Patch Manager | KSP2 Redux | Page 2

upbeat hare Jun 10, 2023, 5:06 PM

#

just "modify" the text asset and it gets cached in a zip file

north mist Jun 10, 2023, 5:06 PM

#

lmao

#

I mean, we're not doing anything wrong, the user has to own the files first to be able to access them that way, and they can do the same with Asset Studio or any other similar tool

upbeat hare Jun 10, 2023, 5:07 PM

#

True

upbeat hare Jun 10, 2023, 6:29 PM

#

Alright cool, I think I implemented a simple ISelectable structure for parts, now its time to implement the actual engine for testing purposes

#

@north mist I am going to move the interface I made to shared instead

#

I am getting a circular dependency error

north mist Jun 10, 2023, 6:31 PM

#

yeah that makes sense

upbeat hare Jun 10, 2023, 8:00 PM

#

Damn, I'm actually writing a whole lotta shit to make my system work

#

I mean I am basically running a programming language so

upbeat hare Jun 10, 2023, 8:19 PM

#

I wonder if I can speed this up w/ worker threads at some point, but my intuiton says no because they are all acting on the same object when they run and I can't really tell if they are acting on different objects

#

public override void ExecuteIn(Environment environment)
{
    foreach (var attribute in Attributes)
    {
        switch (attribute)
        {
            case RequireModAttribute requireModAttribute when
                !environment.GlobalEnvironment.Universe.AllMods.Contains(requireModAttribute.Guid):
            case RequireNotModAttribute requireNotModAttribute when
                environment.GlobalEnvironment.Universe.AllMods.Contains(requireNotModAttribute.Guid):
                return;
        }
    }
    var snapshot = environment.Snapshot();
    var patcher = new SassyTextPatcher(snapshot, this);
    environment.GlobalEnvironment.Universe.RegisterPatcher(patcher);
}

Register patcher is an action in the Universe state of my patching engine that can register a patcher into the execution engine, but what it does is set in the constructor of the universe

upbeat hare Jun 10, 2023, 9:45 PM

#

Oops I made a 500 line file containing interop conversions between Values and managed objects

#

Just so I can write builtin functions in an easier way

north mist Jun 10, 2023, 9:47 PM

#

damn

upbeat hare Jun 10, 2023, 10:30 PM

#

@north mist the next PR I will do (not doing it just yet) will again be insane (this is just one commit because I need to remember to commit more often)

#

Its a lot of the meat of the engine

north mist Jun 10, 2023, 10:35 PM

#

amazing stuff

upbeat hare Jun 10, 2023, 10:37 PM

#

It has a lot of stuff in it, like automatic registration of rulesets, and automatic registration of builtin libraries as well, plus method interop code (that I stole from another one of my projects because why reinvent the wheel)

#

My next step is to set up a local test system to test the patches

upbeat hare Jun 10, 2023, 11:00 PM

#

I has to go but I got it to register a patch from my standalone testing project!

#

Gods we are drawing ever closer to a working sysrem

#

When I'm free I'm actually gonna try and run a patch on multiple json files

north mist Jun 10, 2023, 11:14 PM

#

upbeat hare Gods we are drawing ever closer to a working sysrem

I'm off to bed, I was quite busy today, but I hope to be able to spend more time on this tomorrow

upbeat hare Jun 11, 2023, 12:11 AM

#

no fucking way, no fucking way, no fucking way @north mist I know you are sleeping but I ran my patch on a JSON file and it fucking worked!!!

#

Notice the different in the amount of hydrogen

north mist Jun 11, 2023, 12:12 AM

#

Am in bed but not asleep yet

#

Let's gooooo

#

That's amazing

upbeat hare Jun 11, 2023, 12:12 AM

#

Technically I ran it on all these json files :3

north mist Jun 11, 2023, 12:14 AM

#

In the morning I want to take a look at the actual caching instead of just storing flat JSON files, and into adding support for the specialized cases like the gimbal module

upbeat hare Jun 11, 2023, 12:25 AM

#

Hmm, I need to add a custom integer data type aaa

#

Gimme one minute

#

Cuz I think something like this may error if I don't

upbeat hare Jun 11, 2023, 12:43 AM

#

Integer support has been added, it wasn't as much work as I thought it would be

#

just means I have 3 more cases for every operation ran

upbeat hare Jun 11, 2023, 1:01 AM

#

Anyways if you want to work w/ this tomorrow munix
https://github.com/jan-bures/PatchManager/pull/7

GitHub

MVP For Sassy Patching execution engine by cheese3660 · Pull Reques...

This is an MVP for the sassy patch execution engine, specifically over parts

#

Sorry that its another >100 file change

upbeat hare Jun 11, 2023, 1:47 AM

#

Wait did I really just write an MVP for a DSL/programming language ... in less than a week? Wow I mustve been really at a lack for things to do

upbeat hare Jun 11, 2023, 11:03 AM

#

insane in the best possible way

north mist Jun 11, 2023, 11:04 AM

#

munleyix

upbeat hare Jun 11, 2023, 11:06 AM

#

I should write an architecture breakdown of the engine

north mist Jun 11, 2023, 11:06 AM

#

That would be very helpful

upbeat hare Jun 11, 2023, 11:08 AM

#

Aight time to pull out google slides

#

Me realizing I made a slight mistake that doesnt require too much to change

upbeat hare Jun 11, 2023, 12:22 PM

#

Writing a general purpose patching language ^

#

So a few notes
A ruleset is technically what transforms the textual data into the ISelectable, which ISelectables are the backbone of the engine
There is a general purpose JSON selectable called JTokenSelectable, and a base selectable called BaseSelectable, but rolling your own is not too hard
Then from ISelectables come IModifiables, which are returned from calling OpenModification on a selectable, if it is modifiable otherwise, null, and those modifiables are where the true transformations take place, and there are also some useful ones like JTokenModifiable and CustomJTokenModifiable which is where you can create custom "adapters" in the modifiables fields to more closely match the selectable you are modifying

#

The overview of the engine itself is simple enough
It starts by ingesting all files and storing those starting with _ as libraries then after its ingested all patches from every mod it runs each file that is not a library, which then registers any selection block in there as an ITextPatcher via a function passed to the universe state of the execution engine

#

Then when each of those text patchers is run, it takes the data and transforms it according to the leftmost ruleset on the top block and then runs all the selections henceforfh

#

Then the modifications read and write from the IModifiables

north mist Jun 11, 2023, 12:38 PM

#

upbeat hare So a few notes A ruleset is technically what transforms the textual data into th...

#

pinning just in case

#

but it sounds straightforward enough

upbeat hare Jun 11, 2023, 12:44 PM

#

Anyways the next parts are yours while I start expanding the library of the engine and maybe actually add loops into functions

#

I don't know if I want to enable someone to be able to say do
@while true {} but meh there are other ways to do that in any case

#

Looks like my builtin function code was slightly borked but I fixed it!

#

@use 'builtin:debug';

:parts {
    $ignored: debug-log($current["partName"]);
}

So now a patch program like this works

north mist Jun 11, 2023, 1:00 PM

#

now we need a custom VS Code language server extension for this munley

#

for full intellisense

upbeat hare Jun 11, 2023, 1:03 PM

#

This is the debug-log function btw

/// <summary>
/// Logs a value into the console for debugging
/// </summary>
/// <param name="universe">The universe in which this function is being called</param>
/// <param name="v">The value to log</param>
[SassyMethod("debug-log")]
public static void Log(Universe universe, DataValue v)
{
    universe.MessageLogger(v.ToString());
}

#

It auto fills in the universe parameter when it encounters it

upbeat hare Jun 11, 2023, 1:27 PM

#

Well shit it looks like I'm adding closures somewhat lol

#

Why, I'm not sure

#

though it does mean you can return actions/funcs from builtin functions if you really wish to

#

I don't think I can really do the opposite way though

#

(That is have a function w/ an argument of type Action/Func<...>, unless I want to do dynamic methods)

north mist Jun 11, 2023, 1:48 PM

#

I don't think we need that lmao

upbeat hare Jun 11, 2023, 1:49 PM

#

Ehh ... scss has higher order functions which is what this enables

north mist Jun 11, 2023, 1:51 PM

#

upbeat hare (That is have a function w/ an argument of type Action/Func<...>, unless I want ...

I mean if you wanna do it, sure

upbeat hare Jun 11, 2023, 1:51 PM

#

north mist I mean if you wanna do it, sure

oh you were talking about that part

north mist Jun 11, 2023, 1:51 PM

#

yeah

upbeat hare Jun 11, 2023, 1:52 PM

#

private static int InvokeComparison(Environment env, PatchFunction comparator, DataValue a, DataValue b)
{
    var result = comparator.Execute(env, new List<PatchArgument>
    {
        new PatchArgument {
            ArgumentDataValue = a
        },
        new PatchArgument {
            ArgumentDataValue = b
        },
        
    });
    if (result.IsInteger)
    {
        // A simple are these less 
        return (int)result.Integer;
    }

    if (result.IsReal)
    {
        return (int)result.Real;
    }
    throw new TypeConversionException(result.Type.ToString().ToLowerInvariant(), "integer");
}

/// <summary>
/// Sorts a list, using a comparison function if one is provided otherwise uses a default comparison algorithm
/// </summary>
/// <param name="env"></param>
/// <param name="dataValues"></param>
/// <param name="comparator">The function to sort with, default null (due to closures not existing in the program yet), follows the C# comparison function</param>
/// <returns></returns>
[SassyMethod("list.sort")]
public static List<DataValue> Sort(Environment env, List<DataValue> dataValues, PatchFunction comparator = null)
{
    var copy = new List<DataValue>(dataValues);
    Comparison<DataValue> comparison =
        comparator != null ? (x, y) => InvokeComparison(env, comparator, x, y) : DefaultComparison;
    copy.Sort(comparison);
    return copy;
}

The only place I see it possibly necessary is somewhere like here, but meh

north mist Jun 11, 2023, 1:55 PM

#

I mean can't we just do something similar to meta.get-function() and meta.call()?

upbeat hare Jun 11, 2023, 1:55 PM

#

I am basically doing that yes, but this is more so on the interop side of things I'm talking about

north mist Jun 11, 2023, 1:56 PM

#

gotcha

upbeat hare Jun 11, 2023, 2:05 PM

#

For example

/// <summary>
/// Gets a function from the global environment
/// </summary>
/// <param name="globalEnvironment">The global environment that is automatically filled in</param>
/// <param name="name">The name of the function to get</param>
/// <returns>A closure representing that function</returns>
[SassyMethod("get-function")]
public static PatchFunction GetFunction(GlobalEnvironment globalEnvironment, string name)
{
    return globalEnvironment.AllFunctions[name];
}

/// <summary>
/// Invoke a closure w/ the given arguments
/// </summary>
/// <param name="env">The environment to run the closure in</param>
/// <param name="closure"></param>
/// <param name="arguments"></param>
/// <returns></returns>
[SassyMethod("closure.invoke")]
public static DataValue Invoke(Environment env, PatchFunction closure, List<PatchArgument> arguments)
{
    return closure.Execute(env, arguments);
}

#

I'm not sure how much people are going to use these features, but they are there if someone perchance does want to use them

#

@use 'builtin:debug';
@use 'builtin:functional';
$test-closure: @function($v) {
    $ignored: debug-log($v);
};

:parts {
    $ignored: $test-closure:invoke($current["partName"]);
}

Somehow this works

#

But yeah this is mostly me being slightly bored more than anything

#

Or even this

@use 'builtin:debug';
@use 'builtin:functional';

:parts {
    $x: @function() {
        $y: debug-log($current["partName"]);
    }:invoke();
}

Which I can actually see some use for, if you want to do a long calculation on one value but not mess up some scope, pollute the global scope, or just anything like that

north mist Jun 11, 2023, 2:22 PM

#

that's pretty cool

upbeat hare Jun 11, 2023, 2:31 PM

#

It also allows for an easy map function that can be used like this

$list:map(@function($x) { @return x*2; });

upbeat hare Jun 11, 2023, 2:34 PM

#

north mist that's pretty cool

Thanks!

#

Shit I just made a way to make lists as closures

@use 'builtin:functional';
@use 'builtin:list';
$list-function: get-function("list.create");
$list-123: $list-function:bind(1,2,3);
$list-1234: $list-123:bind(4):invoke(); // Returns [1,2,3,4]
$list-12345: $list-123:bind(4):invoke(5); // Returns [1,2,3,4,5]

The binding is in lieu of the varargs that scss has that I haven't added yet

north mist Jun 11, 2023, 2:49 PM

#

functional sassy patches programming, here we go

#

munley

upbeat hare Jun 11, 2023, 2:50 PM

#

That's the point

north mist Jun 11, 2023, 2:50 PM

#

we should just allow Sassy Patches as a language for standalone SpaceWarp mods

#

lmao

upbeat hare Jun 11, 2023, 2:54 PM

#

Anyways time to add while loops and for loops

#

Though technically I don't need them, its kinda such that I don't want to blow the stack w/ recursion

#

Technically different than sass but I dislike their syntax anyways

@for $list : $val {
  //...
}

@for $list : $idx, $val {
  //...
}

@for $dict : $key, $val {
  //...
}

@for 0,10 : $val {
  //...
}

@while $a > $b {
  //...
}

north mist Jun 11, 2023, 3:04 PM

#

idk, I think the @each $val in $list syntax might be more readable for non-programmers

#

but that's just my opinion

upbeat hare Jun 11, 2023, 3:05 PM

#

It might be, the issue is I want to reserve as little keywords as possible to stop people from having to use strings instead of literals

#

Also this kinda stuff hopefully goes in libraries that non programmers use instead of non programmers having to interact w/ it directly

north mist Jun 11, 2023, 3:06 PM

#

i guess

upbeat hare Jun 11, 2023, 3:06 PM

#

I can understand your disagreement

#

I can reserve @each and do @for x through/to y, it just means that in and through/to have to be reserved

#

Which in the context of KSP2 shouldn't pose too much a problem

north mist Jun 11, 2023, 3:09 PM

#

What I mean is just that I personally envisioned us sticking as closely to SCSS as possible, making more so a superset of it, so that we can mostly stick to their docs and stuff like that, instead of having a language that's familiar yet with some significant syntax differences

#

Though I understand your reasons for not doing that

upbeat hare Jun 11, 2023, 3:12 PM

#

No I get that, ehh, I can do each/for
A minor different in control flow structures though

@else if
// is now
@else-if

north mist Jun 11, 2023, 3:14 PM

#

eh that seems fine

upbeat hare Jun 11, 2023, 3:18 PM

#

Do I add interpolation to selection blocks, w/ the caveat that it makes incremental patching nigh on impossible in the future

north mist Jun 11, 2023, 3:19 PM

#

hm

#

that's a tough one

upbeat hare Jun 11, 2023, 3:35 PM

#

@function test() {
    @each $k,$v in $y {
    }
}

Coolio it'll be something like this for now

#

@function test() {
    @for $x from $y to $z {
    }
}

#

Aaa for loops are "fun"

upbeat hare Jun 11, 2023, 4:10 PM

#

Why the fuck do the part jsons use smart quotes

#

"An “Evolution” in wing design, the HPW-1000 is everything you can ask for in a heavy wing. It can be built to accommodate large cargo craft, passenger jets, and even massive SSTO’s. Bring us your blueprint, schematic, or napkin sketch, and we’ll make it a rea
lity."

