H behaves strangely. Typing 250 results in 249. Pushing the slider to 360 results in 300. Typing 500 gives 319.
Setting S to 0 or nothing resets H to 0, which makes it hard type in the fields directly. Same for V, which resets S and H
circle should adjust colors to current V value or have an extra colored selector for V like this(src)
Alpha is missing (?)
Support for Hex-codes
Thanks. I made a card for these and will get to it soon.
126. Bug: Calling World_SendMessage[ToAll]() in an Automator (before the round is started) crashes the game. I've fixed the crash, however the Automators cannot send messages. What are you trying to use this for?
127. Feature Request: CollisionFilterTriggerPreset that triggers on everything but not Self. I want to add a triggerbox to a moving physics object (the box overlaps the collision model), but it is triggered by its own collision component on spawn. I need a filter presets that ensures that message.triggerId is not the same as message.otherId. Currently its only solvable by setting the collision type to hitbox, which I do not want. This sounds like maybe this should just be the behavior regardless of the preset. I'll chat with the rest of the team about it.
128. Bug: Grid increase/decrease does not work with every keyboard layout (i.e. german) What exactly is the problem? Is it the default keys don't work, but remapping them does, or does it not work even remapped?
129. Q: I am not supposed to add 2 directional lights, right? Only one of them casts a shadow, and which one depends on the current viewing angles. This is limitation of the current implementation. I'll make it so that more than 1 directional light is supported in the (near) future.
130. Remark: Model editor does not open for standard models. Not sure if it should. Works fine for self made ones. This is the intended behavior since you don't have the "source" files for the other models (i.e. the fbx files). In the future I'm going to make it so they open in a "read only" mode like a model viewer.
126. Bug: Calling World_SendMessage[ToAll]() in an Automator (before the round is started) crashes the game. I've fixed the crash, however the Automators cannot send messages. What are you trying to use this for?
As mentioned here, I wanted to have an automator and a script which do roughly the same thing. So to reduce code duplication my firstsecond impulse was to let the automator call the script. This was when I noticed the crash; have duplicated the code now.
Edit: my first impulse was to add the OnSetText() event handler to the automator.
128. Bug: Grid increase/decrease does not work with every keyboard layout (i.e. german) What exactly is the problem? Is it the default keys don't work, but remapping them does, or does it not work even remapped?
Ha, did not even see they can be remapped. Remapping works, but with the default mapping (LeftBracket/RightBracket) on a german keyboard layout, no key or key combination triggers the increase or decrease of the grid size. In particular not even pressing [ or ], which on a german keyboard are mapped to AltGr+8 and 9. Switching to US layout (where [ and ] are to the right of the P) in windows works.
Maybe related: with a german keyboard layout enabled, everything under Options->Bindings still registers the US keys. I.e. hitting + or #, which have their own keys, registers as = and 3. Keys that don't have an equivalent like ü,ö,ä,ß don't register at all. NS2 had this right.
129. Q: I am not supposed to add 2 directional lights, right? Only one of them casts a shadow, and which one depends on the current viewing angles. This is limitation of the current implementation. I'll make it so that more than 1 directional light is supported in the (near) future.
Don't actually need it and it's probably rare that it makes sense (i.e. 2 suns), just noticed the slight glitches.
As mentioned here, I wanted to have an automator and a script which do roughly the same thing. So to reduce code duplication my firstsecond impulse was to let the automator call the script. This was when I noticed the crash; have duplicated the code now.
Edit: my first impulse was to add the OnSetText() event handler to the automator.
Great, so the current build so address this problem without the need to duplicate the code.
Ha, did not even see they can be remapped. Remapping works, but with the default mapping (LeftBracket/RightBracket) on a german keyboard layout, no key or key combination triggers the increase or decrease of the grid size. In particular not even pressing [ or ], which on a german keyboard are mapped to AltGr+8 and 9. Switching to US layout (where [ and ] are to the right of the P) in windows works.
Maybe related: with a german keyboard layout enabled, everything under Options->Bindings still registers the US keys. I.e. hitting + or #, which have their own keys, registers as = and 3. Keys that don't have an equivalent like ü,ö,ä,ß don't register at all. NS2 had this right.
It looks like some changes were made in NS2 after I forked the codebase for Future Perfect. I've integrated these now and made a new build in the "beta" branch. When you get a chance, could you check it out and see if it's working the way you'd expect?
117. Feature Request: A method that returns the position of the center of mass
We could add that, although it would have to do some computations for something like rag doll where there might be multiple bodies. What are you going to use it for?
As mentioned here, I wanted to have an automator and a script which do roughly the same thing. So to reduce code duplication my firstsecond impulse was to let the automator call the script. This was when I noticed the crash; have duplicated the code now.
Edit: my first impulse was to add the OnSetText() event handler to the automator.
Great, so the current build so address this problem without the need to duplicate the code.
Could you elaborate on this? What exactly did you change? Can the automator handle OnSth() events now? Or is it now able to send messages? Only after the game started?
Ha, did not even see they can be remapped. Remapping works, but with the default mapping (LeftBracket/RightBracket) on a german keyboard layout, no key or key combination triggers the increase or decrease of the grid size. In particular not even pressing [ or ], which on a german keyboard are mapped to AltGr+8 and 9. Switching to US layout (where [ and ] are to the right of the P) in windows works.
Maybe related: with a german keyboard layout enabled, everything under Options->Bindings still registers the US keys. I.e. hitting + or #, which have their own keys, registers as = and 3. Keys that don't have an equivalent like ü,ö,ä,ß don't register at all. NS2 had this right.
It looks like some changes were made in NS2 after I forked the codebase for Future Perfect. I've integrated these now and made a new build in the "beta" branch. When you get a chance, could you check it out and see if it's working the way you'd expect?
Reassigning key works now, similar to NS2 (that is keys like "ü" show up as "#41" but work as expected). Still "LeftBracket" and "RightBracket" cannot be triggered with the german layout.
117. Feature Request: A method that returns the position of the center of mass
We could add that, although it would have to do some computations for something like rag doll where there might be multiple bodies. What are you going to use it for?
I have only one solid object and do not care about ragdolls. I need to know the location of the center of mass, so I know where to add pairs of local forces that have point symmetry. Here, I made a video. Note that the 4 hoverpads are not symmetric around the center of mass, which causes extensive swinging.
Could you elaborate on this? What exactly did you change? Can the automator handle OnSth() events now? Or is it now able to send messages? Only after the game started?
I changed it so that calling the function won't crash the game, however the message won't go anywhere. Sending messages between automators is something that we could add, but I haven't seen a use for it yet.
I have only one solid object and do not care about ragdolls. I need to know the location of the center of mass, so I know where to add pairs of local forces that have point symmetry. Here, I made a video. Note that the 4 hoverpads are not symmetric around the center of mass, which causes extensive swinging.
Ok, that looks cool. I added the function (Collision_GetLocalCenterOfMass). It's available in the beta branch right now.
I have only one solid object and do not care about ragdolls. I need to know the location of the center of mass, so I know where to add pairs of local forces that have point symmetry. Here, I made a video. Note that the 4 hoverpads are not symmetric around the center of mass, which causes extensive swinging.
OK so now I want to see a Wipeout mod in FP.... :P
I added the function (Collision_GetLocalCenterOfMass). It's available in the beta branch right now.
Thanks. Works perfectly now.
132. Remark: Nvidia Shadowplay screen recording autodetects the game title as "Tian Long Ba Bu". No idea why or how, but maybe that String looks familiar to someone.
133. Bug: Checkboxes and String fields for automators don't work anymore (are constantly overwritten with previous value)
the builder (drag and dropping into the game window) doesn't like the file extensions for FBX files in capitals. since this varies in versions of 3ds max i assume it will also vary between other modeling programs too. Was a scratching my head over this one for a good ten minutes.
Might also be worth noting that currently .fbx files on my system do not currently have a default program.
the builder (drag and dropping into the game window) doesn't like the file extensions for FBX files in capitals. since this varies in versions of 3ds max i assume it will also vary between other modeling programs too. Was a scratching my head over this one for a good ten minutes.
Might also be worth noting that currently .fbx files on my system do not currently have a default program.
This was change we made because the different captializations caused of lot of additional complexity since the engine itself is case sensitive to make other platforms work consistently. We might go back to supporting this in the future for the sake of user friendliness, but at the moment it was causing too many headaches.
Regarding 132: Tian Long Ba Bu (天龙八部) is a game listed in the Games Supported by GeForce Experience list. So their detection algorithm just gets it wrong.
Regarding 132: Tian Long Ba Bu (天龙八部) is a game listed in the Games Supported by GeForce Experience list. So their detection algorithm just gets it wrong.
