- Most R_SkinAvailable calls should be returning index into demo.skinlist (same numerical value as when demo was recorded), for demo sync
- A handful of general things permit exception for this
- Expose `replaynumskins` (calculated as `(demo.playback ? demo.numskins : numskins)`) to Lua
- There's *always* more that can be done for this, but this is the minimum spec that can at least be somewhat stable
- Fix the last few bugs I could find with thing args
- Move version update code
- Rename internal variables to `thing_[string]args` to make older code merge issues more obvious
SRB2 uses a LOT of mapthing args compared to Hexen (which has none) and ZDoom (which only has them on objects that will never ever activate scripts). So we really badly needed to separate the two if we want attaching scripts to things to be useful.
Adds r_spritefx.cpp
lightlevel should be -255 to 255, relative offset to
normal sector lightlevel.
If RF_ABSOLUTELIGHTLEVEL, mobj_t.lightlevel becomes an
absolute lightlevel in the range 0 to 255.
More opportunities for early rejection in table-wide searches, in anticipation of future work.
- Many circumstances independently implemented string name comparisons. Most of these have been converted to use R_SkinAvailable, with one exception.
- M_CharacterSelectInit already needs to iterate over the characters, so produce the skin hash independently.
Map things are writeable in Lua, which I am pretty certain is a mistake because mapthings are not sent over the network at all. I considered making them net-synced (it would be relatively easy), but it also aligns with another, more "philosophical" issue: Doom generally copies over properties from mapthing_t into mobj_t, and then only refers to it again when needing to respawn an object -- mapthing_t is not really intended to be referred to very often at runtime. At best it's slightly annoying since some objects rely on a spawnpoint for behavior changes, at worst it may make ACS more confusing in the future since Thing and Mobj tags are mixed together or less useful since they wouldn't be able to modify behaviors of objects that are based on args.
So I decided to solve these two issues at the same time; just treat mapthing_t as something to copy values from, like OG Doom does it. This basically just means that special and args are also part of the mobj now instead of the mapthing, which should fill any desire to edit this stuff from Lua, and reduces the number of instances where objects need to check for their spawnpoint to function properly.
Mammoth commit, sorry. I only realised halfway through writing it that SECRET_SKIN was only partially merged.
Ports from 2.2:
- Merge SECRET_SKIN (STJr/SRB2!1474)
- Default skin is now handled by checking all skins for unlock status, and I_Erroring if none are available
- Don't show skin names on game startup, to keep our secrets hidden
- Unlockables now have string variables zallocated.
- For skin names rather than numbers.
- Correctly clean up memory when freeing unlockables and emblems.
Bespoke code:
- For temporary testing. `unlocks.pk3`
- Using this for rapid testing gameboot SOC instead of patch.pk3 because of the intent to turn that into scripts.pk3
- Don't not save gamedata in DEVELOP builds, even if you've used cheats!
- `player->availabilities` is now an array of UINT8
- (MAXSKINS + 7)/8 entries, or 32 bytes.
- Included with XD_ADDPLAYER instead of XD_NAMEANDCOLOR.
- Simplifies a lot of logic with respect to demos, skin changes mid-game, etc.
- Seriously, there's a lot of random places in the code that just iterate over MAXSPLITSCREENPLAYERS and g_localplayers to update availabilities in real time in a way that's not particularly netsafe...
- Lines up with the plan for handling unlocks when returning to menus.
- Was included with XD_ADDBOT, but that actually overruns the netxcmd buffer at first mapload with 7 bots. We might need to consider expanding the size of the netxcmd buffer...
- In demos, can be interpreted as both relative to the original replay and the current skin list depending on boolean context provided to R_SkinUsable.
- Used for SF_IRONMAN (and will crash if all other skins are inaccessible).
- Grand Prix bot randomisation uses the host's unlocks.
- Don't show locked characters on the fancy new character select.
- DXD_JOINDATA for demos
- Replaces the dual-purpose behaviour of DXD_PLAYSTATE
- Contains availabilities
- Handles bot material in a different way
- Forceskin restrictions
- Won't run in demos, because it's assumed recorded DXD_SKIN will handle all the conversions the original match had
- Won't run if K_CanChangeRules says no
- Correctly set `mapvisited` on level visit, even in [fake gasp] MULTIPLAYER/NETGAMES!! 🥹
- Added updating unlockables and extra emblems on `mapvisited` update.
- Currently fails to produce the cecho, but that'll be stripped out entirely in a future commit so I'm not bothered.
Gremlins happened whenever P_TryMove and P_SlideMove/P_BounceMove disagreed on what an object collided with. When TryMove said you collided with a line, but P_BounceMove said that you didn't, then you'd get gremlin'd.
To fix this, P_TryMove now can edit a struct to contain information on what it collides with. P_SlideMove and P_BounceMove no longer try to detect walls on their own and now requires this result from P_TryMove. If a slide/bounce is needed without moving the object, then you'd want to use P_CheckMove to get the result.
Lua is not supported right now.
This makes it significantly easier to save/restore the state of these variables, whenever we need to do so for calling movement functions in the middle of other movement functions. This will also make it easier to move it out of global variable hell if desired later.