Like don't tell me you wouldn't interpret that as a syntax error by just looking at it

#

Anyways, I can't do any real tests until we actually have a way to register and run patchers in ksp2

upbeat hare Jun 12, 2023, 12:51 AM

#

Well imma sleep

upbeat hare Jun 12, 2023, 2:21 PM

#

Munix, any update on your side of things so I can look at integration?

north mist Jun 12, 2023, 2:22 PM

#

I was really busy over the weekend with family stuff, but I'll finish some work stuff now and I'll try to get it up and running asap

#

I (think) I have a good idea of what I want to do

upbeat hare Jun 12, 2023, 4:33 PM

#

Gotcha

north mist Jun 12, 2023, 9:48 PM

#

so I've got most of it in place but my biggest issue right now is that I have no clue how we're going to be assigning all the patch files to their corresponding addressable labels

#

or is the plan just to try to run every patch file on every JSON file?

#

I guess that makes more sense, since technically one patch file could try to patch multiple completely unrelated areas of the game

upbeat hare Jun 12, 2023, 9:57 PM

#

north mist so I've got most of it in place but my biggest issue right now is that I have no...

This is why the ITextPatcher takes the label as well, whatever implements, can match the label (which my system does)

#

But yes we run every patch on every JSON file

#

but only some will fully execute, others will quit early if the label does not match the ruleset

upbeat hare Jun 12, 2023, 10:29 PM

#

north mist I guess that makes more sense, since technically one patch file could try to pat...

And also yes, this

#

But patch files actually all run at the start and register the patch functions which are the ITextPatchers

north mist Jun 13, 2023, 3:13 AM

#

I pushed a hopefully-nearly-finished version

#

there's one issue I know of so far, the zip files whose data is stored in memory for some reason refuse to get saved, or rather, their files stay at 0 B

#

the issue is somewhere in the implementation of the PatchManager.Core.Cache.Archive class

north mist Jun 13, 2023, 11:35 AM

#

alright so that issue is solved and cache is now getting saved fine

#

the issue I have now is that all attempts at running any patches scream Ruleset: parts does not exist!

upbeat hare Jun 13, 2023, 11:36 AM

#

Did you load the parts assembly

north mist Jun 13, 2023, 11:36 AM

#

aaaand I think I know why lmao

#

yeah exactly, I didn't

#

only took me typing it to realize

upbeat hare Jun 13, 2023, 11:37 AM

#

I ran into that same issue w/ my manual testing

north mist Jun 13, 2023, 12:04 PM

#

jesus fucking christ

#

I've been trying to debug for like an hour why my cache keeps getting invalidated/deleted

#

I completely forgot about this piece of shit in the PatchManagerPlugin

#

nooooooooo

upbeat hare Jun 13, 2023, 12:08 PM

#

F

#

cache invalidation is a fun problem

north mist Jun 13, 2023, 12:22 PM

#

I guess this means we're getting somewhere?

#

[Error  : Unity Log] [Debug] (GetAttributeRuleSet) failed to find attribute set called KERBALATTRIBUTES_EVA
[Error  : Unity Log] [General] Object reference not set to an instance of an object

  at KSP.Game.KerbalVarietySystem.GenerateCustomKerbalAttributesFromRawData (System.Collections.Generic.Dictionary`2[TKey,TValue] rawCustomKerbalData, KSP.Game.KerbalAttributes& kerbalAttributes, System.String attributeSetName) [0x00044] in <ef61a348eb874c99b6bdcbcf875cc384>:0
  at KSP.Game.KerbalVarietySystem.CreateCustomKerbalByName (System.String customKerbalName, System.String attributeSetName) [0x00026] in <ef61a348eb874c99b6bdcbcf875cc384>:0
  at KSP.Game.KerbalVarietySystem.TryCreateCustomKerbalByName (System.String customKerbalName, KSP.Game.KerbalAttributes& kerbalAttributes, System.String attributeSetName) [0x00000] in <ef61a348eb874c99b6bdcbcf875cc384>:0
  at KSP.Game.KerbalRosterManager.GenerateCustomKerbalLookups () [0x0005d] in <ef61a348eb874c99b6bdcbcf875cc384>:0
  at KSP.Game.KerbalRosterManager.OnUpdate (System.Single deltaTime) [0x00038] in <ef61a348eb874c99b6bdcbcf875cc384>:0
  at KSP.Game.GameInstance.Update () [0x000bf] in <ef61a348eb874c99b6bdcbcf875cc384>:0

north mist Jun 13, 2023, 2:18 PM

#

well that's unfortunate, I kinda didn't realize until now that by Cecil-patching the game assembly, we lose the ability to debug it with dnSpy

#

and another related thing, though it's obvious, but I thought I'd mention it anyway - disabling a mod in the mod list can't do anything about disabling the mod's preload patchers

#

so when you attempt to disable Patch Manager, you "brick" your game, because the patcher will run, and it indirectly depends on the main plugin assembly having run

upbeat hare Jun 13, 2023, 2:33 PM

#

We can attempt to check the disabled plugins file using the patch manager patcher

north mist Jun 13, 2023, 2:36 PM

#

Yeah, we really should

upbeat hare Jun 13, 2023, 2:36 PM

#

Just check if patchmanagers guid is in the list and return early if it is

upbeat hare Jun 13, 2023, 3:22 PM

#

@north mist do you have the latest changes you made pushed?

north mist Jun 13, 2023, 3:23 PM

#

Nope lemme quickly do that

#

pushed

upbeat hare Jun 13, 2023, 3:35 PM

#

north mist pushed

Cache branch?

north mist Jun 13, 2023, 3:35 PM

#

yep

#

so the two main current issues are that while the cache does get saved fine on the first run, but it doesn't seem to get used, and then on the next start it does get used, but we get that error I posted above

#

I'm assuming the first thing has something to do with rewinding streams and whatnot

#

but the second one is a bit more difficult to debug, as far as I can tell, the files we provide to the game that are related to the error are the exact same as the original ones

upbeat hare Jun 13, 2023, 3:38 PM

#

This is an issue?

#

north mist Jun 13, 2023, 3:39 PM

#

try deleting your build folder and rebuilding

#

doesn't happen to me

upbeat hare Jun 13, 2023, 3:40 PM

#

Fixed

#

why do I keep having these build folder issues aaaaa

north mist Jun 13, 2023, 3:41 PM

#

eh it's probably because I changed some stuff in the project files

upbeat hare Jun 13, 2023, 3:42 PM

#

Version check working lol
But yeah imma test some stuff

#

Damn thats a lot of labels

#

north mist Jun 13, 2023, 3:45 PM

#

I mean for now if we know some specific ones are causing issues, we can just add them to a blacklist or something

#

specifically the kerbal attributes/variety stuff

upbeat hare Jun 13, 2023, 3:46 PM

#

How early do you get these errors

north mist Jun 13, 2023, 3:46 PM

#

when loading a campaign (after launching the game with the cache already existing)

upbeat hare Jun 13, 2023, 3:47 PM

#

Interesting error, but thats expected (I really want copy clicking errors)

north mist Jun 13, 2023, 3:47 PM

#

oh yeah, we need to just filter out stuff that isn't JSON

upbeat hare Jun 13, 2023, 3:49 PM

#

I think the issue is that its calling this function

        public void LoadResourceLocations<T>(object key, Action<IList<IResourceLocation>> resultCallback) where T : UnityEngine.Object
        {
            if (AssetProvider.IsComponent(typeof(T)))
            {
                Debug.LogError("AssetProvider cannot load components/monobehaviours in batch.");
                return;
            }
            Addressables.LoadResourceLocationsAsync(key, typeof(T)).Completed += delegate(AsyncOperationHandle<IList<IResourceLocation>> results)
            {
                if (results.Status != AsyncOperationStatus.Succeeded)
                {
                    Action<IList<IResourceLocation>> resultCallback2 = resultCallback;
                    if (resultCallback2 != null)
                    {
                        resultCallback2(null);
                    }
                    Addressables.Release<IList<IResourceLocation>>(results);
                    return;
                }
                Action<IList<IResourceLocation>> resultCallback3 = resultCallback;
                if (resultCallback3 == null)
                {
                    return;
                }
                resultCallback3(results.Result);
            };
        }

Before LoadAssetByLabel?

#

Meaning we might want to hijack that function as well

#

That may be why it isn't using the cache on first run

#

if (typeof(T) == typeof(TextAsset))
{
    if (!CacheManager.CacheValidLabels.Contains(label))
    {
        PatchingManager.RebuildCache(label);
    }

    var found = Locators.LocateAll(label, typeof(T), out var locations);

    if (found)
    {
        Addressables.LoadAssetsAsync(locations, assetLoadCallback).Completed += onCompletedCallback;
        return;
    }
}

Also I'm gonna make sure we only replace text assets

north mist Jun 13, 2023, 3:53 PM

#

yeah, sure

upbeat hare Jun 13, 2023, 3:54 PM

#

Also this won't backup

var backup = text;
try
{
    var wasPatched = patcher.TryPatch(label, ref text);
    if (wasPatched)
    {
        patchCount++;
    }
}
catch (Exception e)
{
    Console.WriteLine($"Patch errored due to: {e.Message}");
    text = backup;
}

You need to do new string(text)

north mist Jun 13, 2023, 3:54 PM

#

nope, I tested it

#

it will

upbeat hare Jun 13, 2023, 3:54 PM

#

Actually no I might just be being dumb

#

strings are immutable

north mist Jun 13, 2023, 3:54 PM

#

yeah

upbeat hare Jun 13, 2023, 3:56 PM

#

I'm still wondering how we break the kerbal loading

#

And we can't debug at all via dnspy?

#

(I didn't even know that was possible anyways)

north mist Jun 13, 2023, 3:57 PM

#

doesn't seem to be working, when the patcher file is there, it tells me that the assembly is not loaded, when I remove it, I can do the debugging just fine

upbeat hare Jun 13, 2023, 3:58 PM

#

You can manually edit the assembly?

north mist Jun 13, 2023, 3:58 PM

#

wdym?

upbeat hare Jun 13, 2023, 3:59 PM

#

Like, load up the patchmanager.core assembly and the KSP2 assembly into dnspy, and edit the ksp2 function w/ a call to the patchmanager.core one and just not use the preload patcher

north mist Jun 13, 2023, 4:00 PM

#

I mean technically for debugging purposes we could just save the result of the Cecil patching into a DLL, replace the game's assembly with that temporarily and debug with that

upbeat hare Jun 13, 2023, 4:00 PM

#

Thats literally what I was suggesting but done manually

north mist Jun 13, 2023, 4:00 PM

#

yeah

#

this would be a bit easier I was assuming

#

since we're already doing it

#

but just in memory

upbeat hare Jun 13, 2023, 4:01 PM

#

I guess ... editing assemblies in dnSpy is very easy though

north mist Jun 13, 2023, 4:01 PM

#

I wouldn't know, never tried 😆

upbeat hare Jun 13, 2023, 4:02 PM

#

Literallly just press edit method C#

north mist Jun 13, 2023, 4:02 PM

#

heh

upbeat hare Jun 13, 2023, 4:02 PM

#

And you get a code editor

north mist Jun 13, 2023, 4:02 PM

#

cool

upbeat hare Jun 13, 2023, 4:07 PM

#

Only problem w/ manually editing it is that the patcher is internal which I can change lol

#

And boom, simple as that!

#

Literally the exact same code I emitted using mono.cecil

#

And good thing is, even w/ the modified assembly, if we forget to remove the preload patcher it should just run as normal

#

@north mist do you need unity to be set up in debug mode?

north mist Jun 13, 2023, 4:13 PM

#

I'm pretty sure you do

upbeat hare Jun 13, 2023, 4:17 PM

#

Done

#

Lol

#

as if

#

@north mist I found the problem, something is causing the keey in here to be an empty string

north mist Jun 13, 2023, 4:22 PM

#

huh

upbeat hare Jun 13, 2023, 4:23 PM

#

Thats meant to be "KERBALATTRIBUTES_EVA"

#

So something w/ those

north mist Jun 13, 2023, 4:26 PM

#

OH

upbeat hare Jun 13, 2023, 4:26 PM

#

OHHH

north mist Jun 13, 2023, 4:26 PM

#

lmao

upbeat hare Jun 13, 2023, 4:27 PM

#

So somehow the name isn't being passed through to the delegate?

north mist Jun 13, 2023, 4:27 PM

#

#

name corresponds to PrimaryKey

#

so it should start with KerbalAttributes_

#

even if it ends with .zip

upbeat hare Jun 13, 2023, 4:28 PM

#

No it does

#

otherwise it wouldn't be loaded at all

#

WAIT, I think it fails here

#

@north mist do we store the Text Assets name at all?