135. Bug:: Automators are broken.
start editor
start game (F5)
stop game (shift+F5)
change size of floor entity -> no effect
Edit: fixed by yesterday's update
This was actually still broken (the problem was intermittent), but we just released another update which does indeed fix it.
Soul_RiderMod BeanJoin Date: 2004-06-19Member: 29388Members, Constellation, Squad Five Blue
Primarily filesize for the game and steam workshop 100mb limits. I can't make audio heavy games with those restrictions. If not midi, how about maybe adding tracker support? That seems like a good alternative to midi for the games I would like to create, and is something quite well used in the game industry.
I have some ideas for 3d music based games I'd like to try.
Perhaps we can support at least some kind of compressed audio format. I wonder what would be best. Is ogg vorbis decent?
I will add a card in Trello but it will have to be a lower priority task for now as we have some bigger things we need to get to in the short term. Thanks!
137. Bug: World_SendMessage/World_SendMessageToAll nowhere mentioned in api.html or the rest of the docs
138. Bug:: I noticed that a function that is bound the to mousewheel (Action.Next) is always triggered twice. Here is some example code
upcount = 1
function countup()
upcount = upcount + 1
print(upcount)
end
function OnInput(input)
if bit.band(input.actions, Action.Next) ~= 0 then
countup()
end
end
Slowly turning the mousewheel will produce outputs like 2 2 then 3 3, then 4 4
Right now I have a hard time to make sure that a function is only called once per button press.
Until here, that seems like a minor annoyance, but I actually believe this to be a major bug, and here is why:
When you turn your mousewheel rapidly, the output changes
When I dump _G, I actually get two dumps that differ in _G.upcount
Tests indicate that for players connected over the network, the function is only called once.
Making the functions and variables local does not change anything.
139. Feature Request: "if bit.band(input.actions, Action.Next) ~= 0 then ..." is nice and all, but maybe there should be a high level function like IsPressed(Action.Next). While we are at it: when you want to query the state of a button outside of the OnInput function, you need to introduce a global variable and sync it with the input bitmask, why not have the current input state available globally at all times?
140. Remark: Consider the following code
function OnTest(message)
timedfunction(message.var, 0)
end
function timedfunction(var, num)
print(var)
if num < 5 then
timedcallback(1/30, function() timedfunction(var, num+1) end)
end
end
World_SendMessageToAll("Test",{var=Vec3(1,1,1)})
If it's not immediatly clear what's happening here, I am accessing freed chunks of memory here. I am not too familiar with the internals of Lua, but I suppose that the CDATA block message.var is passed as reference and then deallocated after the OnTest function is left. timedcallback then fails to make it's own copy of the object. I can circumvent this by changing to call to timedfunction(message.var*1, 0) [or in my case timedfunction(message.targetCoords*Coords.GetIdentity(), 0)]
141. Bug: Movement code: running diagonally agains small walls (i.e. wall_01_small or 2 crates) and spamming the jump button results in high jump altitudes
142. Q: Is there a way to have a child entity that does not rotate with it's parent, but still moves along with it? Other by manually rotating the child in the opposite direction of the parent.
143. Q: Is there a way to have an object that is not visible to everyone? Think of cloaked skulks of waypoint breadcrumbs.
Additional Remarks to 123.:
Recently I have worked alot with moving child entities, and came to the conclusion that the local version like Self_GetLocalCoords (and its pendants for angles, positions and setters and getters) are the ones you almost always want. Considering how Bone_GetCoords already returns the local coordinates, I propose that the generic function like *_GetCoords, *_SetAngles, ... work in the local reference frame and specialized ones are added to access the global/absolute coordinates.
Apologies in advance, this is probably not going to be the most satisfying set of replies!
136. Bug:: single character variables like
Thanks, I fixed it.
137. Bug: World_SendMessage/World_SendMessageToAll nowhere mentioned in api.html or the rest of the docs
The API documentation is currently only including functions that are implemented in C++. These particular functions are implemented in Lua. We'll include them in the future though.
138. Bug:: I noticed that a function that is bound the to mousewheel (Action.Next) is always triggered twice. Here is some example code
This sounds like it's behaving properly and what you are seeing is the result of running the code on both the server and client. If you change it to print(World_GetName(), upcount) you should be able to see the two different sources. The same code may run on the client multiple times as well due to client side prediction, but before it runs each time it should reset upcount and thus produce the same result.
139. Feature Request: "if bit.band(input.actions, Action.Next) ~= 0 then ..." is nice and all, but maybe there should be a high level function like IsPressed(Action.Next). While we are at it: when you want to query the state of a button outside of the OnInput function, you need to introduce a global variable and sync it with the input bitmask, why not have the current input state available globally at all times?
This has to do with the networking. It would probably make sense to introduce some more concise way of determining of a button is pressed, but adding a global function wouldn't make sense in all contexts. For example, say you have a script on an object which is not a player; what should the functions return in that case? I will give some consideration how this could be done in a way that doesn't lead to confusing behavior.
140. Remark: Consider the following code
Yes you are right, this is a good catch. Another way to do it is like Coords(message.targetCoords) to make a copy of the variable. Fixing problems like this in luajit is one of the hard tasks we still have to solve. What I'd really to to do is make it so that a = b has copy semantics if b is a "structure" (i.e. a Vec3, Coords, etc.).
141. Bug: Movement code: running diagonally agains small walls (i.e. wall_01_small or 2 crates) and spamming the jump button results in high jump altitudes
This is caused by the logic in the movement code that gives you a little extra boost when you're running up a hill in jumping. Without the boost the jump feels very ineffective. The consequence is that jumping on a steep or nearly vertical slope gives you a big boost.
142. Q: Is there a way to have a child entity that does not rotate with it's parent, but still moves along with it? Other by manually rotating the child in the opposite direction of the parent.
Not at the moment, but one feature I've had in mind is make a child entity that doesn't inherit its parent's transform. This is something we could use to place the "ghost" cube for the Minecraft-like building without having the networking delay. This could be an extension to that (i.e. only inherit position, only inherit rotation, inherit full transform). What's your use case?
143. Q: Is there a way to have an object that is not visible to everyone? Think of cloaked skulks of waypoint breadcrumbs.
One way of doing this I've been considering is allowing the camera to include tags that affect what it renders. What do you think about that?
Additional Remarks to 123.:
Recently I have worked alot with moving child entities, and came to the conclusion that the local version like Self_GetLocalCoords (and its pendants for angles, positions and setters and getters) are the ones you almost always want. Considering how Bone_GetCoords already returns the local coordinates, I propose that the generic function like *_GetCoords, *_SetAngles, ... work in the local reference frame and specialized ones are added to access the global/absolute coordinates.
I came to the opposite conclusion (at least that if you didn't have an explicit reason to get the local coordinate frame, you wanted the global one) when I named them, but I'll take your feedback into consideration.
138. Bug:: I noticed that a function that is bound the to mousewheel (Action.Next) is always triggered twice. Here is some example code
This sounds like it's behaving properly and what you are seeing is the result of running the code on both the server and client. If you change it to print(World_GetName(), upcount) you should be able to see the two different sources. The same code may run on the client multiple times as well due to client side prediction, but before it runs each time it should reset upcount and thus produce the same result.
Ok that makes sense. I guess the proper questions is then, why the client side prediction and the server are so easily (and permanently) desynchronized even when client and server are on the same machine, and what I can do to syncronize them again.
Predict 2
Server 2
Predict 3
Server 3
Predict 4
Server 4
Predict 5
Predict 6
Server 5
Predict 7
Server 6
Predict 8
Server 7
Predict 9
Server 8
Predict 10
Server 9
Predict 11
Server 10
Predict 12
Server 11
Predict 13
Predict 14
Server 12
Predict 15
Predict 16
Server 13
Predict 17
Server 14
Predict 18
Server 15
Predict 19
Server 16
Predict 20
Predict 21
Server 17
Predict 22
Server 18
Predict 23
Server 19
Predict 24
Server 20
Predict 25
Server 21
Predict 26
Server 22
Predict 27
Server 23
Predict 28
Server 24
Predict 29
Server 25
Predict 30
Server 26
Predict 31
Predict 32
Server 27
Predict 33
Server 28
Predict 34
Server 29
Predict 35
Server 30
Predict 36
Server 31
139. Feature Request: "if bit.band(input.actions, Action.Next) ~= 0 then ..." is nice and all, but maybe there should be a high level function like IsPressed(Action.Next). While we are at it: when you want to query the state of a button outside of the OnInput function, you need to introduce a global variable and sync it with the input bitmask, why not have the current input state available globally at all times?
This has to do with the networking. It would probably make sense to introduce some more concise way of determining of a button is pressed, but adding a global function wouldn't make sense in all contexts. For example, say you have a script on an object which is not a player; what should the functions return in that case? I will give some consideration how this could be done in a way that doesn't lead to confusing behavior.
