Drawing Debug Volumes — Handmade Hero — Episode Guide

0:02Recap and set the stage for the day

3:06Consider approaching alpha fade and upgrading our understanding of the camera

5:48Run the game to assess our current situation

8:44Determine to fix the camera code

9:41handmade_world_mode.cpp and handmade_render_group.cpp: Make PlayWorld() generate fewer rooms and SetCameraTransform() disable the fog in debug mode

12:39Run the game to see that the fog is disabled, and consider how to define the extents of a room

16:10Determine to extend the debug rectangles into the third dimension as per Miblo's request

17:39Blackboard: Drawing rectangles on sheets

18:29handmade_render_group.cpp: Introduce PushVolumeOutline() based on PushCube()

22:06Blackboard: Cube edges, faces and vertices

23:09handmade_render_group.cpp: Enable PushVolumeOutline() to set the cube vertices

25:16Blackboard: Defining cube vertices and edges

28:14handmade_render_group.cpp: Enable PushVolumeOutline() to call PushLineSegment() for each edge

31:18Blackboard: Cube edges

36:03handmade_render_group.cpp: Introduce PushLineSegment()

41:36Blackboard: Computing thickness for our line segment, an end user-facing approach being to use a distance field in a shader, but using the cross product in a debug scenario

45:03Blackboard: Differential Geometry (gradient, divergence and curl) and Geometric Algebra (dot and cross product)

51:00Blackboard: Inner / dot product

53:04Blackboard: Cross product

1:00:07Blackboard: Solving for the cross product vectors that go in the same direction

1:06:15Hunt for a computer algebra system to solve our equation

1:13:31Consult the documentation for YACAS¹

1:15:20Try to plug our equations into YACAS

1:17:00Blackboard: Computing the angle between two vectors

1:19:41Try to prescribe the angle between the vectors

1:20:35View the supposed solution to our equations

1:22:22Continue the hunt for a computer algebra system

1:25:12Consult the documentation for Maxima

1:27:36Try to plug our equations into Maxima

1:29:18View the non-solution

1:30:09Prescribe that the sine must be 1 and view this solution

1:30:36Reflect on our foray into computer algebra systems

1:32:50Cross product²

1:35:43handmade_math.h: Introduce Cross()

1:39:17Blackboard: How the cross product helps us to produce thickness

1:41:31handmade_render_group.cpp: Enable PushLineSegment() to compute the normalised perpendicular using Cross() and NOZ()

1:44:58handmade_render_group.cpp: Enable PushVolumeOutline() to pass the C and Thickness to PushLineSegment()

1:46:55handmade_world_mode.cpp: Enable PlayWorld() to call PushVolumeOutline()

1:47:55Run the game to see our 3D debug cube

1:48:47handmade_render_group.cpp: Sanity check PushLineSegment()

1:49:39Blackboard: Pre-projecting down to the screen

1:50:57handmade_render_group.cpp: Temporarily make PushLineSegment() keep the Line in line with the camera's Z axis

1:51:39Run the game to see that debug cube

1:52:06handmade_math.h: Sanity check Cross() and consider what is happening

1:53:49handmade_render_group.cpp: Temporarily make PushVolumeOutline() only draw one segment, and run the game to see how that looks

1:56:33handmade_world_mode.cpp: Reduce the thickness and run the game to see the problem when the line is head-on

1:57:34handmade_render_group.cpp: Enable PushLineSegment() to compensate for the LinePerpLength

1:59:31Blackboard: On losing our length as the angle between the edges reduces to 0

2:00:21handmade_render_group.cpp: Continue to enable PushLineSegment() to compensate for the LinePerpLength

2:02:24Run the game to see our current problematic behaviour

2:03:43Blackboard: Computing the perpendicular to the CameraZ

2:05:09Q&A

🗩

2:05:09Q&A

🗩

2:05:09Q&A

🗩

2:06:24insofaras hmd_bot handles the Q&A message and seems to be dead

🗪

2:06:24insofaras hmd_bot handles the Q&A message and seems to be dead

🗪

2:06:24insofaras hmd_bot handles the Q&A message and seems to be dead

🗪

2:06:54desuused Here's your equations solved in Maxima [image], and here's your equations solved with expanding (sinθ)^2 through dot product: [image]. I think this approach is completely wrong when trying to derive cross product

🗪

2:06:54desuused Here's your equations solved in Maxima [image], and here's your equations solved with expanding (sinθ)^2 through dot product: [image]. I think this approach is completely wrong when trying to derive cross product

🗪

2:06:54desuused Here's your equations solved in Maxima [image], and here's your equations solved with expanding (sinθ)^2 through dot product: [image]. I think this approach is completely wrong when trying to derive cross product

🗪

2:07:50ingenero What would be the downside of drawing your debug lines as 3D cylinders or extruded rectangles? It seems like that would look a little nice when viewed from all angles

🗪

2:07:50ingenero What would be the downside of drawing your debug lines as 3D cylinders or extruded rectangles? It seems like that would look a little nice when viewed from all angles

🗪

2:07:50ingenero What would be the downside of drawing your debug lines as 3D cylinders or extruded rectangles? It seems like that would look a little nice when viewed from all angles

🗪

2:10:18vaualbus Also shall we draw the cubes around the room?