#

We have it cached

#

But thats what gets assigned to the key in the dictionary

north mist Jun 13, 2023, 4:30 PM

#

I don't even know how I would provide a name

#

it only takes one parameter and that's the content of the file

#

I mean

#

I can just assign it, for some reason I thought it was readonly

#

but it isn't

upbeat hare Jun 13, 2023, 4:31 PM

#

Yeah, then just do that lol

#

I mean you have the asset name here lol

#

asset.name

#

Seems not impossible

north mist Jun 13, 2023, 4:34 PM

#

yeah I mean I just did

var asset = new TextAsset(archive.ReadFile(file))
{
    name = Path.GetFileNameWithoutExtension(file)
};

#

probably without the "WithoutExtension" part

#

since I don't give them any extensions anyway

upbeat hare Jun 13, 2023, 4:35 PM

#

And there, easy enough
(Even if it was readonly ... reflection)

#

Test it?

north mist Jun 13, 2023, 4:40 PM

#

#

https://tenor.com/view/working-anakin-starwars-gif-4653092

Tenor

upbeat hare Jun 13, 2023, 4:44 PM

#

Aight, push it!

north mist Jun 13, 2023, 4:44 PM

#

just wanna quickly solve the issue with the cache not being used when its first created/invalidated

#

or not, I can just push that later

#

alright, it's there

#

oh and I think I know what the issue is

upbeat hare Jun 13, 2023, 4:48 PM

#

oh?

north mist Jun 13, 2023, 4:48 PM

#

the rebuilding of the cache is async

#

so we try to access it before the zip file has a chance to be saved

#

before this handler has a chance to run

#

this gets called

#

I'm pretty sure

#

and since it can't find anything because there's no entry for the label, it just defaults to the game's assets

upbeat hare Jun 13, 2023, 4:50 PM

#

Anyway to make it not async?

north mist Jun 13, 2023, 4:50 PM

#

I can just add a callback argument to RebuildCache which will contain the call, instead of putting it below

upbeat hare Jun 13, 2023, 4:53 PM

#

Gotcha

#

Anyways I'm gonna test a quick patch

#

(That being my crew capacity patch

north mist Jun 13, 2023, 4:56 PM

#

yep, that's the one that I was running (though I probably tweaked the number just to make the effect more obvious, I think I made it +10)

upbeat hare Jun 13, 2023, 4:56 PM

#

Its the most obvious one

#

I hate the obnoxious debug console from ksp

north mist Jun 13, 2023, 4:57 PM

#

yeah lmao

upbeat hare Jun 13, 2023, 4:58 PM

#

They should just steal space warps

north mist Jun 13, 2023, 4:59 PM

#

though I just discovered today (after having it there for over three months) that you can click to expand/collapse the text of error messages

#

which is pretty nice

upbeat hare Jun 13, 2023, 4:59 PM

#

#

I think I just need to add a generic json patcher and we got something worthy of a testing release

north mist Jun 13, 2023, 5:01 PM

#

yeah, definitely

#

I'm really grateful for your help with the whole DSL side of things, that's really not something I have a lot of experience with

upbeat hare Jun 13, 2023, 5:04 PM

#

north mist I'm really grateful for your help with the whole DSL side of things, that's real...

Surprisingly out of all the things we've done ... its the one I have the most experience w/

north mist Jun 13, 2023, 5:04 PM

#

I can tell lmao

#

you didn't cease to amaze me with the speed with which you were writing this stuff

upbeat hare Jun 13, 2023, 5:05 PM

#

Lol, fair

#

{
    // STUBBED UNTIL DATA IS AVAILABLE
    "attributeName":"FACEPAINT",
    "dependsOn":["BODY"],
    "attributeRangeRuleKey":"STUB_FACEPAINT_OPTIONS"
    //applyFunction":"ApplyFacePaint", --$$
},

THEY USE COMMENTS IN THEIR FREAKING JSON FILES AAAAAAA

north mist Jun 13, 2023, 5:06 PM

#

munley

#

apparently Newtonsoft.Json supports both single- and multi-line comments

#

so I could even use them in the swinfo.json like I wanted

upbeat hare Jun 13, 2023, 5:07 PM

#

Why don't we?

north mist Jun 13, 2023, 5:08 PM

#

I don't know, I just never looked up if it's implemented in the Newtonsoft library I guess lmao

upbeat hare Jun 13, 2023, 5:08 PM

#

I realized it was when implementing the generic json patching stuff I already have

#

I'm just gonna quickly add a way to pass filename to the patchers

#

And thats used as the #name in generic patchers

copper forum Jun 13, 2023, 5:13 PM

#

upbeat hare ```json { // STUBBED UNTIL DATA IS AVAILABLE "attributeName":"FACEPAINT"...

they are using jsonc then.

upbeat hare Jun 13, 2023, 5:14 PM

#

that much is obvious

north mist Jun 13, 2023, 5:16 PM

#

alright, loading of patched assets when cache is invalidated is fixed

upbeat hare Jun 13, 2023, 5:16 PM

#

Cool, I'm gonna have to pull that into my branch'

north mist Jun 13, 2023, 5:17 PM

#

pushed

#

well, I don't think that this is half bad considering I made the repo exactly a week ago

#

and I barely had anything at that point

upbeat hare Jun 13, 2023, 5:21 PM

#

my side is pushed, now time to try a different type of patch though

#

I ... had 50 gb in my recycle bin wtf

upbeat hare Jun 13, 2023, 5:26 PM

#

upbeat hare my side is pushed, now time to try a different type of patch though

gonna do the quarter size solar system

north mist Jun 13, 2023, 5:28 PM

#

now that's a good test

upbeat hare Jun 13, 2023, 5:35 PM

#

north mist now that's a good test

Theres an issue w/ it

#

There is zero ifnformation about orbital parameters in it, and also there are prefabs that need to be scaled I think...

#

Which ... yeah is kinda hard

north mist Jun 13, 2023, 5:36 PM

#

where are you looking for the orbital parameters?

upbeat hare Jun 13, 2023, 5:36 PM

#

in the bodies json

north mist Jun 13, 2023, 5:36 PM

#

you'll probably have to also patch the galaxy definition

#

that's what we were modifying with Hyperion and I'm pretty sure just that allowed us to rescale the planets

upbeat hare Jun 13, 2023, 5:37 PM

#

north mist you'll probably have to also patch the galaxy definition

Where is that?

north mist Jun 13, 2023, 5:38 PM

#

the key is GalaxyDefinition_Default

upbeat hare Jun 13, 2023, 5:38 PM

#

Which doesn't show up in the cache

north mist Jun 13, 2023, 5:38 PM

#

yeah, looks like it doesn't use LoadByLabel

#

GameManager.Instance.Game.Assets.Load<TextAsset>(this._data.SavedGame.GalaxyDefinitionKey, new Action<TextAsset>(this.OnGalaxyDefinitionLoaded), true);

#

we might have to patch all the Load... methods in the AssetProvider

upbeat hare Jun 13, 2023, 5:39 PM

#

Aight, thats gonna be a pain, but yeah I suppose we should

#

I'm just gonna do a quick gravity patch then, see if that works

upbeat hare Jun 13, 2023, 5:40 PM

#

north mist we might have to patch all the `Load...` methods in the AssetProvider

You think you can do that? I can write the cecil patches if necessary

north mist Jun 13, 2023, 5:41 PM

#

sure, I'll look into it

upbeat hare Jun 13, 2023, 5:47 PM

#

I think i did affect the gravity its hard to show a direct showing of

#

[Info   :Patch Manager] Patched AsteroidD01 with 1 patches. Total: 1
[Info   :Patch Manager] Patched Minmus with 1 patches. Total: 2
[Info   :Patch Manager] Patched AsteroidA01 with 1 patches. Total: 3
[Info   :Patch Manager] Patched AsteroidB01 with 1 patches. Total: 4
[Info   :Patch Manager] Patched Mun with 1 patches. Total: 5
[Info   :Patch Manager] Patched Eve with 1 patches. Total: 6
[Info   :Patch Manager] Patched Dres with 1 patches. Total: 7
[Info   :Patch Manager] Patched Ike with 1 patches. Total: 8
[Info   :Patch Manager] Patched Kerbol with 1 patches. Total: 9
[Info   :Patch Manager] Patched Bop with 1 patches. Total: 10
[Info   :Patch Manager] Patched AsteroidC01 with 1 patches. Total: 11
[Info   :Patch Manager] Patched Eeloo with 1 patches. Total: 12
[Info   :Patch Manager] Patched AsteroidE01 with 1 patches. Total: 13
[Info   :Patch Manager] Patched Jool with 1 patches. Total: 14
[Info   :Patch Manager] Patched Gilly with 1 patches. Total: 15
[Info   :Patch Manager] Patched Pol with 1 patches. Total: 16
[Info   :Patch Manager] Patched Kerbin with 1 patches. Total: 17
[Info   :Patch Manager] Patched Duna with 1 patches. Total: 18
[Info   :Patch Manager] Patched Vall with 1 patches. Total: 19
[Info   :Patch Manager] Patched Laythe with 1 patches. Total: 20
[Info   :Patch Manager] Patched Tylo with 1 patches. Total: 21
[Info   :Patch Manager] Patched Moho with 1 patches. Total: 22

Yep

#

@north mist
https://github.com/jan-bures/PatchManager/pull/8

GitHub

Sassy Patching w/ the Generic JSON patching as well, and using the ...

Tested working in game w/ increasing the crew count of crewed parts.

#

I think we did good

upbeat hare Jun 13, 2023, 6:22 PM

#

Anyways @north mist I think I am done for the next hour or 2

north mist Jun 13, 2023, 6:23 PM

#

yeah same, I'm gonna go watch a show with my brother

north mist Jun 14, 2023, 5:18 AM

#

one thing I just realized we need to take into account is that when a mod is added/removed/updated/disabled/enabled, we should also rebuild the cache, since the available parts/other things might have changed

upbeat hare Jun 14, 2023, 10:33 AM

#

north mist one thing I just realized we need to take into account is that when a mod is add...

Space Warp 1.3.0 allows you to easily test this

north mist Jun 14, 2023, 10:45 AM

#

Awesome

upbeat hare Jun 14, 2023, 3:52 PM

#

Any work on patching the other load functions?

#

Cuz I think thats all we need more

north mist Jun 14, 2023, 4:06 PM

#

I haven't really done anything today to be honest, I plopped myself out in the sun in the morning and am still here at 6pm lmao

#

I'll get on it now though

north mist Jun 16, 2023, 10:48 PM

#

alright so I'm having a bit of an issue with this

#

the method for rebuilding the cache is "async" in the sense that it needs to call Addressables.LoadAssetsAsync<TextAsset>(...).Completed += ... and the Completed handler is where the cache archive gets saved

#

however, I need to be able to patch either AsyncOperationHandle<TAsset> LoadAssetAsync<TAsset>(object key) or bool LocateAssetInExternalData(object key, System.Type T, out IResourceLocation location)

#

and those both have a return value and/or out parameter that I'm not able to assign while the methods are running because the archive is not saved yet at that point

#

I tried to just .WaitForCompletion() on the Addressables.LoadAssetsAsync() method call but it threw an error saying something about not being able to reenter an Update method, which "might be caused by calling WaitForCompletion"

#

so that's apparently not an option

#

I can't really see how to do this unless we move the cache rebuilding back out of the AssetProvider methods into a flow action at the start of the game load

north mist Jun 17, 2023, 12:21 PM

#

this means we need to know upfront what labels need to be patched before they're actually being loaded by the AssetProvider

#

and I don't think that's currently possible?

upbeat hare Jun 17, 2023, 2:09 PM

#

Why do you need to call the async functions

#

And why cant you just await them

north mist Jun 17, 2023, 3:05 PM

#

they're asynchronous in the sense that they return an AsyncOperationHandle

#

we need to call them because we need to load the addresable assets if we want to patch them

#

and while there is an API to await the handles (I think), you can't do that in a synchronous method

#

and we're patching synchronous methods (the ones in AssetProvider)

upbeat hare Jun 17, 2023, 3:07 PM

#

north mist and while there is an API to await the handles (I think), you can't do that in a...

I mean but you should be able to

north mist Jun 17, 2023, 3:07 PM

#

how?

#

also, Unity has a synchronous addressables workflow: https://docs.unity3d.com/Packages/[email protected]/manual/SynchronousAddressables.html

#

but like I said, when using it, the game hangs and you get an error about not being able to use WaitForCompletion from inside an Update method

#

which is because AssetProvider.LoadByLabel gets called from an Update method somewhere, directly or indirectly

upbeat hare Jun 17, 2023, 3:09 PM

#

you can do asynccontext.runtask the inner task?

north mist Jun 17, 2023, 3:16 PM

#

I can't find any class like that in .NET, is that a library?

#

for the record I also tried just Addressables.LoadAssetsAsync(...).Task.Result, which again results in a deadlock

upbeat hare Jun 17, 2023, 3:26 PM

#

north mist I can't find any class like that in .NET, is that a library?

Apparently, it wasnt entirely made clear on first look

upbeat hare Jun 17, 2023, 3:26 PM

#

north mist for the record I also tried just `Addressables.LoadAssetsAsync(...).Task.Result`...

Ah...

north mist Jun 17, 2023, 3:27 PM

#

I just get stuck on this

#

when debugging the code, it gets to the line with .Result and hangs there

upbeat hare Jun 17, 2023, 3:28 PM

#

Hmm, im not sure, are we sure we can't just call the synchronous versions of the methods, what do those return?

north mist Jun 17, 2023, 3:28 PM

#

there aren't any

north mist Jun 17, 2023, 3:29 PM

#

north mist also, Unity has a synchronous addressables workflow: https://docs.unity3d.com/Pa...

this is the only way

#

and that throws an error

upbeat hare Jun 17, 2023, 3:31 PM

#

hmm .. idk then

upbeat hare Jun 17, 2023, 4:36 PM

#

We cant really know every label before load nor do we want to load everything that soon do we?

north mist Jun 17, 2023, 4:41 PM

#

I mean it doesn't matter when we do the patching, since what we do is load the original asset, patch it, save it to cache, release it, and load the patched asset from cache, anyway

#

and the patching will only happen once/when new mods are installed

#

as for knowing the label, for the ":parts" patcher type we know that it will be "parts_data", same with any other specialized patcher types

#

and for the generic json one, doesn't the patch need to specify the file the be patched, anyway?

#

in most cases the filename should correspond to the label

#

though I agree it's not optimal

north mist Jul 5, 2023, 4:51 PM

#

alright, we're getting somewhere now

#

the preloader patch is not even needed with this new approach, because the only two methods I'm now patching in AssetProvider are not generic, so a Harmony patch is enough

#

(the new approach being I load all the existing keys when the patch cache is invalid, and I simply try to run the RebuildCache method for all the labels)

#

it's considerably slow when you run it for the first time (the last time I tried took 73s to go through all the labels, and I have a pretty beefy machine)

#

but on subsequent game loads it's barely noticeable

#

obviously, since it's cached

#

and I think I can still optimize the initial patching process

#

there's just one issue now - it seems like it works a bit too well lmao

#

since the game also uses the AssetProvider to create some game objects, and the methods for this internally also use LocateAsset(s)InExternalData, which I use to supply my own patched assets to the game, there seem to be some errors with duplicate objects being created during the first loading, and that freezes the game

#

so I just need to somehow fine-tune this

#

hm, and this time when I tried to run the game after deleting the cache, the rebuilding took 27 seconds

#

and again some issues with the kerbal attributes stuff

#

and this being the freezing exception

maiden wraith Jul 5, 2023, 6:10 PM

#

do u have a patch on minmus?

north mist Jul 5, 2023, 6:17 PM

#

nope

#

but the way it works right now, the cache contains all text assets, not just patched ones

upbeat hare Jul 5, 2023, 7:23 PM

#

north mist it's considerably slow when you run it for the first time (the last time I tried...