I did in no way propose to replace the bitmask with something else. I know why it is there and why it should stay there. All I said is, that having to use logical operations on bitmasks to get a button state is unnecessary complicated, is not very readable and hence makes the game a little bit less accessible. That's why there should be some higher level mechanism that hides the underlying bitmask.
There are multiple ways to achieve that.
The most simple would be to have a function like
function IsKeyPressed(inputs, action)
return bit.band(input.actions, action) ~= 0
end
function OnInput(input)
if IsKeyPressed(input, Action.Next) then
...
But having to pass the input everytime, when it's the only thing you could legitimately pass to that function seems redundant, which is why I proposed to have the current keypresses in a global variable, and then have a global IsPressed(action) function. If no player is "attached" to an entity, the question if a certain key is pressed would of course always yield 'false'. If a player is attached, the global actions would be updated accordingly. However, this implies that only one player can control an entity at a time (which I assume but haven't checked yet).
If this is too limiting, then the current button states could be a property of the player (which is already carrying some data fields), and the call could be made more explicit:
if IsKeyPressed(playerId, Action.Next) then ... -- returns false if playerId<0
Another approach would be to have a bunch of specialized functions
function OnJumpPressed -- executed once
...
end
function OnJumpHeld -- executed continously
...
end
function OnJumpReleased
...
end
function OnDropPressed
...
end
Yes you are right, this is a good catch. Another way to do it is like Coords(message.targetCoords) to make a copy of the variable. Fixing problems like this in luajit is one of the hard tasks we still have to solve. What I'd really to to do is make it so that a = b has copy semantics if b is a "structure" (i.e. a Vec3, Coords, etc.).
I was not aware there is a copy-constructor. Now I dont have to write a copy function to eliminate that unnecessary multiplications. Thanks.
141. Bug: Movement code: running diagonally agains small walls (i.e. wall_01_small or 2 crates) and spamming the jump button results in high jump altitudes
This is caused by the logic in the movement code that gives you a little extra boost when you're running up a hill in jumping. Without the boost the jump feels very ineffective. The consequence is that jumping on a steep or nearly vertical slope gives you a big boost.
I dare to object.
To be honest, I hadn't checked the movement code before (because I am no longer using it). If I had, I would have seen the comment in the part you have mentioned, that the jumpBoost has a potential to be exploited, and wouldn't have made a bug report.
But while it probably has it's own problems, the jumpBoost is not the problem here!
The problem lies few lines above:
if jumpDesired or (now - timeLastJumpDesired) <= queuedJumpTime then
The OnInput function is executed continously (btw, Why?), and the second part (and isOnFloor) is true on multiple of those calls, which causes the queued jump to be added multiple times during 'queuedJumpTime'. Easy fix would be 'timeLastJumpDesired = -1' in the 'if isOnFloor' part (around line 210)
142. Q: Is there a way to have a child entity that does not rotate with it's parent, but still moves along with it? Other by manually rotating the child in the opposite direction of the parent.
Not at the moment, but one feature I've had in mind is make a child entity that doesn't inherit its parent's transform. This is something we could use to place the "ghost" cube for the Minecraft-like building without having the networking delay. This could be an extension to that (i.e. only inherit position, only inherit rotation, inherit full transform). What's your use case?
Well, a picture is worth a thousand words. Here are 60 pictures a second:
So I have my glider (body) and a camera (head). As the body is physics enabled, and you cannot have physics enabled child entities, the camera naturally is a child entity of the glider. Now when the body moves, the camera needs to follow. When the camera rotates, the body needs to follow. But as the camera is a child of the body, any rotation of the body also rotates the camera. As a result, the controls feel almost game-pad like. I tried to keep the camera steady by determining the current rotation of the body and that apply a counter rotation on the camera, but the results are mediocre (0:49-1:12). Ultimatly I would want the horizon line to stay fixed, which means the camera moves with the center of mass, but does not get any of the rotations.
The hiccups in the first 20 seconds during the zooming are caused by issue 138 btw.
Additional Remarks to 123.:
Recently I have worked alot with moving child entities, and came to the conclusion that the local version like Self_GetLocalCoords (and its pendants for angles, positions and setters and getters) are the ones you almost always want. Considering how Bone_GetCoords already returns the local coordinates, I propose that the generic function like *_GetCoords, *_SetAngles, ... work in the local reference frame and specialized ones are added to access the global/absolute coordinates.
I came to the opposite conclusion (at least that if you didn't have an explicit reason to get the local coordinate frame, you wanted the global one) when I named them, but I'll take your feedback into consideration.
Above, you have mentioned the minecraft block placing procedure, which made me thinking. Whether you require global or local coordinates most of the time may depend strongly on your use case, and no matter how you name the functions, there is always someone who will choose the wrong one on the first try.
144. Feature Request: Entity_GetChildByName(entityId, name)
Sure, I can use tags, but it feels dirty.
145. Bug: using_player.lua:
local highlightColor = Color(255, 0, 0, 255) --- Color
I am aware that Hud_HighlightEntity() is currently broken, so this doesn't matter, but this should be either Color(1.00, 0, 0, 1.00) or Color.red
146. Bug: for a physics enabled collision that is part of a player this code
function OnThink()
print(World_GetName(), Collision_GetLinearVelocity(collisionId):GetLength())
end
147. Remark: When i make something a child of the player by using Self_SetParent(), that triggers OnCreate() in the Predict context, which was unexpected. Similarly when i detach the parent again, the OnDestroy method is called in the Predict context.
148. Remark: I needed a vector in the local coordinate system. The current way to do it is
local localVec = coords:GetInverse():TransformDirection(v)
which is fine. But since I only use cartesian coordinate systems, the axes are guaranteed to have length 1. So the above has some unneccessary multiplications and devisions, when effectivly I only need a simple matrix multiplication (matrix times vector):
local localVec = Vec3(v:DotProduct(coords.xAxis), v:DotProduct(coords.yAxis), v:DotProduct(coords.zAxis))
To my surprise, there was none.
149. Remark: (possibly bug)
When an object is inside a trigger area, changing the collision from Solid to Hitbox or Ghost to Hitbox triggers the Trigger (object leaves the area).
This is relevant for the game bounds (when setup in a way that everything is inside and destroyed on leave), as this effectivly means that nothing can be changed to Hitbox type (after creation), as this would immediately cause the destruction of the entity.
150. Remark:
have changed some values while the game was live. not sure what exactly happend, but this caused some asserts at first and then the game crashed. when i tried to load my mod, the assertion errors returned (havent kept a copy). turns out the collision mesh for the object i had edited at last was no longer in the asset browser. I cleared the roaming/FuturePerfect directory to force a rebuild of the collsion mesh, but it didn't work. In the end, I had to delete to following lines in the .build_options file of the model, after that it everything was generated.
152. Remark: Observed multiple times now, that when someone tries to join my game while I am on the beta channel and he is not, he will download a lot of stuff (expected) but then time out. I am not too woried if betas cannot join non-betas, but maybe the timeout doesn't take ongoing downloads into account.
147. Is this creating a problem for you or just seems wrong? These objects are being created/destroyed in the prediction world but this should maybe be hidden from scripts.
148. You want a Matrix type in script basically?
149. Hmm, we will have to think about this one. That behavior is not great.
I have fixed 150.
151. I changed that to use global position/angles. I am not sure why I was using local there.
152. I suspect it is because of how old the non-beta build is at this point but maybe something unintended is going on here with the timeout. I'll see what I can uncover.
We are still deciding on the name thing. It may be a can of worms as we may need to add a lot more of these types of name query functions if we add that one. Names were intended as a debugging mechanism more than anything.
147. Is this creating a problem for you or just seems wrong? These objects are being created/destroyed in the prediction world but this should maybe be hidden from scripts.
No, not a problem at all. Had attached some messages to the creation/destruction events and was surprised of the addtional outputs. But all the calls stay in the predict context, so they don't matter. I make it my habit now to always print World_GetName() with everything. Helps to understand such outputs:
print(World_GetName(), EntityId_ToString(id), Entity_GetIsValid(id)) -- id is networked
Output:
Predict EntityId doesn't exist true
Server EntityId(name: Puck, index: 933 use: 1) true
153. Remark: Documentation should say that World_TestRayOverlap(origin, direction, distance, filter) assumes a normalized direction. I used the current velocity as direction and it took me some time to realize that direction * distance is used.
154. Feature Request: Does Self_GetComponent("Camera") need to be case sensitive? Why even strings when all the componentsTypes are already listed, i.e. ComponentType.Camera?
155. Feature Request: Entity_GetIsValid(nil) should not throw an error, so i can shorten these lines:
if target ~= nil and Entity_GetIsValid(target) then
Also, I had the special case where the argument was a valid Id in the server context, but nil in the predict context.
