what I mean. Think about the real world and the way things actually work. Molecules, fluid dynamics, physics – it’s just way too much stuff to simulate. Even if we get rid of all the stupid sciencey stuff and just consider things like gravity, friction, momentum, and basically anything else from classical physics, the real world is far too complicated to quickly and reliably reproduce on a phone (or a supercomputer, for that matter).

And the best part about this is that it’s not a problem. In fact, even if phones could handle all of physics, we would probably continue faking it. After all, if we do a good enough job of faking it, why would we bother actually implementing it?

This brings me to the actual point – when developing a game, we’re not trying to create a world for you to look at and interact with. Instead, we’re trying to create something that looks enough like a world that you can interact with that you think we actually did create a whole world. This is a very fine line to ride – too far toward realism, and your game will lag, but too far toward fake-ism, and people will be able to tell and won’t like it.

lava

How did you even get there? …How are you not dying?!

Think about a character walking on relatively flat ground. You could spend all of your time designing a system which allows you to near-perfectly imitate physics. Every time the character takes a step, you calculate exactly how their foot hits the ground, and how it changes their path. This process has eaten up most of your development time, and is so intensive that your game can’t run at more than 15 frames per second. But hey, those perfect physics are worth it, right?

Well, no. I mean, in this case, the ground is relatively flat, so you could have the character just walk along a straight line. Sure, his feet might hover above the ground or clip through it at times, but it’s close enough. Even if the ground isn’t flat, the point is that a simplification of what actually happens is always ‘good enough’ for your game, and it helps you save where you really need to – both development time and processor time.

 

 

NoShadow

Wait, what’s making that shadow?

Where FAKE Shadows Slumber

Now, when it comes to Where Shadows Slumber, there are two big areas in which we consistently lie.

Physics. This is the case that applies to most, if not all, games, and Where Shadows Slumber is no exception. Everything you see when you’re playing is a carefully constructed illusion. Obe is never standing on the ground – the ground is conveniently and strategically placed so that it looks like he’s standing on the ground. “Physics” covers any physical interactions or representations of objects. This leads to a huge disconnect between what things look like, and how they work. In fact, nothing in the game serves the dual purpose of actually doing something and actually looking like something. In every case, we simply have two game objects – one which interacts with other objects according to the rules of our game (our simplified “physics”, if you will), and the other which is just there to look pretty.

Shadows. Where Shadows Slumber is, obviously, based on shadows. Someone who has played the game would tell you that “shadows change things”. However, this isn’t exactly true – in fact, the shadows in Where Shadows Slumber have literally no effect on the gameplay whatsoever! This is another instance of the decoupling of an object and its visual representation. We show the dark black shadow as it moves across the world, but using that shadow’s location is far too computationally intense to be doing every frame. We could do it, but this is another case where we don’t need to be 100% realistic, as discussed in my blog posts on how our shadows work (part 1 and part 2), we use a much simpler algorithm to determine if something is in shadow. This saves computation time while not sacrificing quality. It’s all about that trade-off!

 

8rgt8bamhlnvicvzsd3pfjof

Otherwise known as “what happens when two hacks collide.”

Potential Pitfalls of Constantly Lying

While I strongly advocate for this type of simplification, there are cases where it can cause some trouble. A great example of this came up when we were doing the finishing touches on the original demo for the game.

We had added ramps that Obe could walk on to some levels, to give them a little more depth. It  was working very smoothly, and made the world feel less game-y. Separately, we also came to a decision to have a drop shadow for Obe. It felt weird that Obe himself didn’t cast any shadows, but it didn’t make sense for him to, or everything behind him would be in shadow. We ended up with a circular shadow underneath him. Even though it didn’t make sense from a literal standpoint (since the light wasn’t directly above him), we found that players simply knew what it was, and it added realism, since they were so familiar with the concept of shadows being ‘underneath’ something.

StairShadow

Something looks just a little off…

This was all fine and good – both of these ideas were strong ideas (in fact, the latter is a great example of a place where simulating a very fake shadow was much better than attempting to use a realistic one). However, it was when we combined these ideas that we ran into trouble. You see, the drop shadow we made assumed a flat floor – we just plopped it down with a little transparency, and it looked great! Until Obe got to the stairs, that is. Once he started up a ramp, half of the drop shadow ended up being invisible (because it was underneath the ramp), and the other half was at the wrong angle. We had come up with a great simplification, but it ended up totally ruining the illusion!