Still faster than KSP1

north mist Jul 5, 2023, 7:27 PM

#

Even though I switched to Harmony-only in this version, I might still have to use the preload patcher to fix the game object creation issue

maiden wraith Jul 5, 2023, 7:38 PM

#

wait 1m13s w/o mods?

#

and whats the difference when the cache is built?

north mist Jul 5, 2023, 7:50 PM

#

Well it tries to run all patches on all addressable labels

#

And then after that's done once, the results are cached so it doesn't need to be done again until mods change or patches are changed

upbeat hare Jul 5, 2023, 8:48 PM

#

maiden wraith and whats the difference when the cache is built?

Likely normal load times

upbeat hare Jul 6, 2023, 10:09 PM

#

Im thinking we should release this relatively soon, is there anything i should do?

north mist Jul 6, 2023, 10:10 PM

#

I mean, I can push what I have now, if you feel like having a look at those errors

upbeat hare Jul 6, 2023, 10:11 PM

#

Ill take a crack at it tomorrow

upbeat hare Jul 6, 2023, 11:45 PM

#

To elaborate

#

We hardcode a list of things we know are going to be loaded via loadbylabel, and patch those at that time and everything else beforehand, skipping over those keys

#

That was my idea

#

What that means is that, if an update adds a new key, itll be patched, but we can add it to the list if we know itll be loadbylabelled

north mist Jul 6, 2023, 11:49 PM

#

I mean sure, we can do that, but like 99% of all asset loading happens either at game load or at campaign load

#

so I don't see the added value in waiting for the actual labels to be loaded

upbeat hare Jul 6, 2023, 11:49 PM

#

But really, this just affects first load times, and not really, just distributes the tjme

#

So analyzing it that way, yeah there isnt a point

#

We should only cache what we need to though ... disk space and all that

upbeat hare Jul 7, 2023, 3:02 PM

#

Did you push the latest stuff munix?

north mist Jul 7, 2023, 3:56 PM

#

Yep, it's in the cache branch again

upbeat hare Jul 7, 2023, 4:44 PM

#

Aight, gonna check it out now

upbeat hare Jul 7, 2023, 4:45 PM

#

north mist and this being the freezing exception

This one is quite obviously both the cache data and original data being added to the planet definitions

north mist Jul 7, 2023, 4:45 PM

#

I don't see how that could happen though

#

since if a patched version is located, it's returned, and the original method doesn't actually run

#

at least that's how it's supposed to be

#

I could have made a mistake obviously

upbeat hare Jul 7, 2023, 4:47 PM

#

hmm...

#

    [HarmonyPatch(typeof(AssetProvider), nameof(AssetProvider.LocateAssetInExternalData))]
    [HarmonyPrefix]
    // ReSharper disable once InconsistentNaming
    private static bool LocateAssetInExternalData(object key, Type T, out IResourceLocation location, ref bool __result)
    {
        location = null;

        if (Locators.LocateAll(key.ToString(), T, out var patchedLocations))
        {
            location = patchedLocations[0];
            __result = true;
            return false;
        }

        foreach (var registeredResourceLocator in Assets._registeredResourceLocators)
        {
            if (registeredResourceLocator.Locate(key, T, out var locations))
            {
                location = locations[0];
                __result = true;
                return false;
            }
        }

        return false;
    }

#

this always returns false

#

which is wrong I think?

#

wait no, its right

#

my brain failed

north mist Jul 7, 2023, 4:50 PM

#

oh yeah lmao, I mean, I basically just included the original code of the method, it could probably just be

    [HarmonyPatch(typeof(AssetProvider), nameof(AssetProvider.LocateAssetInExternalData))]
    [HarmonyPrefix]
    // ReSharper disable once InconsistentNaming
    private static bool LocateAssetInExternalData(object key, Type T, out IResourceLocation location, ref bool __result)
    {
        location = null;

        if (Locators.LocateAll(key.ToString(), T, out var patchedLocations))
        {
            location = patchedLocations[0];
            __result = true;
            return false;
        }

        return true;
    }

#

I guess

upbeat hare Jul 7, 2023, 4:51 PM

#

you also never set __result to false at any point, but I don't think thats the issue here

maiden wraith Jul 7, 2023, 4:51 PM

#

it defaults as false no?

north mist Jul 7, 2023, 4:52 PM

#

north mist oh yeah lmao, I mean, I basically just included the original code of the method,...

yeah, that should either be before the final return statement, or just keep the method as I edited it here, since that leaves it to the original method to decide if it can find the asset or not

#

if we don't

upbeat hare Jul 7, 2023, 4:53 PM

#

Wait munix

#

when does that error appear

#

because to me it seems as if its an error appearing down the call chain of LoadByLabel

#

Munix, is it possible to split all the rebuilding patch flow actions into one per label?

#

So that it shows that it is doing something

#

        protected override void DoAction(Action resolve, Action<string> reject)
        {
            this._game.UI.SetLoadingBarText(base.Description);
            this._resolve = resolve;
            GameManager.Instance.Game.Assets.Load<TextAsset>(this._data.SavedGame.GalaxyDefinitionKey, new Action<TextAsset>(this.OnGalaxyDefinitionLoaded), true);
        }

#

That's what starts off the chain of calls that eventually causes the crash

#

            foreach (string reference in keys)
            {
                List<IResourceLocation> list = this.LocateAssetsInExternalData(reference);
                if (list.Count > 0)
                {
                    Addressables.LoadAssetsAsync<TAsset>(list, assetLoadCallback, Addressables.MergeMode.Union).Completed += value;
                }
            }
            Addressables.LoadAssetsAsync<TAsset>(keys, assetLoadCallback, Addressables.MergeMode.Union).Completed += value;

#

So essentially, its a bug in KSP2 code @north mist

#

Well, maybe not, but it always loads the keys from addressables even if it exists in ExternalData

north mist Jul 7, 2023, 5:03 PM

#

Ahh, yeah, that makes sense

upbeat hare Jul 7, 2023, 5:04 PM

#

screaming

north mist Jul 7, 2023, 5:04 PM

#

I mean, it would make sense that they want you to be able to add your own items in a mod and still load the original ones from the game

#

Seems like the switch from patching all the individual loading methods in AssetProvider to just patching the LocateAsset(s)InExternalData was premature

upbeat hare Jul 7, 2023, 5:05 PM

#

but how do they expect you to override stuff
Its Async, so modifying it to be able to be overriden would cause race conditions

#

I mean ...
We could modify this function

        public void RegisterBodyFromData(CelestialBodyCore jsonData)
        {
            string bodyName = jsonData.data.bodyName;
            this._celestialBodyLookup.Add(bodyName, jsonData);
        }

To

        public void RegisterBodyFromData(CelestialBodyCore jsonData)
        {
            string bodyName = jsonData.data.bodyName;
            this._celestialBodyLookup.TryAdd(bodyName, jsonData);
        }

#

but that might cause a race condition as I said

maiden wraith Jul 7, 2023, 5:07 PM

#

isnt the parts one async too?

#

almost every insertNameHereProvider has that Register function

#

so PartsProvider, CelestialBodyProvider, ResourceProvider etc

#

so every one of those would be a race condition no?

north mist Jul 7, 2023, 5:08 PM

#

Yeah, it will be better to just go back to our previous approach with the preload patches of the individual AssetProvider loading methods

#

So that we can keep the behavior of the ExternalData methods intact

upbeat hare Jul 7, 2023, 5:12 PM

#

The issue is also that this makes stuff like deletion of planets impossible through this method

#

because it will always load internal data

maiden wraith Jul 7, 2023, 5:14 PM

#

well depends

#

prefix it and, case deletion just dont add it to _celestialBodyLookup

#

not optimal obviously

#

but deletion has another problem

#

to properly delete a planet, you also need to remove any reference to it in other files

#

such as Interstellar map (or wtv it was that had the cordinates) and the galaxy definition

north mist Jul 7, 2023, 5:15 PM

#

It won't if we go back to the previous approach

#

At least it shouldn't

north mist Jul 7, 2023, 5:16 PM

#

upbeat hare The issue is also that this makes stuff like deletion of planets impossible thro...

.

#

Oh wait nevermind, I didn't read the code properly at first lmao

#

That's dumb, why are they calling methods from the Addressables class outside of the AssetProvider, when they have that class for addressables abstractions

#

Jesus

upbeat hare Jul 7, 2023, 5:18 PM

#

No that was from w/in assetprovider

north mist Jul 7, 2023, 5:18 PM

#

Oh well then that's fine

upbeat hare Jul 7, 2023, 5:22 PM

#

So we have to basically rewrite every dang method in asset provider then?

north mist Jul 7, 2023, 5:23 PM

#

Yeah, I already had that basically done before 0.1.3, but then I rewrote it to this

#

I should hopefully still have it shelved

maiden wraith Jul 7, 2023, 5:25 PM

#

upbeat hare So we have to basically rewrite every dang method in asset provider then?

husK

#

i back u up on asking dakota about taht then lmao

#

the thing is, if any change does come from it, we'll only be able to use it in the next update, thats 3 months from now

upbeat hare Jul 7, 2023, 5:36 PM

#

north mist I should hopefully still have it shelved

I hope so

north mist Jul 7, 2023, 5:43 PM

#

And it won't need to be every method, either, I think we should be able to just patch this one specifically causing issues

upbeat hare Jul 8, 2023, 11:10 PM

#

did you get the rewrite done?

#

and is there more I need to do?

north mist Jul 8, 2023, 11:13 PM

#

I do have the method to switch the original one for, now I'm trying to figure out your arcane IL runes

#

in the patcher class

upbeat hare Jul 8, 2023, 11:14 PM

#

Its not as arcane as you think, its just bouncing the call

#

it laods all the arguments and pushes them onto the stack and ccalls

north mist Jul 8, 2023, 11:15 PM

#

think this might work to find the target method?

#

since there are multiple Load ones, I need to find the one with a string[] parameter

upbeat hare Jul 8, 2023, 11:15 PM

#

it should

#

remember to do all the make generic stuff

#

var methodInModule = targetMethod.Module.ImportReference(extractedMethod);
var generic = methodInModule.MakeGeneric(targetMethod.GenericParameters.ToArray());

If you don't you'll get preload errors

north mist Jul 8, 2023, 11:16 PM

#

yeah, I'm just modifying the original one you wrote, so all that should still be in place

#

anything that needs to be changed here you think?

#

upbeat hare Jul 8, 2023, 11:17 PM

#

north mist

make sure its the right number of arguments

north mist Jul 8, 2023, 11:17 PM

#

it has three arguments as well

#

so that should be fine

upbeat hare Jul 8, 2023, 11:17 PM

#

including self?

north mist Jul 8, 2023, 11:17 PM

#

oh yeah, that's a good point lmao

upbeat hare Jul 8, 2023, 11:18 PM

#

tho actually

#

I skipped self there

#

that'd be ldarg 0

#

so your good

north mist Jul 8, 2023, 11:19 PM

#

how does this Ldarg_X stuff even work?

#

if there's more than 3

upbeat hare Jul 8, 2023, 11:19 PM

#

Instruction.Create(OpCodes.Ldarg, 4);

north mist Jul 8, 2023, 11:20 PM

#

ah, makes sense

#

alright, so I'm kinda getting somewhere

#

the patched method ran

#

but it didn't fix the issue

#

lmao

upbeat hare Jul 8, 2023, 11:28 PM

#

Oof

upbeat hare Jul 8, 2023, 11:29 PM

#

north mist if there's more than 3

Also more than 4, 0 is the first, but im skipping that due to it being unnecessary here

#

As assetprovider is all but a static class

north mist Jul 8, 2023, 11:29 PM

#

yeah, makes sense

#

alright so one other things that's kinda annoying right now is that some of the labels look like paths, which is messing with the way I'm saving the zip files with the labels as their names

#

it's not the biggest issue right now but the wall of errors I keep getting because of it is pretty annoying

#

though since the name of the archive is saved in inventory.json, I can just strip them from the file name without causing any issues I guess

upbeat hare Jul 8, 2023, 11:32 PM

#

Easy fix .Replace("/",SOMETHING_ELSE) and vice versa

north mist Jul 8, 2023, 11:32 PM

#

yeah, I don't even need to do it backwards

#

this also seems interesting

#

the method that was supposedly causing the issue was the Load<TAsset>(string[] keys, ...) one, right?

#

oh I guess it could also have been Load<TAsset>(IList<AssetReferenceT<TAsset>> assetRefs, ...)

#

though to be completely honest, I went through the code of CelestialBodyProvider and couldn't find any uses of either of those, and they're the only ones that match that screenshot you sent, I only see calls to Load<TAsset>(string key, ...) and LoadByLabel

upbeat hare Jul 9, 2023, 12:27 AM

#

Can I see what you wrote

#

This load function for referencfe

#

@north mist

#

#

this is the one its calling

#

The object keyObject one

north mist Jul 9, 2023, 12:33 AM

#

yeah, that's what I thought, and that in turn calls LoadAssetAsync

#

#

so I don't see how that could duplicate the data

upbeat hare Jul 9, 2023, 12:34 AM

#

Log the stack every time that is called

#

Is it pushed?

north mist Jul 9, 2023, 12:35 AM

#

not yet

#

lemme just make this change

upbeat hare Jul 9, 2023, 1:14 AM

#

Well I have to dip for the rest of the night, so I'll get to it tomorrow

north mist Jul 9, 2023, 1:18 AM

#

gotcha, I'll try to debug it and fix it

north mist Jul 9, 2023, 1:57 AM

#

huh, so I think I finally managed to patch the actual method that was causing issues, but now the IL patch is not working

Invalid IL code in KSP.Assets.AssetProvider:LoadAssetsAsync<T_REF> (string,System.Action`1<T_REF>): IL_0000: ldarg.1

maiden wraith Jul 9, 2023, 1:58 AM

#

you're passin IL_0000?