156. Remark: Short time of confusion until I realised that Camera_SetFov takes radians. Maybe documentation should mention that.
157. Remark: I want to know if I hit a "Ramp".
As I cannot check for the name (#102, #144), I gave every ramp a tag and then checked for that
function OnContact(message)
print("contact with", EntityId_ToString(message.otherId), Entity_GetHasTag(message.otherId, "ramp"))
end
Result:
contact with EntityId(name: ramp_01_a, index: 3354 use: 1) false
ramp_01_a is a child of "Ramp", so it does not inherit it's tag. I ended up writing a function that recursivly checks all parents for tags.
Of course I am not suggesting that children should inherit the tags of their parents, but maybe there is a more elegant solution.
158. Bug: The World_EndGame() overlay is on top of the entity editor
159. Bug: The only method to unset an entity field such as the "attach to" field is to drop it against the background. Kind of difficult to do from the inside of a room Edit: Backspace works fine
160. Feature Request: on grid increase/decrease also show the snapsize for angles
161. Bug: add script
select script field
type "creat"
drop "add force on create" into the script field
select script field
type "creat"
drop "create player on destroy" into the script field
-> height of "offset distance" field is wrong
(not always reproducable)
162. Bug: add any script to an entity. select the script field and drop "add score on activate" instead. notice how the default score is 0, although it is 1 in the script. works with any script.
163. Bug: create_player_on_destroy.lua:
not changing offsets, will produce errors:
Entity position set to a non-finite value
Entity angles set to a non-finite value
here is why:
local position = coords.origin - (coords.zAxis * offsetDistance)
local lookDir = coords.origin - position
local lookInCoords = Coords.GetLookIn(position, lookDir, Vec3(0, 1, 0))
for offsetDistance = 0, lookdir is Vec3(0,0,0)
this causes lookInCoords to be (Vec3(0,0,0), Vec3(0,0,0), Vec3(0,0,0), someorigin)
which causes lookInCoords:GetAngles() to be nan
solution:
insert
if lookDir == Vec3(0,0,0) then lookDir = Vec3(0,0,1) end
after line 31, or limit the minimum offsetDistance
164. Bug: package manager lacks destinction between local and workshop versions of the same package
165. Bug: could publish a package that depends on other local packages without a warning
166. Feature Request: right click to look around should work while editing an entity
167. Bug: general_spawner.lua:
randomDir unused
spawnRadius unused
spawnAngles unused (!)
use new Entity_AddForce instead of World_SendMessage("AddForce") (although regarding the answer to 106 I am not sure why Entity_AddForce even exists, as it breaks encapsulation)
if spawnMinForce > 0 and spawnMaxForce > 0 then
should be
if ( spawnMinForce ~= 0 or spawnMaxForce ~= 0 ) and spawnMinForce <= spawnMaxForce then
168. Remark: playerspawners for two teams (in two different spawnGroups) dont handle playerJoin (a player is spawned in each spawnGroup), and therefore need to ignore playerJoin and have a central instance manage all player assignments. Maybe I misunderstand the spawnGroup concept here.
Also on player death a "Dead Player" is spawned, that does not inherit the spawnGroup of the destroyed avatar, hence that information is lost.
169. Q: There is OnPlayerJoin. Is there a similar call once a player leaves? I want to maintain a list of all connected players, so i can decide which team new players should be assigned to. (yes, I could check if playerIds are valid everytime i use them)
170. Q: Action.128 = "Drop" <- for what is this field useful? Isn't Action.Drop = 128 all that is ever needed?
171. Remark: I am slowly running out of assignable actions. There are 19 actions, as far as I understand there is room for 13 (45?) more.
172. Bug: Self_SetParent() vs client prediction
When I catch an object, I run this code on the object:
Collision_SetFilter(collisionId, CollisionFilterPreset.Ghost)
Collision_SetPhysics(collisionId, false)
Self_SetParent(message.catcher) -- message.catcher is the player
This is what happens
Note how I shoot in x-direction (world coordinates) and for ~1 frame the puck appears in x-direction in the local coordinates of the player.
Both, the player and the puck are physics-enabled.
Is there anything I can do to prevent this?
0Mat is the texture name that comes right out of MagicaVoxel and if leading numbers are not allowed I need to edit the fbx file, which I would like to avoid. I also believe the error is new, never noticed it before.
174. Remark: I have an entity with 2 Scripts. One has the OnSetText() function, the other one looks like this:
The first PlaySound() call does not emmit a sound, the second one does. If I remove the SendMessage, both do.
The problem is this function which is called by OnSetText:
local function DeleteChildren()
local children = Self_GetChildren()
for i=1,#children do
World_DestroyEntity(children[i])
end
end
I am not sure what happens here. It's almost as if Self_GetId() reports a wrong EntityId, but as far as I can see, it does not.
175. Q: What is the difference between SoundType.Ambient and SoundType.Max? Both ignore the radius and position arguments of World_PlaySound() and I don't notice a difference.
176. Remark: I have an object wich emmits a looping sound. I want to disable/enable this sound by script, which i do now by setting the volume to 0/1. Maybe there should be a start/stop function.
doesnt matter in this case, as the bone is aligned with the local coordinate system, but should be fixed in case someone copies the code or rotates the bone.
local hits = World_TestRayOverlap(Vec3(0,50,0), Vec3(0,-1,0), 200, CollisionFilterQueryPreset.Bullet)
local hit = hits[1]
--local result,hit = World_Raycast(Vec3(0,50,0), Vec3(0,-1,0), 200, CollisionFilterQueryPreset.Bullet)
local o = Vec3(hit.position)
print("hit at", hit.position,"|",o)
print(EntityId_ToString(Self_GetId()),"hit at", hit.position,"|",o)
Output:
hit at Vec3(0, 7.08394, 0) | Vec3(0, 7.08394, 0)
EntityId(name: Glider, index: 3500 use: 14) hit at Vec3(1.84687e+025, 1.48662e+022, 4.36046e+027) | Vec3(0, 7.08394, 0)
Is EntityId_ToString doing something special? No issues if World_Raycast is used instead of World_TestRayOverlap. Should I always make a local copy if I use structs that are passed from other functions? What about this one:
154. I changed it to be case insensitive. I was going to add versions of Self_GetComponents() and Self_GetComponent() that take a ComponentType but there were complications so holding off on that.
155. This is not possible currently.
156. Done.
157. I think you handled this the right way. We can consider adding a helper function to check if any entity in the hierarchy has a tag.
158. Just fixed this one yesterday. It should no longer display when in editor mode.
159. You can select the field by clicking on it and then press the backspace key.
160. Good idea. Added a card.
161. Reproduced. Added a card. I noticed you can deselect the entity and reselect to "fix" this problem.
162. I have noticed this type of thing myself in the past but haven't been able to figure out the repro steps. This didn't do it for me. I tried in the blueprint editor so maybe you have other steps.
163. Fixed. Thanks!
164. Added a card.
165. Hmm, a tricky one! We will have to figure something out.
166. I know what you mean. Added a card.
167. I cleaned up that code. Entity_AddForce exists because it is a very common operation and we didn't want to require every object in the world to have a script to allow forces to be applied. It doesn't break encapsulation that much since force can already be applied to another entity by moving into it.
168. The team spawning code is pretty old and I was never happy with it. Perhaps I should just remove it for now? It would probably be better to have a dedicated "team spawner" script rather than trying to reuse the existing one.
169. There currently is not. I can't think why there shouldn't be so I have added a card to add OnPlayerLeave.
170. I can't find the code right now but I believe the code that generates those automatically fills the table in reverse too. It lets you convert from the code to the name.
171. There is a limit right now. We didn't intend to add a bunch more keys but if you can make a good argument we will of course listen.
172. Hard to tell really what is going on here. Do you set the position of the puck at some point after the code you posted?
173. This was recently fixed. I was using MagicaVoxel too
174. The first play sound causes the sound entity to be attached to the entity. Then in your SetText message receiver you destroy all the children. So the sound entity is a child of your child and is also destroyed. You can add a delay to avoid this with timedcallback perhaps.
175. haha, SoundType.Max is not really supposed to be exposed (not exposed in the UI for instance). It just denotes the number of items in our enum. Ambient and Effect are the only 2 types right now.
176. This is the correct way.
177. Fixed. Thanks!
Entity Ids are structs since an entity can stop existing while something is holding a reference. We update the entity Ids automatically to null them out in this case.
Are you absolutely sure about that?
Seems like whatever creates server.lua only has to add a single check to the function
function Entity_GetIsValid(entityId)
if entityId == nil then
entityId = EntityId_Invalid
end
return (GameLib.FFI_Server_Entity_GetIsValid(world, entityId))
end
The same check already exits in a bunch of other functions (World_SpawnBlueprint, World_PlaySound, World_PlayParticles)
162. I have noticed this type of thing myself in the past but haven't been able to figure out the repro steps. This didn't do it for me. I tried in the blueprint editor so maybe you have other steps.
