* Smoothing groups optimization, save 265ms with configs, more on production & w/ space ruins (#71989)
This one is fun.
On every /turf/Initialize and /atom/Initialize, we try to set
`smoothing_groups` and `canSmoothWith` to a cached list of bitfields. At
the type level, these are specified as lists of IDs, which are then
`Join`ed in Initialize, and retrieved from the cache (or built from
there).
The problem is that the cache only misses about 60 times, but the cache
hits more than a hundred thousand times. This means we eat the cost of
`Join` (which is very very slow, because strings + BYOND), as well as
the preliminary `length` checks, for every single atom.
Furthermore, as you might remember, if you have any list variable set on
a type, it'll create a hidden `(init)` proc to create the list. On
turfs, that costs us about 60ms.
This PR does a cool trick where we can completely eliminate the `Join`
*and* the lists at the cost of a little more work when building the
cache.
The trick is that we replace the current type definitions with this:
```patch
- smoothing_groups = list(SMOOTH_GROUP_TURF_OPEN, SMOOTH_GROUP_FLOOR_ASH)
- canSmoothWith = list(SMOOTH_GROUP_FLOOR_ASH, SMOOTH_GROUP_CLOSED_TURFS)
+ smoothing_groups = SMOOTH_GROUP_TURF_OPEN + SMOOTH_GROUP_FLOOR_ASH
+ canSmoothWith = SMOOTH_GROUP_FLOOR_ASH + SMOOTH_GROUP_CLOSED_TURFS
```
These defines, instead of being numbers, are now segments of a string,
delimited by commas.
For instance, if ASH used to be 13, and CLOSED_TURFS used to be 37, this
used to equal `list(13, 37)`. Now, it equals `"13,37,"`.
Then, when the cache misses, we take that string, and treat it as part
of a JSON list, and decode it from there. Meaning:
```java
// Starting value
"13,37,"
// We have a trailing comma, so add a dummy value
"13,37,0"
// Make it an array
"[13,37,0]"
// Decode
list(13, 37, 0)
// Chop off the dummy value
list(13, 37) // Done!
```
This on its own eliminates 265ms *without space ruins*, with the
combined savings of turf/Initialize, atom/Initialize, and the hidden
(init) procs that no longer exist.
Furthermore, there's some other fun stuff we gain from this approach
emergently.
We previously had a difference between `S_TURF` and `S_OBJ`. The idea is
that if you have any smoothing groups with `S_OBJ`, then you will gain
the `SMOOTH_OBJ` bitflag (though note to self, I need to check that the
cost of adding this is actually worth it). This is achieved by the fact
that `S_OBJ` simply takes the last turf, and adds onto that, meaning
that if the biggest value in the sorting groups is greater than that,
then we know we're going to be smoothing to objects.
This new method provides a limitation here. BYOND has no way of
converting a number to a string at compile time, meaning that we can't
evaluate `MAX_S_TURF + offset` into a string. Instead, in order to
preserve the nice UX, `S_OBJ` now instead opts to make the numbers
negative. This means that what used to be something like:
```dm
smoothing_groups = list(SMOOTH_GROUP_ALIEN_RESIN, SMOOTH_GROUP_ALIEN_WEEDS)
```
...which may have been represented as
```dm
smoothing_groups = list(15, MAX_S_TURF + 3)
```
...will now become, at compile time:
```dm
smoothing_groups = "15,-3,"
```
Except! Because we guarantee smoothing groups are sorted through unit
testing, this is actually going to look like:
```dm
smoothing_groups = "-3,15,"
```
Meaning that we can now check if we're smoothing with objects just by
checking if `smoothing_groups[1] == "-"`, as that's the only way that is
possible. Neat!
Furthermore, though much simpler, what used to be `if
(length(smoothing_groups))` (and canSmoothWith) on every single
atom/Initialize and turf/Initialize can now be `if (smoothing_groups)`,
since empty strings are falsy. `length` is about 15% slower than doing
nothing, so in procs as hot as this, this gives some nice gains just on
its own.
For developers, very little changes. Instead of using `list`, you now
use `+`. The order might change, as `S_OBJ` now needs to come first, but
unit tests will catch you if you mess up. Also, you will notice that all
`S_OBJ` have been increased by one. This is because we used to have
`S_TURF(0)` and `S_OBJ(0)`, but with this new trick, -0 == 0, and so
they conflicted and needed to be changed.
* Sorting how did I miss it
Co-authored-by: Mothblocks <35135081+Mothblocks@users.noreply.github.com>
Co-authored-by: Funce <funce.973@gmail.com>
* Changes the missing food icon test to cover ALL /obj's
* Update implant.dm
* Hopefully fixes all the failing integration tests!