These situations do come up, and pretty often – two great ideas can combine to form one horrible edge case. However, this situation in particular came about due to a bad design process. At some point near the end of the demo’s development period, we realized “Oh shoot, Obe needs a shadow!” We hacked together the drop shadow solution without considering the long-term design implications. The important thing about making this type of simplification is to understand that it is inherently “wrong” on some level, since it doesn’t perfectly respect the way the world works. This is fine, until it comes up against other things, which are themselves “wrong”. In these cases, you must be extra careful to think through your design decisions with respect to everything they’re going to interact with. This is yet another reason why it’s important not to make design decisions or changes toward the end of your project.

 

= = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =

I hope this gave you a bit of an insight into what’s actually happening in Where Shadows Slumber! If your confidence in us is shaken and you have any questions about what else we’re lying about, feel free to contact us! You can always find out more about our game at WhereShadowsSlumber.com, find us on Twitter (@GameRevenant), Facebookitch.io, or Twitch, and feel free to email us directly with any questions or feedback at contact@GameRevenant.com.

Jack Kelly is the head developer and designer for Where Shadows Slumber.

Keeping it All in Your Head

When you study computer science, or first get into toying around with it in your spare time, you find yourself working on a lot of small projects, just to get your feet wet. ‘Hello World’, a program which simply prints the text ‘Hello World!’ is perhaps the most-written program of all time. As your projects get bigger, the code behind them gets more complex, and you, as the software architect, have to keep track of everything that’s going on. This seems like a pretty easy task when working on a guess-the-secret-number game. But what happens when your codebase becomes BIGGER THAN YOUR ENTIRE BRAIN?!

 

Untangling The Web

In case you couldn’t tell from all the words about code, computer science, and programming, this post is gonna be one for the more technical folks out there. However, that doesn’t mean that there aren’t lessons to be learned by anyone else – keeping track of complex systems is a skill that applies to any project management task!

Real-world systems are incredibly complex – even more so than they appear, even after working with them for a while. As you add features, fix bugs, and increase the overall complexity of your code, you suddenly find yourself stuck in a tangled web of your own design. Now, the best way to combat this is simply to write good, clean code and follow good design patterns. However, if you want some advice other than “just do everything exactly right”, then read on!

Where Shadows Slumber isn’t the most complex game, but the implementation behind it is very intricate, and I definitely didn’t do everything exactly right. As the only developer on the project, I have to keep track of everything, which is a lot of stuff. Here are some tips I’ve developed for not going crazy trying to understand a system that you yourself created!

 

Tip 1: Divide and Conquer

The first application of divide and conquer is exactly what it sounds like – take the whole system, divide it up, and give a different part of it to each team member. While this doesn’t really apply to the development of Where Shadows Slumber, it’s still worth mentioning. If you’re in charge of one part of your project, and someone else is in charge of another part, there’s less minutiae for you each to keep track of. You only have to know how the pathfinding (for example) works on a higher level – the intricate details of the exact implementation are left for the ‘pathfinding team’. And if you do need to understand how pathfinding works, there’s someone who knows all about it – and that’s what teammates are for!

The other application of dividing and conquering is what I’ve heard referred to as ‘the Google Maps approach’. When you’re looking at a map of the world, you don’t need to be able to see every single city. But if you’re looking at a map of a state, you probably do want to see them all. So, the amount of detail you get depends on the context in which you’re examining it.

How can we apply this to project management? It’s really just a state of mind. When you’re thinking about your whole project, try to think of it from a more abstract point of view, rather than considering all of the details of the implementation. This kind of thinking happens naturally, but we want to actively embrace it. You want to think of the smaller parts of your project as a ‘black boxes’ – you give them some input, and they give you some output. You don’t know (or care) how it gets figured out, until you need to look at that code – at which point, you shouldn’t be thinking about the rest of the project. By only caring about the part of the project you’re currently working on, you free up a lot of space in your head.

 

Tip 2: Keep It Simple

The best way to prevent your project from becoming too complex is, obviously, to keep it simple!

Honor Societies

This comic is a comic [Image credit: xkcd.com]

“But how do you keep your project simple?” I can hear you asking. The key is in how you think about your code. For the most part – and there are notable exceptions – you should be able to think about or describe the function of different parts of your code with ease. Doing so might require the context within which that piece is working, but given that, it should be relatively simple.

Now, don’t get me wrong – your code itself will probably be very complex. However, it’s important that any code has a specific purpose. If some piece of code doesn’t have an easy-to-determine purpose, consider why it’s there and what it’s doing. If possible, see if you can move parts of it into the appropriate parts of your project.

Additionally, when describing the purpose of a section of your code, make sure it’s a relatively simple purpose – the best way to do this is to avoid the word “and”. If the purpose of a file is “to perform pathfinding and determine nearby enemies”, it would probably be best to split that into two different files.