That baffles me, since it always happens. Blueprint editor and world editor. Whenever a script component is reused for anther script, all values are nulled instead of their default value.
165. Hmm, a tricky one! We will have to figure something out.
I imagine it to be straight forward.
For each package in packages from package.info,
if package does not start with "steam:" and package not in {steam}\FuturePerfect\packages
return warning message
168. The team spawning code is pretty old and I was never happy with it. Perhaps I should just remove it for now? It would probably be better to have a dedicated "team spawner" script rather than trying to reuse the existing one.
Can't tell. The team spawners are certainly not easy to use right now, but it is possible to make them work (e.g. I use them). Probably best to leave them until they get replaced.
171. There is a limit right now. We didn't intend to add a bunch more keys but if you can make a good argument we will of course listen.
Regarding the default keys, there are a bunch of easily reachable keys that are not bound
Alt, Q, F (notice how G is bound), Mouse2 (mousewheel), (Mouse3, Mouse4)
Regarding functionality, there are some functions that probably deserve their own key:
roll left/right (for descent-like games), peek left/right (for shooters)
sneak
quick switch/last weapon
map
flashlight
taunt
onscreen menu/radio menu/buy menu
Here is my situation: 2 keys left and at least 3 actions to assign
assigned actions:
forward,left - move
movementModifier - turbo
drop - drop puck
jump - shoot puck
next,prev - scroll through items
primary - use item (primary mode)
secondary - use item (secondary mode)
jump - shoot
crouch - look behind (hold)
1-9 - inventory access
free actions:
reload
use
actions I want to assign a key to:
free look/look around (modifier)
change camera perspective (10 different ones)
show map
172. Hard to tell really what is going on here. Do you set the position of the puck at some point after the code you posted?
I don't set the position (well I do, but disabled that for this video). I updated the workshop with a minimal version.
I think it is important here that the object to catch is physics enabled and the catching entity is the player. The faulty location is the global coordinates (in the video: +16 in z direction) added relative to the player in his reference frame.
Comments
- H behaves strangely. Typing 250 results in 249. Pushing the slider to 360 results in 300. Typing 500 gives 319.
- Setting S to 0 or nothing resets H to 0, which makes it hard type in the fields directly. Same for V, which resets S and H
- circle should adjust colors to current V value or have an extra colored selector for V like this(src)
- Alpha is missing (?)
- Support for Hex-codes
Thanks. I made a card for these and will get to it soon.126. Bug: Calling World_SendMessage[ToAll]() in an Automator (before the round is started) crashes the game.
I've fixed the crash, however the Automators cannot send messages. What are you trying to use this for?
127. Feature Request: CollisionFilterTriggerPreset that triggers on everything but not Self. I want to add a triggerbox to a moving physics object (the box overlaps the collision model), but it is triggered by its own collision component on spawn. I need a filter presets that ensures that message.triggerId is not the same as message.otherId. Currently its only solvable by setting the collision type to hitbox, which I do not want.
This sounds like maybe this should just be the behavior regardless of the preset. I'll chat with the rest of the team about it.
128. Bug: Grid increase/decrease does not work with every keyboard layout (i.e. german)
What exactly is the problem? Is it the default keys don't work, but remapping them does, or does it not work even remapped?
129. Q: I am not supposed to add 2 directional lights, right? Only one of them casts a shadow, and which one depends on the current viewing angles.
This is limitation of the current implementation. I'll make it so that more than 1 directional light is supported in the (near) future.
130. Remark: Model editor does not open for standard models. Not sure if it should. Works fine for self made ones.
This is the intended behavior since you don't have the "source" files for the other models (i.e. the fbx files). In the future I'm going to make it so they open in a "read only" mode like a model viewer.
131. Bug: Start editor, add package, close package edit view -> crash
Fixed.
Edit: my first impulse was to add the OnSetText() event handler to the automator.
Ha, did not even see they can be remapped. Remapping works, but with the default mapping (LeftBracket/RightBracket) on a german keyboard layout, no key or key combination triggers the increase or decrease of the grid size. In particular not even pressing [ or ], which on a german keyboard are mapped to AltGr+8 and 9. Switching to US layout (where [ and ] are to the right of the P) in windows works.
Maybe related: with a german keyboard layout enabled, everything under Options->Bindings still registers the US keys. I.e. hitting + or #, which have their own keys, registers as = and 3. Keys that don't have an equivalent like ü,ö,ä,ß don't register at all. NS2 had this right.
Don't actually need it and it's probably rare that it makes sense (i.e. 2 suns), just noticed the slight glitches.
At last, I want to bring attention to #117
It looks like some changes were made in NS2 after I forked the codebase for Future Perfect. I've integrated these now and made a new build in the "beta" branch. When you get a chance, could you check it out and see if it's working the way you'd expect?
We could add that, although it would have to do some computations for something like rag doll where there might be multiple bodies. What are you going to use it for?
Reassigning key works now, similar to NS2 (that is keys like "ü" show up as "#41" but work as expected). Still "LeftBracket" and "RightBracket" cannot be triggered with the german layout.
I have only one solid object and do not care about ragdolls. I need to know the location of the center of mass, so I know where to add pairs of local forces that have point symmetry. Here, I made a video. Note that the 4 hoverpads are not symmetric around the center of mass, which causes extensive swinging.
Ok, that looks cool. I added the function (Collision_GetLocalCenterOfMass). It's available in the beta branch right now.
OK so now I want to see a Wipeout mod in FP.... :P
132. Remark: Nvidia Shadowplay screen recording autodetects the game title as "Tian Long Ba Bu". No idea why or how, but maybe that String looks familiar to someone.
133. Bug: Checkboxes and String fields for automators don't work anymore (are constantly overwritten with previous value)
134. Remark: messages.html: missing OnCreate, OnDestroy
133. I noticed this myself yesterday. It has been added to our bug list. Thanks!
134. Done!
Might also be worth noting that currently .fbx files on my system do not currently have a default program.
135. Bug:: Automators are broken.
- start editor
- start game (F5)
- stop game (shift+F5)
- change size of floor entity -> no effect
Edit: fixed by yesterday's updateI know it's a long shot and probably a low priority if it were to be done, but if you don't ask..
I have some ideas for 3d music based games I'd like to try.
I will add a card in Trello but it will have to be a lower priority task for now as we have some bigger things we need to get to in the short term. Thanks!
137. Bug: World_SendMessage/World_SendMessageToAll nowhere mentioned in api.html or the rest of the docs
138. Bug:: I noticed that a function that is bound the to mousewheel (Action.Next) is always triggered twice. Here is some example code
upcount = 1 function countup() upcount = upcount + 1 print(upcount) end function OnInput(input) if bit.band(input.actions, Action.Next) ~= 0 then countup() end endSlowly turning the mousewheel will produce outputs like 2 2 then 3 3, then 4 4Right now I have a hard time to make sure that a function is only called once per button press.
Until here, that seems like a minor annoyance, but I actually believe this to be a major bug, and here is why:
When you turn your mousewheel rapidly, the output changes
Tests indicate that for players connected over the network, the function is only called once.
Making the functions and variables local does not change anything.
139. Feature Request: "if bit.band(input.actions, Action.Next) ~= 0 then ..." is nice and all, but maybe there should be a high level function like IsPressed(Action.Next). While we are at it: when you want to query the state of a button outside of the OnInput function, you need to introduce a global variable and sync it with the input bitmask, why not have the current input state available globally at all times?
140. Remark: Consider the following code
function OnTest(message) timedfunction(message.var, 0) end function timedfunction(var, num) print(var) if num < 5 then timedcallback(1/30, function() timedfunction(var, num+1) end) end end World_SendMessageToAll("Test",{var=Vec3(1,1,1)})Output: If it's not immediatly clear what's happening here, I am accessing freed chunks of memory here. I am not too familiar with the internals of Lua, but I suppose that the CDATA block message.var is passed as reference and then deallocated after the OnTest function is left. timedcallback then fails to make it's own copy of the object. I can circumvent this by changing to call to timedfunction(message.var*1, 0) [or in my case timedfunction(message.targetCoords*Coords.GetIdentity(), 0)]141. Bug: Movement code: running diagonally agains small walls (i.e. wall_01_small or 2 crates) and spamming the jump button results in high jump altitudes
142. Q: Is there a way to have a child entity that does not rotate with it's parent, but still moves along with it? Other by manually rotating the child in the opposite direction of the parent.
143. Q: Is there a way to have an object that is not visible to everyone? Think of cloaked skulks of waypoint breadcrumbs.