#

that doesnt seem right

#

what r u doing to check if youre on the right il?

north mist Jul 9, 2023, 1:59 AM

#

what I have now is

var coreType = coreAssembly.MainModule.Types.First(t => t.Name == "AssetProviderPatch");
var extractedMethod = coreType.Methods.First(m => m.Name == "LoadAssetsAsync");

var targetType = assemblyDefinition.MainModule.Types.Single(t => t.Name == "AssetProvider");
var targetMethod = targetType.Methods.Single(m => m.Name == "LoadAssetsAsync" && m.HasGenericParameters);
// Remove every single instruction from the body of the methods
// Emit call to our extracted method
var methodInModule = targetMethod.Module.ImportReference(extractedMethod);
var generic = methodInModule.MakeGeneric(targetMethod.GenericParameters.ToArray());
targetMethod.Body.Instructions.Clear();
targetMethod.Body.Instructions.Add(Instruction.Create(OpCodes.Ldarg_1));
targetMethod.Body.Instructions.Add(Instruction.Create(OpCodes.Ldarg_2));
targetMethod.Body.Instructions.Add(Instruction.Create(OpCodes.Call, generic));
targetMethod.Body.Instructions.Add(Instruction.Create(OpCodes.Ret));

updated from this which worked fine

var coreType = coreAssembly.MainModule.Types.First(t => t.Name == "AssetProviderPatch");
var extractedMethod = coreType.Methods.First(m => m.Name == "LoadByLabel");

var targetType = assemblyDefinition.MainModule.Types.Single(t => t.Name == "AssetProvider");
var targetMethod = targetType.Methods.Single(m => m.Name == "LoadByLabel" && m.HasGenericParameters);
// Remove every single instruction from the body of the methods
// Emit call to our extracted method
var methodInModule = targetMethod.Module.ImportReference(extractedMethod);
var generic = methodInModule.MakeGeneric(targetMethod.GenericParameters.ToArray());
targetMethod.Body.Instructions.Clear();
targetMethod.Body.Instructions.Add(Instruction.Create(OpCodes.Ldarg_1));
targetMethod.Body.Instructions.Add(Instruction.Create(OpCodes.Ldarg_2));
targetMethod.Body.Instructions.Add(Instruction.Create(OpCodes.Ldarg_3));
targetMethod.Body.Instructions.Add(Instruction.Create(OpCodes.Call, generic));
targetMethod.Body.Instructions.Add(Instruction.Create(OpCodes.Ret));

maiden wraith Jul 9, 2023, 1:59 AM

#

ah youre doing it like that... i dont think i can help you xD

#

wait

#

couldn't tell ya

#

my guess is its not finding the right method

#

thus the ILCode 000 being invalid

#

which from what i know is always valid for a valid method

#

but im no expert on this

north mist Jul 9, 2023, 2:02 AM

#

hm, that's weird, this is the signature of the method

#

and there's only one, no overloads

maiden wraith Jul 9, 2023, 2:03 AM

#

iirc async methods need other opcodes too

north mist Jul 9, 2023, 2:03 AM

#

it's not async

upbeat hare Jul 9, 2023, 2:03 AM

#

maiden wraith you're passin IL_0000?

thats the location

north mist Jul 9, 2023, 2:03 AM

#

it returns an AsyncOperationHandle

#

but not actually marked "async"

maiden wraith Jul 9, 2023, 2:03 AM

#

yeah i mean IL_0000 is always valid no?

maiden wraith Jul 9, 2023, 2:04 AM

#

north mist but not actually marked "async"

yeah got mislead by the name

upbeat hare Jul 9, 2023, 2:05 AM

#

maiden wraith yeah i mean IL_0000 is always valid no?

that means its the first op

north mist Jul 9, 2023, 2:05 AM

#

if it helps in any way, this is the method in AssetProviderPatch that I'm trying to call

upbeat hare Jul 9, 2023, 2:06 AM

#

whats the stack size

#

tho it shouldnt matter

north mist Jul 9, 2023, 2:08 AM

#

upbeat hare Jul 9, 2023, 2:10 AM

#

I meant the IL stack size, but it defaults to 8 so it dont matter

north mist Jul 9, 2023, 2:10 AM

#

oh lmao

upbeat hare Jul 9, 2023, 2:16 AM

#

Push the entire code and ill figure it out tomorrow

north mist Jul 9, 2023, 2:17 AM

#

yeah I was just about to do that, it's useless to be trying to figure this out at 4am when I'm barely awake

upbeat hare Jul 9, 2023, 2:58 AM

#

Wait a second

#

y'know loadassetsasync doesn't get called from that chain, right?

#

Wait it does

#

i'm dumb

#

But no, its actually load asset async

#

not loadassetsasync

#

also gods, we need to make it more visible what patch manager is doing, I keep feeling like its busted while loading

#

Looking at an emitted copy of the assembly in DNSpy, there is zero issue w/ this???

#

Excuse me, wtf

#

I fixed

#

it

maiden wraith Jul 9, 2023, 3:18 AM

#

how?

upbeat hare Jul 9, 2023, 3:18 AM

#

        targetMethod.Body.Instructions.Add(Instruction.Create(OpCodes.Ldarg_S,1));
        targetMethod.Body.Instructions.Add(Instruction.Create(OpCodes.Ldarg_S,2));

#

I changed it to Ldarg_S

#

for some reason, I was just testing something, and it loaded

#

wtf

#

C# why

#

WHY

#

As for why LoadCelestialBodyData still crashes it

#

Well, because, @north mist patched LoadAssetsAsync rather than LoadAssetAsync

maiden wraith Jul 9, 2023, 3:21 AM

#

lmao

upbeat hare Jul 9, 2023, 3:22 AM

#

Why did that fix it, like seriously

#

But wait, LOadAssetAsync should already be looking into external data???"

#

Wait, then how is this breaking at all

#

[Info   :Galaxy Definition Load Actions] Loading from: GalaxyDefinition_Default
[Info   :Galaxy Definition Load Actions]   at PatchManager.Core.Patches.Runtime.GalaxyDefinitionTestPatch.OnGalaxyDefinitionLoaded (UnityEngine.TextCore.Text.TextAsset asset) [0x00023] in <6bb337ee19bc4b21b57fd2d85604a659>:0 
  at KSP.Game.Load.LoadCelestialBodyDataFilesFlowAction.DMD<KSP.Game.Load.LoadCelestialBodyDataFilesFlowAction::OnGalaxyDefinitionLoaded> (KSP.Game.Load.LoadCelestialBodyDataFilesFlowAction , UnityEngine.TextAsset ) [0x00000] in <f07798ad7fdc4402ab44d85c99f8e7b0>:0 
  at KSP.Assets.AssetProvider+<>c__DisplayClass11_0`1[TAsset].<Load>b__1 (UnityEngine.ResourceManagement.AsyncOperations.AsyncOperationHandle`1[TObject] handle) [0x00000] in <f07798ad7fdc4402ab44d85c99f8e7b0>:0 
  at DelegateList`1[T].Invoke (T res) [0x00000] in <11e2f4eda1124a9f83368725a730d057>:0 
  at UnityEngine.ResourceManagement.AsyncOperations.AsyncOperationBase`1[TObject].InvokeCompletionEvent () [0x00000] in <11e2f4eda1124a9f83368725a730d057>:0 
  at UnityEngine.ResourceManagement.AsyncOperations.AsyncOperationBase`1[TObject].UnityEngine.ResourceManagement.AsyncOperations.IAsyncOperation.InvokeCompletionEvent () [0x00000] in <11e2f4eda1124a9f83368725a730d057>:0 
  at UnityEngine.ResourceManagement.ResourceManager.ExecuteDeferredCallbacks () [0x00000] in <11e2f4eda1124a9f83368725a730d057>:0 
  at UnityEngine.ResourceManagement.ResourceManager.Update (System.Single unscaledDeltaTime) [0x00000] in <11e2f4eda1124a9f83368725a730d057>:0 
  at MonoBehaviourCallbackHooks.Update () [0x00000] in <11e2f4eda1124a9f83368725a730d057>:0

#

I only get one load message

#

wait a second

#

Wait a fucking second

#

we've been barking up the wrong tree

#

I have a suspicion I need to confirm

#

We've somehow been duplicating the celestial bodies?

#

lemme check this

#

So, I think we might still need to be patching loadbylabel

#

[Info   :Galaxy Definition Load Actions] AsteroidD01 added 2x
[Info   :Galaxy Definition Load Actions] Minmus added 2x
[Info   :Galaxy Definition Load Actions] AsteroidA01 added 2x
[Info   :Galaxy Definition Load Actions] AsteroidB01 added 2x
[Info   :Galaxy Definition Load Actions] Mun added 2x
[Info   :Galaxy Definition Load Actions] Eve added 2x
[Info   :Galaxy Definition Load Actions] Dres added 2x
[Info   :Galaxy Definition Load Actions] Ike added 2x
[Info   :Galaxy Definition Load Actions] Kerbol added 2x
[Info   :Galaxy Definition Load Actions] Bop added 2x
[Info   :Galaxy Definition Load Actions] AsteroidC01 added 2x
[Info   :Galaxy Definition Load Actions] Eeloo added 2x
[Info   :Galaxy Definition Load Actions] AsteroidE01 added 2x
[Info   :Galaxy Definition Load Actions] Jool added 2x
[Info   :Galaxy Definition Load Actions] Gilly added 2x
[Info   :Galaxy Definition Load Actions] Pol added 2x
[Info   :Galaxy Definition Load Actions] Kerbin added 2x
[Info   :Galaxy Definition Load Actions] Duna added 2x
[Info   :Galaxy Definition Load Actions] Vall added 2x
[Info   :Galaxy Definition Load Actions] Laythe added 2x
[Info   :Galaxy Definition Load Actions] Tylo added 2x
[Info   :Galaxy Definition Load Actions] Moho added 2x

#

Yep, we are duplicating every body

#

        __instance._game.Assets.LoadByLabel<TextAsset>("celestial_bodies",null, x =>
        {
            source.LogInfo($"Result callback called {n++} times");
            foreach (var text in x)
            {
                if (numbers.ContainsKey(text.name))
                {
                    numbers[text.name] += 1;
                }
                else
                {
                    numbers.Add(text.name,1);
                }

                source.LogInfo($"{text.name} added {numbers[text.name]}x");
            }

            GameManager.Instance.Game.Assets.ReleaseAsset(x);
        });

#

debug code for that

#

Sop the issue ... ultimately stems from LoadByLabel

#

now i sleep

north mist Jul 9, 2023, 7:47 AM

#

upbeat hare Sop the issue ... ultimately stems from LoadByLabel

Yes, I know, that's what my new code was fixing

#

Since LoadByLabel calls LoadAsset__s__Async

#

To be fair I don't know if it did fix it, but at least I had the right method

north mist Jul 9, 2023, 9:04 AM

#

upbeat hare ```cs targetMethod.Body.Instructions.Add(Instruction.Create(OpCodes.Ldar...

this just gives me another error

#

#

#

wtf

north mist Jul 9, 2023, 9:23 AM

#

the only thing that's different between LoadByLabel/Load (which I was able to patch successfully) and LoadAssetsAsync is the fact that the latter has a return type of AsyncOperationHandle<IList<T>>, while the first two were void

north mist Jul 9, 2023, 9:43 AM

#

aaahhhh this is so frustrating

upbeat hare Jul 9, 2023, 10:56 AM

#

north mist this just gives me another error

H-how??

#

I was able to compile and run it

#

I wouldve gotten a preload error otherwise?

#

Wait it literally doesnt like the opcode??

north mist Jul 9, 2023, 11:21 AM

#

Yeah I have no clue, neither this version nor the one with _1 and _2 run for me

upbeat hare Jul 9, 2023, 11:43 AM

#

wait I'm dumb, I got the same error, hence why I was able to load

#

but I didn't see a preload error which is what it shouldve been?

#

I know the fix to that error

#

I didn't know that Mono.cecil is that strict

#

Lemme try something

#

(never debug code when your half asleep)

#

Also spacewarps console is acting very glitchy for me

#

[Error  :   BepInEx] Failed to run [PatchManager.PreloadPatcher.Patcher] when patching [Assembly-CSharp]. This assembly will not be patched. Error: System.Reflection.TargetInvocationException: Exception has been thrown by the target of an invocation. ---> System.ArgumentOutOfRangeException: Specified argument was out of the range of valid values.
  at Mono.Collections.Generic.Collection`1[T].get_Item (System.Int32 index) [0x00009] in <6034b380a22b41a596c9dc29d282c0a9>:0 
  at PatchManager.PreloadPatcher.Patcher.Patch (Mono.Cecil.AssemblyDefinition& assemblyDefinition) [0x00183] in C:\Users\arall\PatchManager\src\PatchManager.PreloadPatcher\Patcher.cs:74 
  at (wrapper managed-to-native) System.Reflection.MonoMethod.InternalInvoke(System.Reflection.MonoMethod,object,object[],System.Exception&)
  at System.Reflection.MonoMethod.Invoke (System.Object obj, System.Reflection.BindingFlags invokeAttr, System.Reflection.Binder binder, System.Object[] parameters, System.Globalization.CultureInfo culture) [0x00032] in <695d1cc93cca45069c528c15c9fdd749>:0 
   --- End of inner exception stack trace ---
  at System.Reflection.MonoMethod.Invoke (System.Object obj, System.Reflection.BindingFlags invokeAttr, System.Reflection.Binder binder, System.Object[] parameters, System.Globalization.CultureInfo culture) [0x00048] in <695d1cc93cca45069c528c15c9fdd749>:0 
  at System.Reflection.MethodBase.Invoke (System.Object obj, System.Object[] parameters) [0x00000] in <695d1cc93cca45069c528c15c9fdd749>:0 
  at BepInEx.Preloader.Patching.AssemblyPatcher+<>c__DisplayClass11_0.<AddPatchersFromDirectory>b__3 (Mono.Cecil.AssemblyDefinition& pAss) [0x0000c] in <fc9d7fbc6dcb44cf87be11d8d92ae161>:0 
  at BepInEx.Preloader.Patching.AssemblyPatcher.PatchAndLoad (System.String[] directories) [0x001b5] in <fc9d7fbc6dcb44cf87be11d8d92ae161>:0

#

Ah this error makes no sense, but also makes sense

#

there are no arguments to the method we are patching????

north mist Jul 9, 2023, 11:52 AM

#

https://tenor.com/view/confused-meme-girl-uhmmm-what-now-gif-16804319

Tenor

upbeat hare Jul 9, 2023, 11:52 AM

#

targetMethod.Parameters[1]

north mist Jul 9, 2023, 11:52 AM

#

wtf

upbeat hare Jul 9, 2023, 11:52 AM

#

I was using that

north mist Jul 9, 2023, 11:52 AM

#

it has to have two

#

even when I was debug logging the FullName of both extractedMethod and targetMethod, they had the correct signatures

#

with the same return type and the same two parameters

upbeat hare Jul 9, 2023, 11:54 AM

#

Wait no I'm being dumb again

#

[Info   :Patch Manager Preload] Parameter: key
[Info   :Patch Manager Preload] Parameter: assetLoadCallback

#

Apparently the Parameters array doesn't contain the this

north mist Jul 9, 2023, 11:54 AM

#

ah

#

by the way I also tried changing the extractedMethod to not be static, and added Ldarg_0, but no bueno

upbeat hare Jul 9, 2023, 11:55 AM

#

And we are back to square one

#

[EXC 07:54:54.478] InvalidProgramException: Invalid IL code in KSP.Assets.AssetProvider:LoadAssetsAsync<T_REF> (string,System.Action`1<T_REF>): IL_0000: ldarg.s   1


    KSP.Assets.AssetProvider.LoadByLabel[T] (System.String label, System.Action`1[T] assetLoadCallback, System.Action`1[T] resultCallback) (at <f07798ad7fdc4402ab44d85c99f8e7b0>:0)
    KSP.Game.DifficultyOptionsDataManager.Load () (at <f07798ad7fdc4402ab44d85c99f8e7b0>:0)
    KSP.Game.GameManager.InitializeDifficultyOptionsManager (System.Action resolve, System.Action`1[T] reject) (at <f07798ad7fdc4402ab44d85c99f8e7b0>:0)
    KSP.Game.Flow.GenericFlowAction.DoAction (System.Action resolve, System.Action`1[T] reject) (at <f07798ad7fdc4402ab44d85c99f8e7b0>:0)
    KSP.Game.Flow.FlowAction.Do (System.Action`1[T] resolve, System.Action`2[T1,T2] reject) (at <f07798ad7fdc4402ab44d85c99f8e7b0>:0)
    KSP.Game.Flow.SequentialFlow.NextFlowAction () (at <f07798ad7fdc4402ab44d85c99f8e7b0>:0)
    KSP.Game.Flow.SequentialFlow.Update () (at <f07798ad7fdc4402ab44d85c99f8e7b0>:0)

north mist Jul 9, 2023, 11:55 AM

#

yeah I'm just really confused by all this

#

it worked just fine with Load and LoadByLabel, so why doesn't it work with LoadAssetsAsync

#

the only visible difference is that those were void and LoadAssetsAsync returns AsyncOperationHandle

upbeat hare Jul 9, 2023, 12:02 PM

#

And its so damn odd that the base starts w/ a ldarg just fine??