Water skipping is enabled for Orbinaut, Jawz, and Ballhog currently. Jawz can water run, as long as their target is on/above the plane they're at -- once their target goes into the water, they'll start skipping.
Squashed SRB2 2.2 Public uncapped without UDMF merge
Implement interpolation at the renderer level
Instead of interpolating thinkers, we interpolate mobjs inside the
renderer. Further interpolation is TBI.
Place Frame Interpolation in "Experimental" video options header
This seems like an appropriate way to describe the feature for now.
Add smooth level platter under interpolation, `renderdeltatics`
`renderdeltatics` can be used as a standard delta time in any place,
allowing for smooth menus. It will always be equal to `realtics`
when frame interpolation is turned off, producing consistent
framerate behavior everywhere it is used.
Add smooth rendering to save select screen
Add smooth rendering to Record/NiGHTS Attack, F_SkyScroll
Ensure viewsector is accurate to viewx/viewy
This fixes a potential crash in OpenGL when changing between levels.
Ensure + commands get executed before map start
Always have precise_t defined
Fix misc dropshadow issues
Reset view interpolation on level load
Remove unnecessary precipmobj thinker hack
Add reset interpolation state functions
Reset precip interpolation on snap to ceil
Reset mobj interp state on TeleportMove
Only swap view interp state if a tick is run
Run anti-lag chasecam at tic frequency
Fixes jittery and unstable chasecam in high latency netgames
Homogenize mobj interpolations
Add sector plane level interpolations
Add SectorScroll interpolator
Add SideScroll interpolator
Add Polyobj interpolator
Intialize interpolator list at a better time
Delete interpolators associated with thinkers
Interpolate mobj angles and player drawangle
Interpolate HWR_DrawModel
Add functions to handle interpolation
Much less code duplication
P_InitAngle, to fix angle interpolation on spawning objects
Fully fix drop shadows
It used the thing's floorz / ceilingz directly -- that wouldn't account for interpolated coordinates.
Do not speed up underwater/heatwave effect in OpenGL
Closer OpenGL underwater/heatwave effect to Software
Interpolate from time of previous tic
Previously interpolated from last 35th of a second, which
may be offset from game time due to connection lag.
Consider this the proper fix to 6ecac4159a too.
Calculate FPS stuff even if frame is skipped
I decided ultimately to actually keep the frame skip optimization disabled, because I think it is actually a little bit helpful that you can still get accurate rendering perfstats while paused, however if we decide otherwise then we can have this optimization back without making the game act like it's lagging.
Keep rect in memory
Feel better about this than creating one all da time
Lots of FPS stuff
- Disabled VSync, due to the numerous problems it has.
- Instead, added an FPS cap.
- Frame interpolation is now tied to fpscap != 35.
- By default, the FPS cap is set to the monitor's refresh rate.
- Rewrote the FPS counter.
(This also consolidates several more commits ahead of this
fixing various issues. -eid)
Misc changes after Kart cherry-picks
Fix renderdeltatics with new timing data
Update mobj oldstates before all thinkers
Allow FPS cap values
Adjust how FPS cap is checked to improve FPS stability
Fix precip crash from missing vars
Improve the framerate limiter's timing for extreme stable FPS
Handle the sleep at the end of D_SRB2Loop instead of the start
Simplifies logic in the other parts of the loop, and fixes problems with it frequently waiting too long.
Reset mobj interp state on add
Add mobj interpolator on load netgame
Move mobj interpolators to r_fps
Dynamic slope interpolators
I_GetFrameTime to try and improve frame pace
(It doesn't feel that much better though.)
Move I_FinishUpdate to D_SRB2Loop to sync screen updates with FPS cap, use timestamps in I_FrameCapSleep to simplify the code
Fix plane interpolation light level flickering
Fix flickering plane interpolation for OpenGL in the exact same way
Funny OpenGL renderer being at least 50% copy-pasted Software code :)
P_SetOrigin & P_MoveOrigin to replace P_TeleportMove
Convert P_TeleportMove use to origin funcs
Revert "P_InitAngle, to fix angle interpolation on spawning objects"
This reverts commit a80c98bd164a2748cbbfad9027b34601185d93f5.
Waypoint polyobjects interpolate z & children
Add interpolation to more moving plane types
Adds interpolation to the following:
- Crumbling platforms
- Mario blocks
- Floatbob platforms (this one works really strangely due to two thinkers, maybe double-check this one?)
Reset overlays interp states each TryRunTics
Interpolate model interpolation (lol)
Use interp tracer pos for GL linkdraw
Papersprite angle interpolation
Makes the ending signpost smooth
Move intermission emerald bounce to ticker
Bring back shadows on polyobjects
Also optimizes the method used so rings can show their shadows too. Using just the subsector is a tad bit imprecise admittedly but any more precise methods get really laggy.
Fix a bunch of ticking in hu_ drawing functions
Revert "Reset overlays interp states each TryRunTics"
This reverts commit a71a216faa20e8751b3bd0157354e8d748940c92.
Move intro ticking out of the drawer
Adjust 1up monitor icon z offsets
Fixes interpolation issues with 1up monitors.