🗪

2:10:18vaualbus Also shall we draw the cubes around the room?

🗪

2:10:18vaualbus Also shall we draw the cubes around the room?

🗪

2:10:47handmade_entity.cpp: Make UpdateAndRenderEntities() draw collision volumes

2:12:06Run the game to see what that does

2:12:37handmade_render_group.cpp: Make PushVolumeOutline() take the ObjectTransform into account

2:13:32Run the game to see the collision volumes

2:15:10handmade_render_group.cpp: Enable SetCameraTransform() to set the DebugXForm in both modes, and run the game to see the collision volumes in the game mode

2:16:44desuused You mentioned that you use discriminated unions. Have you seen Rust's implementation of sum types?

🗪

2:16:44desuused You mentioned that you use discriminated unions. Have you seen Rust's implementation of sum types?

🗪

2:16:44desuused You mentioned that you use discriminated unions. Have you seen Rust's implementation of sum types?

🗪

2:16:59jezzi23 I was looking at how the vector types are set up in Handmade Hero as unions. As far as I know, it's undefined behaviour to write value to one type of a union and then read as another type. Is doing something like this well defined - vec1.x = 5.0f and then read as vec1.E[0] - or is it one of the cases where we rely that a reasonable compiler will do the right thing?

🗪

2:16:59jezzi23 I was looking at how the vector types are set up in Handmade Hero as unions. As far as I know, it's undefined behaviour to write value to one type of a union and then read as another type. Is doing something like this well defined - vec1.x = 5.0f and then read as vec1.E[0] - or is it one of the cases where we rely that a reasonable compiler will do the right thing?

🗪

2:16:59jezzi23 I was looking at how the vector types are set up in Handmade Hero as unions. As far as I know, it's undefined behaviour to write value to one type of a union and then read as another type. Is doing something like this well defined - vec1.x = 5.0f and then read as vec1.E[0] - or is it one of the cases where we rely that a reasonable compiler will do the right thing?

🗪

2:20:39desuused Would you describe discriminated union to the general public?

🗪

2:20:39desuused Would you describe discriminated union to the general public?

🗪

2:20:39desuused Would you describe discriminated union to the general public?

🗪

2:20:50Blackboard: Struct, as a representations of data in memory

2:24:12Blackboard: Union, as a combination of structs

2:26:21Blackboard: Discriminated union, as a union containing a type field

2:30:03Blackboard: Inheritance in C++

2:32:32graham1642 Can you give a motivating example for these discriminating unions?

🗪

2:32:32graham1642 Can you give a motivating example for these discriminating unions?

🗪

2:32:32graham1642 Can you give a motivating example for these discriminating unions?

🗪

2:33:11Blackboard: A "single dispatch" example in which a discriminated union and inheritance work equally well

2:38:54Blackboard: The tractability of Single vs Double Dispatch

2:40:00Blackboard: A "double dispatch" example in which a discriminated union excels

2:46:06Blackboard: Generating a sparse matrix containing the cases you care about

2:49:34Blackboard: Discriminated unions shine in enabling the ability to write code which looks at the way in which types are set among more than one object

2:53:44cubercaleb How would you feel if you wrote the functions for colliding each pair or types, and allowed the compiler to generate the sparse matrix of function calls for you?

🗪

2:53:44cubercaleb How would you feel if you wrote the functions for colliding each pair or types, and allowed the compiler to generate the sparse matrix of function calls for you?

🗪

2:53:44cubercaleb How would you feel if you wrote the functions for colliding each pair or types, and allowed the compiler to generate the sparse matrix of function calls for you?

🗪

2:55:09Recommend researching category theory, with a call for Pseudonym73

3:00:12cynokron Are you hoping JAI will have discriminated unions?

🗪

3:00:12cynokron Are you hoping JAI will have discriminated unions?

🗪

3:00:12cynokron Are you hoping JAI will have discriminated unions?

🗪

3:00:32Wind down here

🗩

3:00:32Wind down here

🗩

3:00:32Wind down here

🗩

Keyboard Navigation

Global Keys

Menu toggling

In-Menu and Index Controls

Quotes and References Menus and Index

Quotes, References and Credits Menus

Filter Menu

Filter and Link Menus

Credits Menu