#

Lemme try something else

upbeat hare Jul 9, 2023, 12:48 PM

#

Ahh, great, a new error

#

Just crashing w/o any error

#

Ahh

        public AsyncOperationHandle<IList<T>> LoadAssetsAsync<T>(string key, Action<T> assetLoadCallback)
        {
            /*
An exception occurred when decompiling this method (06008365)

ICSharpCode.Decompiler.DecompilerException: Error decompiling UnityEngine.ResourceManagement.AsyncOperations.AsyncOperationHandle`1<System.Collections.Generic.IList`1<T>> KSP.Assets.AssetProvider::LoadAssetsAsync<T>(System.String,System.Action`1<T>)

 ---> System.NullReferenceException: Object reference not set to an instance of an object.
   at ICSharpCode.Decompiler.ILAst.ILInlining.InlineOneIfPossible(ILBlockBase block, List`1 body, Int32 pos, Boolean aggressive) in D:\a\dnSpy\dnSpy\Extensions\ILSpy.Decompiler\ICSharpCode.Decompiler\ICSharpCode.Decompiler\ILAst\ILInlining.cs:line 218
   at ICSharpCode.Decompiler.ILAst.ILInlining.InlineAllInBlock(ILBlock block) in D:\a\dnSpy\dnSpy\Extensions\ILSpy.Decompiler\ICSharpCode.Decompiler\ICSharpCode.Decompiler\ILAst\ILInlining.cs:line 144
   at ICSharpCode.Decompiler.ILAst.ILInlining.InlineAllVariables() in D:\a\dnSpy\dnSpy\Extensions\ILSpy.Decompiler\ICSharpCode.Decompiler\ICSharpCode.Decompiler\ILAst\ILInlining.cs:line 108
   at ICSharpCode.Decompiler.ILAst.ILAstOptimizer.Optimize(DecompilerContext context, ILBlock method, AutoPropertyProvider autoPropertyProvider, StateMachineKind& stateMachineKind, MethodDef& inlinedMethod, AsyncMethodDebugInfo& asyncInfo, ILAstOptimizationStep abortBeforeStep) in D:\a\dnSpy\dnSpy\Extensions\ILSpy.Decompiler\ICSharpCode.Decompiler\ICSharpCode.Decompiler\ILAst\ILAstOptimizer.cs:line 227
   at ICSharpCode.Decompiler.Ast.AstMethodBodyBuilder.CreateMethodBody(IEnumerable`1 parameters, MethodDebugInfoBuilder& builder) in D:\a\dnSpy\dnSpy\Extensions\ILSpy.Decompiler\ICSharpCode.Decompiler\ICSharpCode.Decompiler\Ast\AstMethodBodyBuilder.cs:line 123
   at ICSharpCode.Decompiler.Ast.AstMethodBodyBuilder.CreateMethodBody(MethodDef methodDef, DecompilerContext context, AutoPropertyProvider autoPropertyProvider, IEnumerable`1 parameters, Boolean valueParameterIsKeyword, StringBuilder sb, MethodDebugInfoBuilder& stmtsBuilder) in D:\a\dnSpy\dnSpy\Extensions\ILSpy.Decompiler\ICSharpCode.Decompiler\ICSharpCode.Decompiler\Ast\AstMethodBodyBuilder.cs:line 99
   --- End of inner exception stack trace ---
   at ICSharpCode.Decompiler.Ast.AstMethodBodyBuilder.CreateMethodBody(MethodDef methodDef, DecompilerContext context, AutoPropertyProvider autoPropertyProvider, IEnumerable`1 parameters, Boolean valueParameterIsKeyword, StringBuilder sb, MethodDebugInfoBuilder& stmtsBuilder) in D:\a\dnSpy\dnSpy\Extensions\ILSpy.Decompiler\ICSharpCode.Decompiler\ICSharpCode.Decompiler\Ast\AstMethodBodyBuilder.cs:line 99
   at ICSharpCode.Decompiler.Ast.AstBuilder.<>c__DisplayClass90_0.<AddMethodBody>b__0() in D:\a\dnSpy\dnSpy\Extensions\ILSpy.Decompiler\ICSharpCode.Decompiler\ICSharpCode.Decompiler\Ast\AstBuilder.cs:line 1528
*/;
        }

#

Oh I know what happened

#

I forgot to copy over the variables

upbeat hare Jul 9, 2023, 1:22 PM

#

aight, I give up, this is insanity

north mist Jul 9, 2023, 1:22 PM

#

nooooooooo

upbeat hare Jul 9, 2023, 1:23 PM

#

Is there a way to just not use this method

north mist Jul 9, 2023, 1:23 PM

#

well we can patch every other method where it's used

#

lmao

upbeat hare Jul 9, 2023, 1:24 PM

#

at this point that might just be cleaner

north mist Jul 9, 2023, 1:26 PM

#

oh, it seems like the only place where it's actually used is just LoadByLabel anyway, lol

#

which we already were patching previously

#

so this was kinda unnecessary anyway

#

munley

upbeat hare Jul 9, 2023, 1:27 PM

#

dies internally

north mist Jul 9, 2023, 1:28 PM

#

I mean it's also called by LoadAllRaw, but 1) that actually isn't used anywhere, and 2) it seems to me like that's a bug anyway, the ...Raw methods are supposed to call Addressables methods directly

upbeat hare Jul 9, 2023, 1:30 PM

#

ao just undo all of what we just did to repatch loadbylabel

#

Anyways, imma go grab some brekky

north mist Jul 9, 2023, 1:43 PM

#

and done, patching of all labels now works

upbeat hare Jul 9, 2023, 1:44 PM

#

It was that simple

north mist Jul 9, 2023, 1:44 PM

#

a bit more polishing and I think we can make an experimental release

upbeat hare Jul 9, 2023, 1:44 PM

#

Now I can finally make a test patch to scale up the solar system

#

Does it still build cache on first run?

north mist Jul 9, 2023, 1:45 PM

#

yeah

upbeat hare Jul 9, 2023, 1:46 PM

#

In that case a necessary polish would be splitting every label into its own flow action

north mist Jul 9, 2023, 1:46 PM

#

and not saving unpatched labels into the cache

upbeat hare Jul 9, 2023, 1:46 PM

#

That too

upbeat hare Jul 9, 2023, 1:46 PM

#

upbeat hare In that case a necessary polish would be splitting every label into its own flow...

Mostly to stop it from "hanging"

north mist Jul 9, 2023, 1:47 PM

#

yeah, makes sense

#

alright, for now i merged it all into main, and I gotta go for a bit, but I'll be back later

upbeat hare Jul 9, 2023, 1:48 PM

#

Gotcha

#

I just realized a possible issue

#

Mods can define their own labels

#

And their bundles are loaded later

upbeat hare Jul 9, 2023, 9:53 PM

#

Aight, you said its on main branch?

north mist Jul 9, 2023, 9:54 PM

#

yeah

upbeat hare Jul 9, 2023, 9:56 PM

#

Aight got it

#

Wait, I see no actions being registered at all

north mist Jul 9, 2023, 9:57 PM

#

I haven't done that yet, if you mean the inidividual flow actions for each label

upbeat hare Jul 9, 2023, 9:58 PM

#

Wait, I forgot to update

#

We could do it as a per mod flow action, but I think that might run too late? for some things to be patched

#

Ah, I see whats going on, kinda

#

for the cache rebuilding

north mist Jul 9, 2023, 10:01 PM

#

yeah I really don't know what the best time to do the patching is, we don't want to do it too early, before mod catalogs are registered, but also not too late

upbeat hare Jul 9, 2023, 10:02 PM

#

which is why the lazy patching system was decent, but also not really possible to maintain?

north mist Jul 9, 2023, 10:02 PM

#

yeah, it's a shame but I don't think we could easily work around the async stuff

upbeat hare Jul 9, 2023, 10:04 PM

#

Hmm, we have to patch after all addressables are loaded, its kinda just necessary I think, if we want to be able to patch mod stuff

#

Including modded parts for that matter

#

as for the not caching everything ... thats simple enough we just need a value in the cache when something is the original value rather than being cached?

#

but looking at the architecture of this, maybe not, as we are completely overriding the resource locator system

#

What all is loaded before campaign load?

#

Actually wait, the simplest way to reduce cache size, is just to remove labels that have zero changes

#

I'm gonna quickly implement that rq

north mist Jul 9, 2023, 10:12 PM

#

alright, awesome

upbeat hare Jul 9, 2023, 10:18 PM

#

just zipping up the current cache so I can inspect it for future purposes

#

A whole lot smaller

north mist Jul 9, 2023, 10:21 PM

#

yeah, that's much better

upbeat hare Jul 9, 2023, 10:21 PM

#

I can't really do sublabel optimization at this point, though I do want to be able to

#

Also, we were duplicatiing information? Because stuff can have more than one label I think

#

but thats only in the inventory.json and isn't easily fixable

#

Do you mind if I don't pretty print the inventory file to save space?

#

as that does get to almost a megabyte in size w/ everything patched

#

Ehh, it doesn't matter too much

#

So @north mist addressables in spacewarp get added at the end of everything :/

#

so anything that gets loaded beforehand necessarily has to get skipped?

north mist Jul 9, 2023, 10:28 PM

#

upbeat hare Also, we were duplicatiing information? Because stuff can have more than one lab...

that, that was another thing I needed to optimize

#

well it's in inventory.json on purpose

#

because technically someone could request a specific asset both by its own label, and by its group label

#

although I don't know why that would happen, but it's technically possible

north mist Jul 9, 2023, 10:30 PM

#

upbeat hare so anything that gets loaded beforehand necessarily has to get skipped?

I don't think I get this

upbeat hare Jul 9, 2023, 10:32 PM

#

essentially, spacewarp loads addressables as late as it can in the progress, which is after everything else has been loaded, meaning that we need to patch after that, but files that are read before then will be missed in that case

north mist Jul 9, 2023, 10:32 PM

#

yeah

#

sounds like it might be useful to split the patching process after all

upbeat hare Jul 9, 2023, 10:33 PM

#

we need a balanced patching system of at load time, and at read time for that ... which may be a bit complex

#

not necessary for an experimental release, but will be later

north mist Jul 9, 2023, 10:33 PM

#

well for starters we could just patch specifically the things that we know are being read before SpaceWarp registers catalogs

#

and leave the rest after that happens

upbeat hare Jul 9, 2023, 10:34 PM

#

Fair

#

lemme write the stuff for post spacewarp, and then I'll leave you to doing presw stuff

north mist Jul 9, 2023, 10:35 PM

#

sure

upbeat hare Jul 9, 2023, 10:35 PM

#

the issue is then, that we still have to read patch files as early as possible, which is fine, thats not the bottleneck

north mist Jul 9, 2023, 10:35 PM

#

yeah, I'm kinda assuming that we won't be able to make it fully lazy-loaded possibly ever

#

but we can at least optimize as best as we can

upbeat hare Jul 9, 2023, 10:40 PM

#

I might have to un-async RebuildAllCache...

#

Oh gods, patches will still load despite error state in space warp

upbeat hare Jul 9, 2023, 11:05 PM

#

Alright this is insanity, unities rendering is single-threaded right?

#

That means then, that I have to generate all the loading actions on the fly :3

upbeat hare Jul 9, 2023, 11:22 PM

#

I see why itt akes so long munix

#

@north mist get this, there is an extreme amount of labels, that just iterating through them takes very long

north mist Jul 9, 2023, 11:23 PM

#

yeah, I've noticed that

#

I wish we had a way to filter through them somehow

#

I mean I guess we could add like a blacklist of labels that we don't want to try to patch

#

if we know they don't contain TextAssets etc

#

that way we won't have to attempt to load their assets

#

not sure how much it would shorten the loading time though

upbeat hare Jul 9, 2023, 11:27 PM

#

And lazy patching ... is impossible

#

Anyways @north mist

#

One thing we could do, filter out any labels that look like they are just hash strings

#

The only useful thing those add are just the privacy policies

#

Anda also ones that are integers

#

Ahhh

[Info   :Patch Manager] 0 mod libraries loaded!
[Info   :Patch Manager] 0 patchers registered!