Delta time choose player menu animations
Add drawerlib deltaTime function
Interpolate afterimages further back
Use old sleep in dedicated mode
Clamp cechotimer to 0
Fixes issues with cechos staying on-screen and glitching out
(NiGHTS items for example).
Revert "Remove unnecessary precipmobj thinker hack"
This reverts commit 0e38208620d19ec2ab690740438ac2fc7862a49e.
Fix frame pacing when game lags behind
The frame timestamp should've been made at the start of the frame, not the end.
Fix I_FrameCapSleep not respecting cpusleep
Jonathan Joestar bruh
Allow dedicated to use precise sleep timing again
Instead of only using one old sleep, just enforce framerate cap to match TICRATE.
Make Lua TeleportMove call MoveOrigin
Reset Metal fume interp state on appear
Add interpdebug
Put interpdebug stuff in perfstats instead
Add timescale cvar
Slow the game down to debug animations / interpolation problems! Speed it up if you need to get somewhere quickly while mapping!
Enable timescale outside of DEVELOP builds
It has NETVAR, so it should be fine -- put an end to useful debugging features excluded in multiplayer!
Force interpolation when timescale != 1.0
Reset old_z in MT_LOCKON think
Fixes interpolation artifacting due to spawn pos.
Fix cutscenes in interp
Fix boss1 laser in interp
Interpolate mobj scale
Precalculate refresh rate
Slower PCs can have issue querying mode over and over. This might kinda suck for windowed mode if you have different refresh rate displays but oh well
Fix interp scaling crashing software
Reset interp scale when Lua sets .scale
Disable angle interp on fresh mobjs
Fix interp scale crash for hires sprites
Interp shadow scales
Copy interp state in P_SpawnMobjFromMobj
Fix multiplayer character select
Don't interpolate mobj state if frac = 1.0
Fix Mario block item placement
Interpolate spritescale/offset x/y
Fix offset copies for SpawnMobjFromMobj
THANKS SAL
Add Lua HUD drawlists
Buffers draw calls between tics to ensure hooks
run at the originally intended rate.
Rename drawerlib deltaTime to getDeltaTime
Make renderisnewtic is false between tics
I know what I'm doing! I swear
Completely refactor timing system
Time is now tracked internally in the game using I_GetPreciseTime
and I_UpdateTime. I_Time now pulls from this internal timer. The
system code no longer needs to keep track of time itself.
This significantly improves frame and tic timing in interp mode,
resulting in a much smoother image with essentially no judder at
any framerate.
Ensure mobj interpolators reset on level load
Ensure view is not interpolated on first frame
Disable sprite offset interpolation (for now)
Refactor timing code even more
System layer is greatly simplified and framecap
logic has been moved internally. I_Sleep now
takes a sleep duration and I_SleepDuration
generically implements a precise sleep with spin
loop.
Adjust spawned mobj z by gravity inversion
Only check VERTICALFLIP for old_z calc
Fix Marathon Mode menu for uncapped
Move screenshot code before timing
Only play emerald hunt sounds on new tics
Restore deleted zofs (fixes signpost sparkles etc)
Revert to before screenUpdate boolean was added
Was done this way for the sake of an older version of uncapped's timing; now that the new timing is much better, I decided it should remain as close to pre-uncapped as it can be.
Run UpdateLevelInterpolators in preticker
FixesKartKrew/Kart-Public#10
Run UpdateMobjInterpolators in preticker
Reset view interp in preticker
Add per-split player view resets
All my work thus far on solving the drawflag/renderflag/additive/subtractive conundrum.
Outstandng problems:
* Bad additive/subtractive tables means that they appear opaque except under certain conditions.
* No support for FOFs, Polyobjects, or linedefs in OpenGL yet.
* All OpenGL support mostly done blind, may or may not function in practice.
If nothing else, the hard engineering problems are solved and it's just bug hammering...
Index and take length of tag lists like a table, 1-indexed. There are three
methods which may be used on tag lists:
list:iterate() - returns an iterator over the tags in the list
list:has(tag) - returns a boolean whether the tag is in the list
list.shares(list2) - returns whether two lists share a tag
"find" is also an alias to "has". Each method may be accessed from the global
taglist library too, e.g. taglist.iterate(list)
Tag lists may be compared with an equality operator too. This will tell you if
the two lists are composed of identical tags.
Accessible from sector.taglist, line.taglist and mapthing.taglist.
The global "tags" can be iterated upon for every unique tag which is set in the
level. If a tag is set on a sector/line/thing, it will be included. Taking the
length of "tags" will give you the number of these unique tags. (If a tag is
set on multiple sectors/lines/things, it will only be counted once though.)
For sectors, lines and mapthings, call the field "tagged". This function takes
one argument, which is the tag. The return value can be iterated over for all
the sectors/lines/things with that tag. The length can also be taken for the
number of such objects. If no argument is given, the global tag is default.
I remember when I tried to fix wolfs' branch I was having issues, but this ended up working pretty much fine? Don't remember why the old branch was messed up though (it was simply too old & not substantial enough to try and work off of imo)