By keeping your code as simple as possible, at least from an organizational perspective, you won’t have to strain yourself every time you try to remember what your code is trying to do.

 

Tip 3: Organization

Speaking of organization, keeping your project organized is one of the best ways to keep it under control. This can be tricky and surprisingly time-consuming, which is why people so often shy away from it, but it can also be crucial to your success. The key here is to create sensible patterns, and then follow them.

ORGANIZED

Everything is right where it should be!

The easiest way to apply this is in directory structure. Make a decision toward the start of your project how you’re going to organize everything, and then stick to it. For Where Shadows Slumber, as you can see, we sort most things by world. All of the levels, materials, and textures for World 1 are in the same folder, since they all apply to the same levels.

However, notice that there are some folders which are not organized by world. Scripts and prefabs are examples of things which span across worlds. While a model or texture might be specific to a certain world, the shadowCharacter.cs script, or the pathfinding node prefab are not, so why should they be sorted by world?

Thinking through your project and deciding on a directory structure that makes sense can make it a lot easier to understand what’s happening in your project. And, every so often, you should re-examine your organization, make sure it still makes sense, and make sure you’re actually following it. In this way, you can be organized, stay organized, and know that your organization is actually effective.

There are also organizational paradigms that you can apply to your code to keep it clean. One of my favorites is the idea of data ownership. The idea is that every piece of data in your project (the location of the character, a bullet’s speed, the number of points a player has, etc.) should have an owner. It’s usually pretty easy to figure out who the owner should be, but sometimes it can be tricky – and it’s in those cases where it’s important to know. If my shadowCharacter.cs script is the owner of the character’s position and velocity, then no other code should be allowed to mess with those values. That way, if there’s a problem with the character’s position, you know exactly where to look.

This is just one example of an organizational coding pattern, but the concept behind them all is pretty similar – at every point, you want to make it easier to understand what your code is doing. It’s a whole lot easier to make changes, fix bugs, and implement new features when the things that your code is doing actually make sense to you.

 

Tip 4: Not Too Complex… Yet

Every project starts out small and simple, and yet we constantly find that our projects have gotten out of hand, growing into sentient monsters, taking over our lives and ruining any chance we had of success – who knew that project management was so much like parenthood?

Incredibly clever comments aside, if our project starts out simple and ends up complex, there has to be some point when it started to go awry, right? And if so, there’s probably a point when we could have noticed it going wrong and steered it back on track. Thus, it’s important, as you’re working, to be constantly considering the state that your codebase is in. Every so often, ask yourself; is this code still clean? The more often you ask yourself this question, the sooner you’ll know when you start heading in the wrong direction – it’s much easier to fix this problem if it’s only just started to go wrong!

This is the concept of technical debt – every so often, you add in some bad code, just so that you can meet a deadline, or get some functionality off the ground. Every time you do that, you’re increasing your technical debt – and if you don’t pay that debt, it adds up until your code is unmanageable. It’s always good to keep your technical debt in the back of your mind, and address it as often as possible.

In my experience, there’s one really good way to determine if your technical debt is getting to the point where it’s impairing your development. If you ever find yourself writing overly complex code, it probably means that you have an overly complex system.

The longer the conditional the better

Genius!

Again, I’m not advocating against complex code in general, as it has its place. But if you find yourself writing complex code to do something simple, or repeatedly thinking “it should be easier to do this”, that’s a big red flag. If you ever fix a bug by ‘trying something’, but you don’t know exactly why it fixed the problem, that’s a sign that you don’t fully understand your code, and code that has exceeded your grasp is exactly what leads to very subtle (read: hard to fix) bugs.

When you get to this point, you should take a step back (and maybe a break), get a fresh look at your code, and spend some time cleaning it up. No one likes spending time on housekeeping tasks, but trust me, it’s a lot more fun spending an hour here and there cleaning up your code than it is mucking through an overgrown garden of technical debt.

 

…And Beyond

This is by no means a definitive list, nor will every part of it apply to you or your project. Rather, these are just some of the philosophies I try to keep in mind as I’m coding and software architect-ing. There are plenty of others, but hopefully adding these to your repertoire will help you reign in your projects and keep them from becoming too complex!

 

= = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = =

If you have questions about managing complex systems, or want to share your own tips, feel free to contact us! You can always find out more about our game at WhereShadowsSlumber.com, find us on Twitter (@GameRevenant), Facebookitch.io, or Twitch, and feel free to email us directly with any questions or feedback at contact@GameRevenant.com.

Jack Kelly is the head developer and designer for Where Shadows Slumber.