#

Alright so its not much faster

#

but it at least doesn't go through the almost useless stuff

#

which is more visually appealing

#

But anyways! it does show the amount of patches

#

It's just going to hang a bit on the labels it is patching

#

just had to bump patch manager up to spec 1.3 lol

#

Shit I broke wwise again accidentally

#

Idfk how

#

but I did

maiden wraith Jul 10, 2023, 12:00 AM

#

upbeat hare Anyways <@366972069440913418>

what did you do? just added a UITK text?

upbeat hare Jul 10, 2023, 12:00 AM

#

No, I patched the loadingbar to inject tips into there

#

I might add it as a spacewarp thing later on

maiden wraith Jul 10, 2023, 12:00 AM

#

heh fair

#

yes pls

#

not in PM scope but what about a timer for how long each loading screen takes? maybe even a debug feature ig

north mist Jul 10, 2023, 12:02 AM

#

upbeat hare But anyways! it does show the amount of patches

oh yeah, that's looking nice

upbeat hare Jul 10, 2023, 12:02 AM

#

Except, the way I did it, broke WWISE somehow

maiden wraith Jul 10, 2023, 12:02 AM

#

oh as always

north mist Jul 10, 2023, 12:02 AM

#

it happens if you create an instance of a flow action too early

#

that's why I have the wrapper API for it in Patch Manager

upbeat hare Jul 10, 2023, 12:03 AM

#

Can't I just harmony patch that away w/ space warp

north mist Jul 10, 2023, 12:04 AM

#

no clue, but it would be nice

upbeat hare Jul 10, 2023, 12:05 AM

#

Likely not...

#

It does mean, I need to pass a state inbetween awake, and initialized though

#

as the only time you are legally allowed to add space warp general loading actions then is in PreInitialized

#

And I need to use SW for this

#

Also, I know what causes the bug

#

I2Localization kills WWISE

#

As when you create a flow action, it attempts to access I2Localization, which may not be loaded yet, so through some chain of events WWISE dies

north mist Jul 10, 2023, 12:10 AM

#

yeah, I'm assuming it has to do with the fact that I2 can also be used to localize audio

#

so it tries to init wwise

#

without the correct settings

upbeat hare Jul 10, 2023, 12:10 AM

#

No wait, I legit can't use general loading actions at all, they are broken in space warp

#

Aight time to bug fix that

#

So essentially, this will rely on 1.3.1

upbeat hare Jul 10, 2023, 12:12 AM

#

upbeat hare No wait, I legit can't use general loading actions at all, they are broken in sp...

You can't create them too early, but once too early is over (in preinitialize), there is no time to create them, as they would already be added

north mist Jul 10, 2023, 12:13 AM

#

yeah this is the only place where I found them usable

#

at the end of GameManager.StartBootstrap

upbeat hare Jul 10, 2023, 12:13 AM

#

They are usable anywhere in gamemanager.startbootstrap

#

but that'll already have been called by the time I need to do stuff

north mist Jul 10, 2023, 12:14 AM

#

yeah that makes sense

upbeat hare Jul 10, 2023, 12:15 AM

#

So, given that API is literally broken, I'm gonna bug fix it

#

Screw it, I'll release 1.3.1 tonight

#

as there are 2 things I want to add

#

@north mist do you mind if I release a spacewarp 1.3.1, it'll at least make semantic versioning line up if anything else

#

not the complete right place to ask, but it is kinda necessary for me to continue some stuff

north mist Jul 10, 2023, 12:25 AM

#

yeah, definitely go for it

upbeat hare Jul 10, 2023, 12:31 AM

#

wait, I can just replace the download on spacedock

#

But ehh I needed to bump the version anyways

north mist Jul 10, 2023, 12:32 AM

#

upbeat hare *wait, I can just replace the download on spacedock*

yeah but that's really ew

upbeat hare Jul 10, 2023, 12:39 AM

#

So anyways munix, the QoL branch has all of what I just did

#

Some of it is ugly AF but is actually necessary to get the amount of patches to show in the loading tips

north mist Jul 10, 2023, 12:40 AM

#

awesome, thanks

upbeat hare Jul 10, 2023, 11:34 PM

#

Alright, what do we need now

#

A list of labels to patch before everything?

#

Then a few example patches to show off how to patch

#

Actually, lemme write my larger solar system patch :3

upbeat hare Jul 10, 2023, 11:54 PM

#

I'll scream if this actually works

maiden wraith Jul 11, 2023, 2:01 AM

#

did you scream?

#

i dont think you screamed

upbeat hare Jul 11, 2023, 4:21 AM

#

I did not scream

maiden wraith Jul 12, 2023, 10:49 PM

#

for sound patching there's 2 alternatives

#

the way that wwise works for sound, is that you provide either an unique name or an unique id, and it will play that sound when triggered

#

now, i dont know of a reliable way to retrieve that name or id yet at runtime or while datamining, which is a problem

#

solution A

just copy the gameobject with the component KSPPartAudio from the target part
solution B
Find the reference somehow and provide somwhere (ie wiki 👀 ) a lookup table of part to wwise sounds names

the patch could be something like this to copy everything

part[MyPartName]
{
@AUDIO = COPY_AUDIO[swivel]
}

or

part[MyPartName]
{
@AUDIO.OnDestroyed = "Swivel_Engine_Destroyed"

@AUDIO.OnEngineStart= 1728947489
}

#

i dont know scss syntax so 99% my syntax is wrong but you get the idea

#

and this will be widely used because things like part collision, part destroyed, ambient sounds are all using wwise, so if you want your parts to make noise once destroyed (via collision or heat or wtv) you need to either create your own sounds or use the game's reference

upbeat hare Jul 13, 2023, 12:23 AM

#

This would require a substantial rewrite of PM

maiden wraith Jul 13, 2023, 12:23 AM

#

it doesnt need to be literally like this btw

#

and if it requires that then its better another solution

#

tho the way that audio is assignes, is via a child gameobject of the prefab

#

maiden wraith Jul 13, 2023, 11:22 PM

#

so @upbeat hare , i found a way to get all the sound references from all parts

#

what could we do on patch manager to patch them in?

#

w/o much work on your side

upbeat hare Jul 13, 2023, 11:24 PM

#

I think thats a fundamental misunderstanding of how patch manager works
The sound references aren't stored in a text based format, right @north mist ?

north mist Jul 13, 2023, 11:24 PM

#

I have no clue about that tbh

#

I need to look into the json files

maiden wraith Jul 13, 2023, 11:25 PM

#

i dont think so ngl

#

they are stored in the monobehaviour as a Ak.Wwise.Event

#

there might be some connection from that to a text based format

#

but again, we can get each sound just with the name of the event

upbeat hare Jul 13, 2023, 11:26 PM

#

That is definitely not the aim of patch manager

#

Stored in the monobehaviour

maiden wraith Jul 13, 2023, 11:27 PM

#

heh sure

#

tho i remember in the start that there was a talk about custom patches like for LFO etc

#

i guess the most we could do in that regard is to copy a plume from one place to another

north mist Jul 13, 2023, 11:28 PM

#

maiden wraith tho i remember in the start that there was a talk about custom patches like for ...

yeah, definitely

#

we'll need to do a shit ton of custom patching to actually get this to be useful

#

that's why I tried to make it modular so we can split off those special patch types into separate assemblies and allow people to make their own

maiden wraith Jul 13, 2023, 11:29 PM

#

then sounds could fit those custom patches? idk i might be misunderstanding patch manager i think

north mist Jul 13, 2023, 11:31 PM

#

well right now all it does is load all TextAssets from addressable labels, runs all patches on them to transform the JSON, saves it to cache, and then patches the AssetProvider to read assets from the cache if they're in it

#

so it will need a lot of work to do anything beyond the scope of json files

#

but I always kinda expected we'll have to do that, anyway

maiden wraith Jul 13, 2023, 11:32 PM

#

well thats were i come in, for parts, effects, and now sounds

#

but idek how to start that lmao

#

have to take a deeper look into patch manager

#

and into Scss syntax. something like this #1115274490490929172 message would be amazing

#

but as cheese said it would take a lot of rewriting

#

which i'd rather avoid

upbeat hare Jul 14, 2023, 12:14 PM

#

hmm, I think we can parellize the patches on any given label

maiden wraith Jul 14, 2023, 1:45 PM

#

i think its a good idea, that is, the labels actions are independent

#

which im 99% sure they are but yeah

upbeat hare Jul 14, 2023, 2:00 PM

#

Am tryna get patch manager to work reeee

#

Tryna do the 2.5 scale solar system

maiden wraith Jul 14, 2023, 2:01 PM

#

whats not working?

#

iirc for that you need to mess with PQS for collision to work decently

upbeat hare Jul 14, 2023, 2:01 PM

#

maiden wraith whats not working?

My f-in json ruleset :3

maiden wraith Jul 14, 2023, 2:02 PM

#

ah lmao

#

oh btw, we will have an API (or better, an abstract class) to inherit so that other mods can process patches right?
something like

public abstract class PMPatchProcessor{

public abstract void ReceivePatch(string Selector, string target, string bla bla bla bla);

}

upbeat hare Jul 14, 2023, 2:08 PM

#

Thats what rulesets are for :3
But hmm

maiden wraith Jul 14, 2023, 2:08 PM

#

gredat my keyboard is fucked noo_why

#

im adsking because i might make a sound manadger or sound helper wtv and i could process the patches there

#

but for the time being i'd just have config files to copy sound from parts to other parts

upbeat hare Jul 14, 2023, 2:39 PM

#

2.5x gravity definitely wobblifies the rockets more

upbeat hare Jul 14, 2023, 3:37 PM

#

Damn, I'm not sure if I can modify the PQS with patches alone

#

I might have to write a simple code patch for that

maiden wraith Jul 14, 2023, 3:53 PM

#

Yup probably not, its not even in the json data i think

upbeat hare Jul 14, 2023, 3:53 PM

#

everything else I can do in JSON

maiden wraith Jul 14, 2023, 3:54 PM

#

U know that reminds me of the comment made on the forum post

upbeat hare Jul 14, 2023, 3:54 PM

#

Can you tell me though where the PQS stuff is stored at runtime

maiden wraith Jul 14, 2023, 3:54 PM

#

We should show these things to people to solidy the fact that we cant just edit configs and it will work

#

It requires more than that to modify parta now

north mist Jul 14, 2023, 4:26 PM

#

maiden wraith We should show these things to people to solidy the fact that we cant just edit ...

yeah but then they'll just say that it's another reason why KSP 2 sucks and why they won't mod it

upbeat hare Jul 14, 2023, 5:07 PM

#

Why the hell won't my roslyn patch compile, I'll just make a dll mod for this test

#

Conclusion: I suffer from a deficiency in brain cells

maiden wraith Jul 14, 2023, 5:40 PM

#

north mist yeah but then they'll just say that it's another reason why KSP 2 sucks and why ...

Well fuck them then, they (not all) expect the game to be KSP1 after all

#

They clearly dont understand that the game was rebuilt in a better way

#

But yeah ur right 😔😔

north mist Jul 15, 2023, 12:03 AM

#

By the way I've been thinking about how to best handle the patching of 3rd party mods

#

It's not a good idea for every mod, even if it only has one config file that could be patched, to have to have it in an addressable bundle, but for now that's the only way Patch Manager would find it

#

So I'm thinking we should still keep using the addressables and resource location systems but get rid of the bundles part - it could be as simple as adding a special flow action for each mod that would work similarly to the flow action for loading addressables, but it wouldn't add the ResourceLocator from a bundle catalog, but instead a custom ResourceLocator and ResourceProvider used similarly to how Patch Manager already does, to serve JSON files using the Addressables system from the disk instead from bundles

#

And Patch Manager should then not even need any special treatment for most of those, and would be able to patch any arbitrary mod JSON

#

Then it you wanted to make a mod that makes use of a library mod, your mod could register its JSON files' ResourceLocator such that they all have a label that the library mod then loads to get all configs from all the mods making use of it/extending it, already patched

upbeat hare Aug 16, 2023, 11:41 PM

#

What do we need to do for release?

rapid parrot Aug 24, 2023, 12:37 AM

#

Can I make a request for the filtering capabilities that I can't seem to find in MM for ksp1.

The ability to filter things in an inverted way.

so kinda like

@PART[*]:HAS[@MODULE[ModuleEnginesFX]:HAS[@PROPELLANT[ElectricCharge],!PROPELLANT[*]]]

and that would ONLY find engines that have ElectricCharge but NO secondary propellant.
atm this one doesn't work i'm guessing because its finding the ElectricCharge one as part of the * too.

and while I can make it work by listing EVERY other type of fuel it is VERY annoying to have to do so.

upbeat hare Aug 24, 2023, 12:41 AM

#

Hmm, there is already a way to do that under what I am working on, but its syntax is completely different, basically you'd have to do a conditional statement inside the block and check that the length of the amount of propellants is 1

rapid parrot Aug 24, 2023, 12:41 AM

#

ahh nice. then all good

#

🙂

north mist Aug 24, 2023, 3:02 PM

#

@upbeat hare do you still have that language specification document somewhere?

upbeat hare Aug 24, 2023, 3:04 PM

#

I do yes, I'm gonna retitle it to remove sassy from the name, as I don't want to run afoul of trademarks I might not know about, then I'm going to have to update it quite a bit

north mist Aug 24, 2023, 3:05 PM

#

Alright, that would be awesome, thanks

upbeat hare Aug 24, 2023, 3:06 PM

#

north mist Alright, that would be awesome, thanks

What should I use instead of Sassy?

north mist Aug 24, 2023, 3:06 PM

#

Then we can start putting together some examples of configs and patches in various formats like we were talking about, and do a bit of a survey

#

I mean we could always just use something like Patch Manager Language as an interim name

upbeat hare Aug 24, 2023, 3:08 PM

#

I need a few example patches to put in there

#

do you still have the crew capacity one

north mist Aug 24, 2023, 3:12 PM

#

I have these two:

:parts {
    @if $current["crewCapacity"] > 0 {
        crewCapacity: +10;
    }
}

:parts {
    * > ResourceContainers > * {
        capacityUnits: *50;
    }
}

#

but the second one never actually worked

#

even though I thought I copied it from you

upbeat hare Aug 24, 2023, 3:12 PM

#

Wait, it didn't work?

north mist Aug 24, 2023, 3:12 PM

#

not as far as I remember

upbeat hare Aug 24, 2023, 3:13 PM

#

I've tested it?

copper forum Aug 24, 2023, 3:13 PM

#

how would this look in your one? whats the conversion like? want to make a site that you can create them on.

north mist Aug 24, 2023, 3:13 PM

#

upbeat hare I've tested it?

well it might have changed the JSON but the game could have just used the original values from the prefab or something

#

there are some places in the game that we'll need to patch

#

not sure if this is one of them

upbeat hare Aug 24, 2023, 3:14 PM

#

north mist well it might have changed the JSON but the game could have just used the origin...

I think you are missing the initialunits part of the patch

#

:parts {
    * > ResourceContainers > * {
        capacityUnits: *5;
        initialUnits: *5;
    }
}

upbeat hare Aug 24, 2023, 3:15 PM

#

copper forum how would this look in your one? whats the conversion like? want to make a site ...