Additional Remarks to 123.:
Recently I have worked alot with moving child entities, and came to the conclusion that the local version like Self_GetLocalCoords (and its pendants for angles, positions and setters and getters) are the ones you almost always want. Considering how Bone_GetCoords already returns the local coordinates, I propose that the generic function like *_GetCoords, *_SetAngles, ... work in the local reference frame and specialized ones are added to access the global/absolute coordinates.
136. Bug:: single character variables like
Thanks, I fixed it.
137. Bug: World_SendMessage/World_SendMessageToAll nowhere mentioned in api.html or the rest of the docs
The API documentation is currently only including functions that are implemented in C++. These particular functions are implemented in Lua. We'll include them in the future though.
138. Bug:: I noticed that a function that is bound the to mousewheel (Action.Next) is always triggered twice. Here is some example code
This sounds like it's behaving properly and what you are seeing is the result of running the code on both the server and client. If you change it to print(World_GetName(), upcount) you should be able to see the two different sources. The same code may run on the client multiple times as well due to client side prediction, but before it runs each time it should reset upcount and thus produce the same result.
139. Feature Request: "if bit.band(input.actions, Action.Next) ~= 0 then ..." is nice and all, but maybe there should be a high level function like IsPressed(Action.Next). While we are at it: when you want to query the state of a button outside of the OnInput function, you need to introduce a global variable and sync it with the input bitmask, why not have the current input state available globally at all times?
This has to do with the networking. It would probably make sense to introduce some more concise way of determining of a button is pressed, but adding a global function wouldn't make sense in all contexts. For example, say you have a script on an object which is not a player; what should the functions return in that case? I will give some consideration how this could be done in a way that doesn't lead to confusing behavior.
140. Remark: Consider the following code
Yes you are right, this is a good catch. Another way to do it is like Coords(message.targetCoords) to make a copy of the variable. Fixing problems like this in luajit is one of the hard tasks we still have to solve. What I'd really to to do is make it so that a = b has copy semantics if b is a "structure" (i.e. a Vec3, Coords, etc.).
141. Bug: Movement code: running diagonally agains small walls (i.e. wall_01_small or 2 crates) and spamming the jump button results in high jump altitudes
This is caused by the logic in the movement code that gives you a little extra boost when you're running up a hill in jumping. Without the boost the jump feels very ineffective. The consequence is that jumping on a steep or nearly vertical slope gives you a big boost.
142. Q: Is there a way to have a child entity that does not rotate with it's parent, but still moves along with it? Other by manually rotating the child in the opposite direction of the parent.
Not at the moment, but one feature I've had in mind is make a child entity that doesn't inherit its parent's transform. This is something we could use to place the "ghost" cube for the Minecraft-like building without having the networking delay. This could be an extension to that (i.e. only inherit position, only inherit rotation, inherit full transform). What's your use case?
143. Q: Is there a way to have an object that is not visible to everyone? Think of cloaked skulks of waypoint breadcrumbs.
One way of doing this I've been considering is allowing the camera to include tags that affect what it renders. What do you think about that?
Additional Remarks to 123.:
Recently I have worked alot with moving child entities, and came to the conclusion that the local version like Self_GetLocalCoords (and its pendants for angles, positions and setters and getters) are the ones you almost always want. Considering how Bone_GetCoords already returns the local coordinates, I propose that the generic function like *_GetCoords, *_SetAngles, ... work in the local reference frame and specialized ones are added to access the global/absolute coordinates.
I came to the opposite conclusion (at least that if you didn't have an explicit reason to get the local coordinate frame, you wanted the global one) when I named them, but I'll take your feedback into consideration.
Ok that makes sense. I guess the proper questions is then, why the client side prediction and the server are so easily (and permanently) desynchronized even when client and server are on the same machine, and what I can do to syncronize them again.
Server 2
Predict 3
Server 3
Predict 4
Server 4
Predict 5
Predict 6
Server 5
Predict 7
Server 6
Predict 8
Server 7
Predict 9
Server 8
Predict 10
Server 9
Predict 11
Server 10
Predict 12
Server 11
Predict 13
Predict 14
Server 12
Predict 15
Predict 16
Server 13
Predict 17
Server 14
Predict 18
Server 15
Predict 19
Server 16
Predict 20
Predict 21
Server 17
Predict 22
Server 18
Predict 23
Server 19
Predict 24
Server 20
Predict 25
Server 21
Predict 26
Server 22
Predict 27
Server 23
Predict 28
Server 24
Predict 29
Server 25
Predict 30
Server 26
Predict 31
Predict 32
Server 27
Predict 33
Server 28
Predict 34
Server 29
Predict 35
Server 30
Predict 36
Server 31
I did in no way propose to replace the bitmask with something else. I know why it is there and why it should stay there. All I said is, that having to use logical operations on bitmasks to get a button state is unnecessary complicated, is not very readable and hence makes the game a little bit less accessible. That's why there should be some higher level mechanism that hides the underlying bitmask.
There are multiple ways to achieve that.
The most simple would be to have a function like
function IsKeyPressed(inputs, action) return bit.band(input.actions, action) ~= 0 end function OnInput(input) if IsKeyPressed(input, Action.Next) then ...But having to pass the input everytime, when it's the only thing you could legitimately pass to that function seems redundant, which is why I proposed to have the current keypresses in a global variable, and then have a global IsPressed(action) function. If no player is "attached" to an entity, the question if a certain key is pressed would of course always yield 'false'. If a player is attached, the global actions would be updated accordingly. However, this implies that only one player can control an entity at a time (which I assume but haven't checked yet).
If this is too limiting, then the current button states could be a property of the player (which is already carrying some data fields), and the call could be made more explicit:
Another approach would be to have a bunch of specialized functions
I was not aware there is a copy-constructor. Now I dont have to write a copy function to eliminate that unnecessary multiplications. Thanks.
I dare to object.
To be honest, I hadn't checked the movement code before (because I am no longer using it). If I had, I would have seen the comment in the part you have mentioned, that the jumpBoost has a potential to be exploited, and wouldn't have made a bug report.
But while it probably has it's own problems, the jumpBoost is not the problem here!
The problem lies few lines above: The OnInput function is executed continously (btw, Why?), and the second part (and isOnFloor) is true on multiple of those calls, which causes the queued jump to be added multiple times during 'queuedJumpTime'. Easy fix would be 'timeLastJumpDesired = -1' in the 'if isOnFloor' part (around line 210)
Well, a picture is worth a thousand words. Here are 60 pictures a second:
So I have my glider (body) and a camera (head). As the body is physics enabled, and you cannot have physics enabled child entities, the camera naturally is a child entity of the glider. Now when the body moves, the camera needs to follow. When the camera rotates, the body needs to follow. But as the camera is a child of the body, any rotation of the body also rotates the camera. As a result, the controls feel almost game-pad like. I tried to keep the camera steady by determining the current rotation of the body and that apply a counter rotation on the camera, but the results are mediocre (0:49-1:12). Ultimatly I would want the horizon line to stay fixed, which means the camera moves with the center of mass, but does not get any of the rotations.
The hiccups in the first 20 seconds during the zooming are caused by issue 138 btw.
You mean like a blacklist? Dont render anything whith the following tags...? Sounds good.
Above, you have mentioned the minecraft block placing procedure, which made me thinking. Whether you require global or local coordinates most of the time may depend strongly on your use case, and no matter how you name the functions, there is always someone who will choose the wrong one on the first try.
Sure, I can use tags, but it feels dirty.
145. Bug: using_player.lua:
local highlightColor = Color(255, 0, 0, 255) --- Color
I am aware that Hud_HighlightEntity() is currently broken, so this doesn't matter, but this should be either Color(1.00, 0, 0, 1.00) or Color.red
146. Bug: for a physics enabled collision that is part of a player this code
function OnThink() print(World_GetName(), Collision_GetLinearVelocity(collisionId):GetLength()) endgenerates this output147. Remark: When i make something a child of the player by using Self_SetParent(), that triggers OnCreate() in the Predict context, which was unexpected. Similarly when i detach the parent again, the OnDestroy method is called in the Predict context.
148. Remark: I needed a vector in the local coordinate system. The current way to do it is which is fine. But since I only use cartesian coordinate systems, the axes are guaranteed to have length 1. So the above has some unneccessary multiplications and devisions, when effectivly I only need a simple matrix multiplication (matrix times vector): To my surprise, there was none.
149. Remark: (possibly bug)
When an object is inside a trigger area, changing the collision from Solid to Hitbox or Ghost to Hitbox triggers the Trigger (object leaves the area).
This is relevant for the game bounds (when setup in a way that everything is inside and destroyed on leave), as this effectivly means that nothing can be changed to Hitbox type (after creation), as this would immediately cause the destruction of the entity.