* Fixes more missing icons
* Even more icon fixes
* Hopefully that was all of them
* Okay now SURELY that's all of them
* I'm tired of this shit man
* Hopefully that's all, for real this time!
Co-authored-by: ShizCalev <ShizCalev@users.noreply.github.com>
Co-authored-by: Zonespace <41448081+Zonespace27@users.noreply.github.com>
Co-authored-by: GoldenAlpharex <jerego1234@hotmail.com>
Co-authored-by: GoldenAlpharex <58045821+GoldenAlpharex@users.noreply.github.com>
* Save 2.2s minimum (with zero ruins, likely a good bit more in production) of atom init time (#69564)
Pre-sort smoothing_groups and canSmoothWith
Without any ruins, these sorts were taking more than 0.6s, and the bulk of the runtime cost of sortTim during init time.
This only happens on init and they are never changed apart from that, so pre-sorts everything and adds a unit test (in the form of #ifdef UNIT_TESTS, because you can't initial a list) to ensure that they are proper.
Keep visibilityChanged() to mapload only for turf/Initialize
Saves about 0.4s worst case scenario (e.g. with no ruins). Very expensive code (175k loop iterations) for 0 side effects.
Space areas now have the fullbright overlay, not the space turfs
Saves about 0.8s worst case scenario. Seems to work fine with starlight.
Remove is_station_level check for window spawners assigning RCD memory.
Saves about 0.3s worst case scenario. The logic for this isn't consistent since neither walls nor floors check this (for performance), plus some minor micro-opts to spawners.
Optimize is_station_level
Doubles in speed, used heavily in /turf/open/floor and in other initialization procs. Bit hard to tell exactly how much is saved, though.
* Save 2.2s minimum (with zero ruins, likely a good bit more in production) of atom init time
* Hopefully fixes the broken CI
* Okay now it shouldn't be failing CI anymore (hopefully)
* Fixes even more issues with smoothing_groups, this time hopefully for good
* Okay NOW it's going to pass CI, surely...
* Okay haha what if it passes this time? :)
Co-authored-by: Mothblocks <35135081+Mothblocks@users.noreply.github.com>
Co-authored-by: GoldenAlpharex <jerego1234@hotmail.com>
Co-authored-by: GoldenAlpharex <58045821+GoldenAlpharex@users.noreply.github.com>
* This tail refactor turned into an organ refactor. Funny how that works.
* Firstly, fixing all the conflicts.
* Fixes all our maps (hopefully)
* Actually, this should fix pod people hair :)
* Almost everything is working, just two major things to fix
* Fixed a certain kind of external organ
* Cleaning up some more stuff
* Turned tail_cat into tail because why the fuck are they separate?
* Moved all the tails into tails.dmi because that was just dumb to have like 3 in a different file
* Adds relevant_layers to organs to help with rendering
* Makes stored_feature_id also check mutant_bodyparts
* Fixes the icon_state names of ALL the tails (pain)
* Fixes wagging, gotta refactor most mutant bodyparts later on
* I Love Added Failures
* Fixed some organs that slipped through my searches
* This could possibly fix the CI for this?
* It doesn't look like it did fix it
* This will make it pass, even if it's ugly as sin.
* Fixed Felinids having a weird ghost tail
* Fixes instances of snouts and tails not being properly colored
Co-authored-by: Kapu1178 <75460809+Kapu1178@users.noreply.github.com>
Co-authored-by: GoldenAlpharex <jerego1234@hotmail.com>
* xeno weed changes (#64477)
lowered the grow time limit from 15-20 seconds to 5-10 seconds
weeds can't grow in water now
documented the variables and procs
use the cooldown system
children weeds of the parent node will be destroyed on parent destroyed
* xeno weed changes
Co-authored-by: jjpark-kb <55967837+jjpark-kb@users.noreply.github.com>
* aliens can no longer clear eggs before the hugger bursts out (#62597)
* aliens can no longer clear eggs before the hugger bursts out
Co-authored-by: Fikou <23585223+Fikou@users.noreply.github.com>
* Makes canister leaking and blowing up use a component and element respectively (#59075)
I want to use this behavior on some other things so into a component and element it goes. Gas leaking is handled by a component so it can process whereas the object breaking and causing an explosion is handled by an element. Some minor changes were made so canisters were more consistent in leaking.
* Makes canister leaking and blowing up use a component and element respectively
Co-authored-by: Emmett Gaines <ninjanomnom@gmail.com>
* Converts the atmos senstive component over to connect_loc (#58266)
* Makes all uses of atmos_senstive pass in mapload as context
* Converts atmos senstive to connect_loc, does some general cleanup to the element, and makes it check the state of the tile the thing is on assuming creation didn't happen as a part of map loading
* Updates connect loc to match the new arg list
* Converts the atmos senstive component over to connect_loc
Co-authored-by: LemonInTheDark <58055496+LemonInTheDark@users.noreply.github.com>
* Adds a few walls and airlocks to the big derelict space ruin to keep atmos diffs down when entered. (#54749)
* Oh right the commit
* Removes snowflake pathing for alien resin window.