I had a patch for something similar to that, I just forgot where I stored it, give me a moment

upbeat hare Aug 24, 2023, 3:21 PM

#

copper forum how would this look in your one? whats the conversion like? want to make a site ...

I haven't written a specific adapter for Celestial Bodies yet, so it does require some conditionals for the moment, but I can fix that easily

:json celestial_bodies > data {
  @if $current["bodyName"] == "Mun" {
    rotationPeriod: 800;
    isRotating: true;
    isTidallyLocked: false;
  }

  @if $current["bodyName"] == "Tylo" {
    rotationPeriod: 500;
    isRotating: true;
    isTidallyLocked: false;
  }

  @if $current["bodyName"] == "Duna" {
    rotationPeriod: 400;
    isRotating: true;
    isTidallyLocked: false;
  }
}

#

Once I do that specific adapter it would look closer to

:bodies {
  #Mun {
    rotationPeriod: 800;
    isRotating: true;
    isTidallyLocked: false;
  }
  #Tylo {
    rotationPeriod: 500;
    isRotating: true;
    isTidallyLocked: false;
  }
  #Duna {
    rotationPeriod: 400;
    isRotating: true;
    isTidallyLocked: false;
  }
}

or

:bodies #Mun {
  rotationPeriod: 800;
  isRotating: true;
  isTidallyLocked: false;
}
:bodies #Tylo {
  rotationPeriod: 500;
  isRotating: true;
  isTidallyLocked: false;
}
:bodies #Duna {
  rotationPeriod: 400;
  isRotating: true;
  isTidallyLocked: false;
}

north mist Aug 24, 2023, 3:24 PM

#

now I'm interested (although I should probably know this), do we have a way to for example easily copy a body and just change some parameters, like its orbit?

#

it would probably require some game patches, but I mean more so if the syntax allows it

upbeat hare Aug 24, 2023, 3:27 PM

#

Orbits are changed in the galaxy definitions which have been giving me trouble

#

But, I'm trying to think of the syntax for that stuff, it shouldn't be too hard to come up with

#

@use "builtin:list";
:json GalaxyDefinition_Default > CelestialBodies > OrbitProperties {
    semiMajorAxis: $value*2.5;
}

@function ComputeObjectData($localSimObjects) {
    @return $localSimObjects:map(
        @function($object) {
            @if ($object["RelativeTo"] == "") {
                @return {
                    "Name": $object["Name"],
                    "RelativeTo": "",
                    "ReferenceFrame": $object["ReferenceFrame"],
                    "LocalPosition": {
                        "x": $object["LocalPosition"]["x"]*2.5,
                        "y": $object["LocalPosition"]["y"]*2.5,
                        "z": $object["LocalPosition"]["z"]*2.5
                    },
                    "LocalRotation": $object["LocalRotation"],
                    "FixedGuid": $object["FixedGuid"]
                };
            } @else {
                @return $object;
            }
        }
    );
}

:json celestial_bodies > data {
    radius: $value*2.5;
    MinTerrainHeight: $value*2.5 if $value != null else null;
    MaxTerrainHeight: $value*2.5  if $value != null else null;
    TerrainHeightScale: $value*2.5 if $value != null else null;
    atmosphereDepth: $value*2.5 if $value != null else null;
    inverseRotThresholdAltitude: $value*2.5 if $value != null else null;
    //gravityASL: $value*2.5 if $value != null else null;
    oceanAltitude: $value*2.5 if $value != null else null;
    LocalSimObjectsData: ComputeObjectData($value) if $value != null else null;
}

This patch for example is what was giving me a lot of trouble w/ scaled assets and stuff

#

but thats a very non trivial patch as well lol, using the functional programming stuff

#

But back to the question
The problem with copying/adding new parts/bodies, is that there is a lot of nontrivial stuff that goes into them

#

And a lot of information outside of just the json

north mist Aug 24, 2023, 3:34 PM

#

yeah, absolutely, I mean, that definitely doesn't need to be a part of the initial release, more like something to think about for later

upbeat hare Aug 24, 2023, 3:34 PM

#

Like youd need to find the parts prefab and copy it ... which needs to be cached somehow

#

At least planets have their asset keys baked into the json

north mist Aug 24, 2023, 3:35 PM

#

quite a while back I was talking with Lux in vc and I feel like we figured out pretty well how to patch most of the stuff needed to be able to make a part just with a JSON file, a mesh and textures

#

though it will require a lot of Harmony patching

upbeat hare Aug 24, 2023, 3:36 PM

#

At that point our load time issues will be tenfold

#

Speaking of which, my thoughts are, we have a list of commonly used labels (corresponding to the rulesets we have), then patch files can define any labels they want to patch outside of that list explicitly

north mist Aug 24, 2023, 3:38 PM

#

yeah, I think that would be a reasonable compromise

upbeat hare Aug 24, 2023, 3:38 PM

#

I can easily write something like that :3

#

But I have class soon

north mist Aug 24, 2023, 3:39 PM

#

that's fine, I mean this thing is not running away anywhere, we've been sitting on it for months and months, so no point in rushing it now 😅

upbeat hare Aug 24, 2023, 3:40 PM

#

I'll just write it in class, better than going over taylor polynomials ... something I already know

north mist Aug 24, 2023, 3:41 PM

#

specifically, the idea of this has been hovering around since half of March lmao

#

from what I could find

upbeat hare Aug 25, 2023, 3:13 PM

#

@north mist what looks better for defining labels to be loaded

@load 'x', 'y', 'z';
// or
@patch 'x', 'y', 'z';
// or
@labels 'x', 'y', 'z';

north mist Aug 25, 2023, 3:15 PM

#

I think @patch is the most easily understandable one

upbeat hare Aug 25, 2023, 3:41 PM

#

alright, @north mist the commit to the label_declarations branch should have that system
by default, since there is only one specific ruleset defined, parts_data is the only label thats patched, but rulesets can declare in the attributes certain labels that should be added to the always patch list, and patches can also declare labels

#

Once I make a celestial body ruleset as well, then that will be added to the always patch list

#

Now I need to just quickly test it

#

Alright, loading time was a lot quicker, though something didn't work ... hmm

north mist Aug 25, 2023, 3:50 PM

#

upbeat hare alright, <@366972069440913418> the commit to the `label_declarations` branch sho...

alright, that sounds great!

upbeat hare Aug 25, 2023, 3:50 PM

#

[LOG 11:47:18.069] [System] Patch Manager: parts_data completed in 6.4749s. gosh 7 seconds to go over every single part though

#

[Info : Console] Testing: resourceContainers against ResourceContainers ahh this was the error

#

Capitalization

#

Thats why that patch you were trying to test with didn't work

#

It might be a bit quicker if we take out a lot of the debug logging I did

#

[Info : Console] Testing: resourceContainers against resourceContainers -> True
Aight cool, so that part works

#

[Info   :   Console] Patch of parts_data:adapter_3v_m3_rightcone_methalox_2v-m3 errored due to: PatchManager.SassyPatching.Exceptions.InvalidVariableReferenceException: C:\Program Files (x86)\Steam\steamapps\common\Kerbal Space Program 2\BepInEx\plugins\PatchManager\patches\test1.patch:3:23: $value does not exist in current scope
  at PatchManager.SassyPatching.Nodes.Expressions.Unary.Implicit.Compute (PatchManager.SassyPatching.Execution.Environment environment) [0x00024] in C:\Users\arall\PatchManager\src\PatchManager.SassyPatching\Nodes\Expressions\Unary\Implicit.cs:35 
  at PatchManager.SassyPatching.Nodes.Statements.SelectionLevel.Field.ExecuteOn (PatchManager.SassyPatching.Execution.Environment environment, PatchManager.SassyPatching.Interfaces.ISelectable selectable, PatchManager.SassyPatching.Interfaces.IModifiable modifiable) [0x000f2] in C:\Users\arall\PatchManager\src\PatchManager.SassyPatching\Nodes\Statements\SelectionLevel\Field.cs:59 
  at PatchManager.SassyPatching.Nodes.Statements.SelectionBlock.ExecuteOnSingleSelection (PatchManager.SassyPatching.Execution.Environment environment, PatchManager.SassyPatching.Interfaces.ISelectable selection, PatchManager.SassyPatching.DataValue parentDataValue) [0x00077] in C:\Users\arall\PatchManager\src\PatchManager.SassyPatching\Nodes\Statements\SelectionBlock.cs:145 
  at PatchManager.SassyPatching.Nodes.Statements.SelectionBlock.ExecuteOn (PatchManager.SassyPatching.Execution.Environment environment, PatchManager.SassyPatching.Interfaces.ISelectable selectable, PatchManager.SassyPatching.Interfaces.IModifiable modifiable) [0x000e8] in C:\Users\arall\PatchManager\src\PatchManager.SassyPatching\Nodes\Statements\SelectionBlock.cs:123 
  at PatchManager.SassyPatching.Nodes.Statements.SelectionBlock.ExecuteFresh (PatchManager.SassyPatching.Execution.Environment snapshot, System.String datasetType, System.String name, System.String& dataset) [0x000af] in C:\Users\arall\PatchManager\src\PatchManager.SassyPatching\Nodes\Statements\SelectionBlock.cs:66 
  at PatchManager.SassyPatching.Execution.SassyTextPatcher.TryPatch (System.String patchType, System.String name, System.String& patchData) [0x00001] in C:\Users\arall\PatchManager\src\PatchManager.SassyPatching\Execution\SassyTextPatcher.cs:51 
  at PatchManager.Core.Assets.PatchingManager.PatchJson (System.String label, System.String assetName, System.String text) [0x00034] in C:\Users\arall\PatchManager\src\PatchManager.Core\Assets\PatchingManager.cs:54

#

[Info   :   Console] Trying to find capacityUnits in {
  "name": "SolidFuel",
  "capacityUnits": 2.81,
  "initialUnits": 2.81,
  "NonStageable": false
}

Okay so why does $value not exist?

#

no, $value exists ???

#

Ohhh, I did a dumb

#

and I also amasked an error

#

the real error is I forgot to implement Real and Integer implicit multiplication

#

meaning that just woulda workd if it was *50.0

#

I still wish we could shunt patching to a completely separate thread altogether and have the game not freeze during it so I can get an actually accurate patch counter

#

Anyways @north mist I fixed that test patch ... it was a combination of improper capitalization in the patch, and my own dumbassery

north mist Aug 25, 2023, 4:22 PM

#

ahhh awesome!

upbeat hare Aug 25, 2023, 4:29 PM

#

So, now that we have solved the load time issue?

maiden wraith Aug 25, 2023, 4:38 PM

#

release? :O

north mist Aug 25, 2023, 5:01 PM

#

so do we want to release this as is and then do the survey/discussion about the config/patching formats later?

maiden wraith Aug 25, 2023, 5:04 PM

#

this way people can experiment with it

#

and not just have an unexperienced opinion

north mist Aug 25, 2023, 5:05 PM

#

I mean

#

the things we wanted to ask about, they can't really experiment with either way

#

because I don't think that PM in its current state supports creating new configs from scratch

upbeat hare Aug 25, 2023, 5:08 PM

#

It does not yet ... but im starting to have an idea

#

But like how do you create like a new mesh if you copy a part config ... or do you just have to have the mesh/prefab in the bundles already

#

And if you create a part ... you have to add that to the list of parts to be patched at the end

north mist Aug 25, 2023, 5:11 PM

#

I mean, that should be solved by the dependency order

#

also, I think that specifically the creation and copying of parts should be a goal for a later point, since that will be one of the most complex things you could possibly do with PM

#

I meant more so the creation of completely new custom configs, like for example Lux being able to create a config file for LFO with Patch Manager

upbeat hare Aug 25, 2023, 5:13 PM

#

That should be easier to implement

#

Thoughts on a syntax like this?

:+lfo("name") {
}

north mist Aug 25, 2023, 5:14 PM

#

and then other mods should be able to specify they need LFO to be processed first, and then they could patch the LFO default config

north mist Aug 25, 2023, 5:15 PM

#

upbeat hare Thoughts on a syntax like this? ```scss :+lfo("name") { } ```

as in, create a new asset called "name" with the "lfo" label?

#

or with a completely custom LFO adapter?

#

that's another thing I wanted to discuss

upbeat hare Aug 25, 2023, 5:16 PM

#

north mist as in, create a new asset called "name" with the "lfo" label?

Essentially, the semantics are slightly different being what you just said as an adapter

#

Something like

:+json("label","name") {
}

Would be generic

north mist Aug 25, 2023, 5:17 PM

#

ah gotcha

upbeat hare Aug 25, 2023, 5:17 PM

#

And any adapter can specify a list of parameters they take for creation

north mist Aug 25, 2023, 5:17 PM

#

yeah, I think that could work

#

although, what's our syntax for deleting?

#

if we have one

#

I really sound like I've never seen this project before in my life lmao

#

I haven't really kept up with the syntax tbh

upbeat hare Aug 25, 2023, 5:19 PM

#

@delete; in the body of the selection block, though I havent yet added that fully in the engine

north mist Aug 25, 2023, 5:19 PM

#

I'm just wondering if we should maybe make those two kinda similar, since they mirror each other

upbeat hare Aug 25, 2023, 5:20 PM

#

One requires a lot more information than the other

north mist Aug 25, 2023, 5:21 PM

#

also, just to make sure I understand correctly, @delete is only used to delete the top-level object, aka the whole asset, or other parts of the hierarchy as well, such as individual modules?

upbeat hare Aug 25, 2023, 5:22 PM

#

Can delete anywhere in the hierarchy

north mist Aug 25, 2023, 5:22 PM

#

and do we have anything for adding things within the hierarchy?

upbeat hare Aug 25, 2023, 5:23 PM

#

+element_type/name at the end of a selection statement

north mist Aug 25, 2023, 5:23 PM

#

(hope I'm not being annoying with all these questions, just want to make sure I have the full picture)

upbeat hare Aug 25, 2023, 5:23 PM

#

upbeat hare `+element_type/name` at the end of a selection statement

(Also not fully handled yet, but the interfaces are there)

north mist Aug 25, 2023, 5:23 PM

#

alright, then I guess :+adapter makes total sense

upbeat hare Aug 25, 2023, 5:25 PM

#

We need a way to specify in the cache that an asset has been deleted

north mist Aug 25, 2023, 5:25 PM

#

we'll need some very extensive docs lmao

maiden wraith Aug 25, 2023, 5:30 PM

#

i'd love something more

#

complex? i think

#

like

#

lfo:action

north mist Aug 25, 2023, 5:30 PM

#

you can just do that in the body

maiden wraith Aug 25, 2023, 5:30 PM

#

lfo:plume{
}
lfo:frost{
}

#Patch Manager