150. Remark:
have changed some values while the game was live. not sure what exactly happend, but this caused some asserts at first and then the game crashed. when i tried to load my mod, the assertion errors returned (havent kept a copy). turns out the collision mesh for the object i had edited at last was no longer in the asset browser. I cleared the roaming/FuturePerfect directory to force a rebuild of the collsion mesh, but it didn't work. In the end, I had to delete to following lines in the .build_options file of the model, after that it everything was generated.
collision=[ { name="" convex_decomposition=false bodies=[ { name="" attach_point="" linear_damping=0.0 angular_damping=0.0 shapes=[ { type="Mesh" radius=-1.#QNAN half_height=0.0 half_extents=[-1.#QNAN 0.0 -1.#QNAN] min_bone_weight=0.5 mass=0.0 density=1.0 } ] } ] } ]151. Bug: attached a player spawner to some entity, was spawned near (0,0,0) because of player_spawner.lua: Should be global positions and angles.
152. Remark: Observed multiple times now, that when someone tries to join my game while I am on the beta channel and he is not, he will download a lot of stuff (expected) but then time out. I am not too woried if betas cannot join non-betas, but maybe the timeout doesn't take ongoing downloads into account.
145. I'll fix that.
I'm looking into 146 now
147. Is this creating a problem for you or just seems wrong? These objects are being created/destroyed in the prediction world but this should maybe be hidden from scripts.
148. You want a Matrix type in script basically?
149. Hmm, we will have to think about this one. That behavior is not great.
I have fixed 150.
151. I changed that to use global position/angles. I am not sure why I was using local there.
152. I suspect it is because of how old the non-beta build is at this point but maybe something unintended is going on here with the timeout. I'll see what I can uncover.
Thanks!
https://trello.com/c/JrMy1lbI/1564-timeout-while-downloading-from-workshop
https://trello.com/c/8UE5lWVF/1563-when-an-object-is-inside-a-trigger-area-changing-the-collision-from-solid-to-hitbox-or-ghost-to-hitbox-triggers-the-trigger-obje
https://trello.com/c/2adM3zTH/1562-add-entity-getchildbyname-entityid-name
We are still deciding on the name thing. It may be a can of worms as we may need to add a lot more of these types of name query functions if we add that one. Names were intended as a debugging mechanism more than anything.
local function Multiply(v, w) if w.origin ~= nil then local result = Coords() result.xAxis.x = v.xAxis.x * w.xAxis.x + v.yAxis.x * w.xAxis.y + v.zAxis.x * w.xAxis.z result.xAxis.y = v.xAxis.y * w.xAxis.x + v.yAxis.y * w.xAxis.y + v.zAxis.y * w.xAxis.z result.xAxis.z = v.xAxis.z * w.xAxis.x + v.yAxis.z * w.xAxis.y + v.zAxis.z * w.xAxis.z result.yAxis.x = v.xAxis.x * w.yAxis.x + v.yAxis.x * w.yAxis.y + v.zAxis.x * w.yAxis.z result.yAxis.y = v.xAxis.y * w.yAxis.x + v.yAxis.y * w.yAxis.y + v.zAxis.y * w.yAxis.z result.yAxis.z = v.xAxis.z * w.yAxis.x + v.yAxis.z * w.yAxis.y + v.zAxis.z * w.yAxis.z result.zAxis.x = v.xAxis.x * w.zAxis.x + v.yAxis.x * w.zAxis.y + v.zAxis.x * w.zAxis.z result.zAxis.y = v.xAxis.y * w.zAxis.x + v.yAxis.y * w.zAxis.y + v.zAxis.y * w.zAxis.z result.zAxis.z = v.xAxis.z * w.zAxis.x + v.yAxis.z * w.zAxis.y + v.zAxis.z * w.zAxis.z result.origin.x = v.xAxis.x * w.origin.x + v.yAxis.x * w.origin.y + v.zAxis.x * w.origin.z + v.origin.x result.origin.y = v.xAxis.y * w.origin.x + v.yAxis.y * w.origin.y + v.zAxis.y * w.origin.z + v.origin.y result.origin.z = v.xAxis.z * w.origin.x + v.yAxis.z * w.origin.y + v.zAxis.z * w.origin.z + v.origin.z return result else local result = Vec3() result.x = w:DotProduct(v.xAxis) result.y = w:DotProduct(v.yAxis) result.z = w:DotProduct(v.zAxis) return result end end(possibly with some better code that checks if w is of type Coords or Vec3)or some extra function
local function InverseTransformDirection(self, v) local result = Vec3() result.x = v.x * self.xAxis.x + v.y * self.xAxis.y + v.z * self.xAxis.z result.y = v.x * self.yAxis.x + v.y * self.yAxis.y + v.z * self.yAxis.z result.z = v.x * self.zAxis.x + v.y * self.zAxis.y + v.z * self.zAxis.z return result end(possibly with a better name, TransformDirection already exists)For comparison, here is what coords:GetInverse():TransformDirection(v) calculates
local xScale = 1.0 / (self.xAxis.x * self.xAxis.x + self.xAxis.y * self.xAxis.y + self.xAxis.z * self.xAxis.z) -- = 1 local yScale = 1.0 / (self.yAxis.x * self.yAxis.x + self.yAxis.y * self.yAxis.y + self.yAxis.z * self.yAxis.z) -- = 1 local zScale = 1.0 / (self.zAxis.x * self.zAxis.x + self.zAxis.y * self.zAxis.y + self.zAxis.z * self.zAxis.z) -- = 1 local result = Coords() result.xAxis.x = self.xAxis.x * 1; result.xAxis.y = self.yAxis.x * 1; result.xAxis.z = self.zAxis.x * 1; result.yAxis.x = self.xAxis.y * 1; result.yAxis.y = self.yAxis.y * 1; result.yAxis.z = self.zAxis.y * 1; result.zAxis.x = self.xAxis.z * 1; result.zAxis.y = self.yAxis.z * 1; result.zAxis.z = self.zAxis.z * 1; local v = Vec3() result.origin.x = -1 * ( self.origin.x * result.xAxis.x + self.origin.y * result.yAxis.x + self.origin.z * result.zAxis.x ) result.origin.y = -1 * (self.origin.x * result.xAxis.y + self.origin.y * result.yAxis.y + self.origin.z * result.zAxis.y ) result.origin.z = -1 * (self.origin.x * result.xAxis.z + self.origin.y * result.yAxis.z + self.origin.z * result.zAxis.z ) local result2 = Vec3() result2.x = v.x * result.xAxis.x + v.y * result.yAxis.x + v.z * result.zAxis.x result2.y = v.x * resultxAxis.y + v.y * result.yAxis.y + v.z * result.zAxis.y result2.z = v.x * result.xAxis.z + v.y * result.yAxis.z + v.z * result.zAxis.z return result2But again, such multiplication are only meaningful if the coordinate axes are normalized. As far as I see, scaled axes are used nowhere, yet.
154. Feature Request: Does Self_GetComponent("Camera") need to be case sensitive? Why even strings when all the componentsTypes are already listed, i.e. ComponentType.Camera?
155. Feature Request: Entity_GetIsValid(nil) should not throw an error, so i can shorten these lines: Also, I had the special case where the argument was a valid Id in the server context, but nil in the predict context.
156. Remark: Short time of confusion until I realised that Camera_SetFov takes radians. Maybe documentation should mention that.
157. Remark: I want to know if I hit a "Ramp".
As I cannot check for the name (#102, #144), I gave every ramp a tag and then checked for that
function OnContact(message) print("contact with", EntityId_ToString(message.otherId), Entity_GetHasTag(message.otherId, "ramp")) endResult: ramp_01_a is a child of "Ramp", so it does not inherit it's tag. I ended up writing a function that recursivly checks all parents for tags.Of course I am not suggesting that children should inherit the tags of their parents, but maybe there is a more elegant solution.
158. Bug: The World_EndGame() overlay is on top of the entity editor
159. Bug: The only method to unset an entity field such as the "attach to" field is to drop it against the background. Kind of difficult to do from the inside of a room Edit: Backspace works fine
160. Feature Request: on grid increase/decrease also show the snapsize for angles
161. Bug: add script
select script field
type "creat"
drop "add force on create" into the script field
select script field
type "creat"
drop "create player on destroy" into the script field
-> height of "offset distance" field is wrong
(not always reproducable)
162. Bug: add any script to an entity. select the script field and drop "add score on activate" instead. notice how the default score is 0, although it is 1 in the script. works with any script.