* Missed one.
* Actually this whole map is just a var-edit hell
* Adds a few walls and airlocks to the big derelict space ruin to keep atmos diffs down when entered.
Co-authored-by: ArcaneMusic <41715314+ArcaneMusic@users.noreply.github.com>
* Turns transparency for turfs into an element, and gives it to datum materials. (#54250)
* Turns transparency for turfs into an element, and gives it to datum materials.
Co-authored-by: Qustinnus <Floydje123@hotmail.com>
* Opacity refactor (#52881)
Moves all opacity var manipulation to a proc which sends a signal.
light_blocker element for movable opaque atoms made, which tracks its movement and updates the affected turfs for proper lighting updates.
has_opaque_atom boolean replaced by the opacity_sources lazylist to keep track of the sources, and a directional_opacity which serves a similar function but also allows for future expansion with on-border opaque objects (not yet implemented).
Some opacity-related sight procs optimized as a result of this.
Some variables moved to the object's definition.
A define or two added into the mix for clarity.
Some code cleaning, like turning booleans into their defines.
One file renamed for clarity.
Changelog
cl
balance: Mechs no longer block sight. It's a non-trivial cost for the lighting system with little to no gain.
/cl
* Opacity refactor
Co-authored-by: Rohesie <rohesie@gmail.com>
* Icon smooth refactor (#52864)
bitflag list construct added: an associative list of bitflags for cheap and quick element comparison between two lists using the same system.
canSmoothWith list turned into a bitflag list.
smoothing_groups list added to substitute the type path list.
smoothing procs turned into atom procs, refactored and optimized a bit.
smooth directions redefined in order to fit in 8 bits for a future smoothing system
some variable names changed, foreseeing a second smoothing system
SMOOTH_OBJ flag added, for things that need to scan turfs for smoothing. The old locate() optimization has the risk of returning false negatives by finding a child and returning null while there might be one of the wanted type as well, as it doesn't match the type exactly.
SMOOTH_TRUE and SMOOTH_MORE condensed into SMOOTH_CORNERS. The old behavior can be replicated using smoothing groups without loss.
Does very minor code cleanup.
Processing-wise didn't find a noticeable difference. The system loses on init a bit by setting the bitflag_lists, and by scanning whole turf contents for object smoothing (increasing accuracy), and gains by making less checks per target to smooth, through the same bitflag_lists.
Memory-wise there should be a small improvement, given that on the old system we had 63512 canSmoothWith lists (a few typelists, most unique), and on this new system canSmoothWith + smoothing_groups are both bitflag_lists from the same pool, totaling 46 in number.
Could be tested a bit to see if I missed any icons not properly smoothing.
* Icon smooth refactor
Co-authored-by: Rohesie <rohesie@gmail.com>
* Makes all CanPass procs call parent
* Makes CanPass more extendable and gives the mover a say in the matter
* Replace CanPass with CanAllowThrough to use the new system
Regex replace `(?<!proc)/CanPass\(` => `/CanAllowThrough(`
* Simple optimization pass
removes materials list from items, uses custom_materials instead. This might introduce some bugs so we should testmerge this for a while (and Ill test stuff locally as much as I can)
this also adds material crafting to sheets. Test case being chairs. In the future we can add stuff like tables, walls, doors etc.
also applies materials to everything, with fixes, which can close#46299
About The Pull Request
Converts every single usage of playsound's vary parameter to use the boolean define instead of 1 or 0. I'm tired of people copypasting the incorrect usage.
Also changes a couple of places where a list was picked from instead of using get_sfx internal calls
This was done via regex:
(playsound\(.+,.+,.+, ?)1( ?\)| ?,.+\)) to match 1
(playsound\(.+,.+,.+, ?)0( ?\)| ?,.+\)) to match 0
full sed commands:
/(playsound\(.+,.+,.+, ?)1( ?\)| ?,.+\))/\1TRUE\2/ 1 to TRUE
/(playsound\(.+,.+,.+, ?)0( ?\)| ?,.+\))/\1FALSE\2/ 0 to FALSE
I'm not very good with regex and these could probably be optimized, but they worked.
Why It's Good For The Game
Code usability
* New proximity monitor datum
* Replace the old systems
* Fixes#26254
* Ports a New and spawn to Initialize
* Switch to Moved
* Remove unused var
* Speedup and fixes
* More speed
* qdel in movable
* Uneeded now
* Fix
* Use the faster RANGE_TURFS