163. Bug: create_player_on_destroy.lua:
not changing offsets, will produce errors:
Entity position set to a non-finite value
Entity angles set to a non-finite value
here is why:
local lookDir = coords.origin - position
local lookInCoords = Coords.GetLookIn(position, lookDir, Vec3(0, 1, 0))
for offsetDistance = 0, lookdir is Vec3(0,0,0)
this causes lookInCoords to be (Vec3(0,0,0), Vec3(0,0,0), Vec3(0,0,0), someorigin)
which causes lookInCoords:GetAngles() to be nan
solution:
insert
if lookDir == Vec3(0,0,0) then lookDir = Vec3(0,0,1) end
after line 31, or limit the minimum offsetDistance
164. Bug: package manager lacks destinction between local and workshop versions of the same package
165. Bug: could publish a package that depends on other local packages without a warning
166. Feature Request: right click to look around should work while editing an entity
167. Bug: general_spawner.lua:
randomDir unused
spawnRadius unused
spawnAngles unused (!)
use new Entity_AddForce instead of World_SendMessage("AddForce") (although regarding the answer to 106 I am not sure why Entity_AddForce even exists, as it breaks encapsulation) should be same for torque should use math.randomf(0,2*math.pi) (I think)
168. Remark: playerspawners for two teams (in two different spawnGroups) dont handle playerJoin (a player is spawned in each spawnGroup), and therefore need to ignore playerJoin and have a central instance manage all player assignments. Maybe I misunderstand the spawnGroup concept here.
Also on player death a "Dead Player" is spawned, that does not inherit the spawnGroup of the destroyed avatar, hence that information is lost.
169. Q: There is OnPlayerJoin. Is there a similar call once a player leaves? I want to maintain a list of all connected players, so i can decide which team new players should be assigned to. (yes, I could check if playerIds are valid everytime i use them)
170. Q: Action.128 = "Drop" <- for what is this field useful? Isn't Action.Drop = 128 all that is ever needed?
171. Remark: I am slowly running out of assignable actions. There are 19 actions, as far as I understand there is room for 13 (45?) more.
172. Bug: Self_SetParent() vs client prediction
When I catch an object, I run this code on the object: This is what happens
Note how I shoot in x-direction (world coordinates) and for ~1 frame the puck appears in x-direction in the local coordinates of the player.
Both, the player and the puck are physics-enabled.
Is there anything I can do to prevent this?
173. Bug: referee/referee.fbx.build_options:9 : Syntax error: Expected '='
8 materials={ 9 0Mat="referee/referee.material" 10 }0Mat is the texture name that comes right out of MagicaVoxel and if leading numbers are not allowed I need to edit the fbx file, which I would like to avoid. I also believe the error is new, never noticed it before.174. Remark: I have an entity with 2 Scripts. One has the OnSetText() function, the other one looks like this:
function OnTrigger(message) World_SendMessage("SetText", Self_GetId(), {text=""}) World_PlaySound("pads/boost.wav", Vec3(0,0,0), SoundType.Effect, 50, 0.8, Self_GetId()) -- 1 -- World_PlaySound("pads/boost.wav", Self_GetPosition(), SoundType.Effect, 50, 0.8) -- 2 endThe first PlaySound() call does not emmit a sound, the second one does. If I remove the SendMessage, both do.The problem is this function which is called by OnSetText:
local function DeleteChildren() local children = Self_GetChildren() for i=1,#children do World_DestroyEntity(children[i]) end endI am not sure what happens here. It's almost as if Self_GetId() reports a wrong EntityId, but as far as I can see, it does not.175. Q: What is the difference between SoundType.Ambient and SoundType.Max? Both ignore the radius and position arguments of World_PlaySound() and I don't notice a difference.
176. Remark: I have an object wich emmits a looping sound. I want to disable/enable this sound by script, which i do now by setting the volume to 0/1. Maybe there should be a start/stop function.
177. Bug: discombobulizer and buying_station:
Bone_GetCoords(Self_GetBoneComponent("Spawn")) * Self_GetCoords()should beSelf_GetCoords() * Bone_GetCoords(Self_GetBoneComponent("Spawn"))doesnt matter in this case, as the bone is aligned with the local coordinate system, but should be fixed in case someone copies the code or rotates the bone.Regarding 140. (access of freed memory):
local hits = World_TestRayOverlap(Vec3(0,50,0), Vec3(0,-1,0), 200, CollisionFilterQueryPreset.Bullet) local hit = hits[1] --local result,hit = World_Raycast(Vec3(0,50,0), Vec3(0,-1,0), 200, CollisionFilterQueryPreset.Bullet) local o = Vec3(hit.position) print("hit at", hit.position,"|",o) print(EntityId_ToString(Self_GetId()),"hit at", hit.position,"|",o)Output: Is EntityId_ToString doing something special? No issues if World_Raycast is used instead of World_TestRayOverlap. Should I always make a local copy if I use structs that are passed from other functions? What about this one: Unstuck recieves another id than the calling function was sending. Are entityIds structs? I thought they were integers interpreted as doubles?154. I changed it to be case insensitive. I was going to add versions of Self_GetComponents() and Self_GetComponent() that take a ComponentType but there were complications so holding off on that.
155. This is not possible currently.
156. Done.
157. I think you handled this the right way. We can consider adding a helper function to check if any entity in the hierarchy has a tag.
158. Just fixed this one yesterday. It should no longer display when in editor mode.
I will continue on these items on Monday!
160. Good idea. Added a card.
161. Reproduced. Added a card. I noticed you can deselect the entity and reselect to "fix" this problem.
162. I have noticed this type of thing myself in the past but haven't been able to figure out the repro steps. This didn't do it for me. I tried in the blueprint editor so maybe you have other steps.
163. Fixed. Thanks!
164. Added a card.
165. Hmm, a tricky one! We will have to figure something out.
166. I know what you mean. Added a card.
167. I cleaned up that code. Entity_AddForce exists because it is a very common operation and we didn't want to require every object in the world to have a script to allow forces to be applied. It doesn't break encapsulation that much since force can already be applied to another entity by moving into it.
168. The team spawning code is pretty old and I was never happy with it. Perhaps I should just remove it for now? It would probably be better to have a dedicated "team spawner" script rather than trying to reuse the existing one.
169. There currently is not. I can't think why there shouldn't be so I have added a card to add OnPlayerLeave.
170. I can't find the code right now but I believe the code that generates those automatically fills the table in reverse too. It lets you convert from the code to the name.
171. There is a limit right now. We didn't intend to add a bunch more keys but if you can make a good argument we will of course listen.
172. Hard to tell really what is going on here. Do you set the position of the puck at some point after the code you posted?
173. This was recently fixed. I was using MagicaVoxel too
174. The first play sound causes the sound entity to be attached to the entity. Then in your SetText message receiver you destroy all the children. So the sound entity is a child of your child and is also destroyed. You can add a delay to avoid this with timedcallback perhaps.
175. haha, SoundType.Max is not really supposed to be exposed (not exposed in the UI for instance). It just denotes the number of items in our enum. Ambient and Effect are the only 2 types right now.
176. This is the correct way.
177. Fixed. Thanks!
Entity Ids are structs since an entity can stop existing while something is holding a reference. We update the entity Ids automatically to null them out in this case.
Seems like whatever creates server.lua only has to add a single check to the function
function Entity_GetIsValid(entityId) if entityId == nil then entityId = EntityId_Invalid end return (GameLib.FFI_Server_Entity_GetIsValid(world, entityId)) endThe same check already exits in a bunch of other functions (World_SpawnBlueprint, World_PlaySound, World_PlayParticles)That baffles me, since it always happens. Blueprint editor and world editor. Whenever a script component is reused for anther script, all values are nulled instead of their default value.
I imagine it to be straight forward.
For each package in packages from package.info, if package does not start with "steam:" and package not in {steam}\FuturePerfect\packages return warning messageCan't tell. The team spawners are certainly not easy to use right now, but it is possible to make them work (e.g. I use them). Probably best to leave them until they get replaced.Regarding the default keys, there are a bunch of easily reachable keys that are not bound
Alt, Q, F (notice how G is bound), Mouse2 (mousewheel), (Mouse3, Mouse4)
Regarding functionality, there are some functions that probably deserve their own key:
roll left/right (for descent-like games), peek left/right (for shooters)
sneak
quick switch/last weapon
map
flashlight
taunt
onscreen menu/radio menu/buy menu
Here is my situation: 2 keys left and at least 3 actions to assign
forward,left - move
movementModifier - turbo
drop - drop puck
jump - shoot puck
next,prev - scroll through items
primary - use item (primary mode)
secondary - use item (secondary mode)
jump - shoot
crouch - look behind (hold)
1-9 - inventory access
free actions:
reload
use
actions I want to assign a key to:
free look/look around (modifier)
change camera perspective (10 different ones)
show map
I think it is important here that the object to catch is physics enabled and the catching entity is the player. The faulty location is the global coordinates (in the video: +16 in z direction) added relative to the player in his reference frame.