CommNext

[thin mauve]

yup, native

[tepid dust]

And this is the mono dll?

Aaaa

Ill look more when home

[verbal leaf]

it would seem so

[tepid dust]

Wheres the internal calls source

[verbal leaf]

https://github.com/dotnet/runtime/blob/d4bab6ba129819657b60732e36b0c18b5d765c3b/src/coreclr/pal/inc/pal.h#L4258

no clue what Pal is

ah Platform Adaptation Layer

[tepid dust]

I meant in monoo

[verbal leaf]

sorryy

[verbal leaf]

just the C math.h standard library

https://en.cppreference.com/w/c/numeric/math/fma

seems to just be a direct call to this

[tepid dust]

Which should do what it says on the tin with msvc

[thin mauve]

Thanks, Rider.

But maybe this is the hint I misconfigured the project?

[tepid dust]

Nah

Fma just dont exist

Idk

[thin mauve]

Ok in the end I just lost my patience and compiled a native C dll

[tepid dust]

Lmao

The nuclear option

[thin mauve]

I'm sure this mean handling a shittruck of problems

but

it seems to be working

let me see

[thin mauve]

for... THIS

#include "compute.h"
#include <math.h>

double FusedMultiplyAdd(double a, double b, double c)
{
    return fma(a, b, c);
}

The good thing: it is working
The bad thing: it is not solving the issue

[tepid dust]

Well is there still roundoff error in the discriminant

Or further down the line

What about under/overflow

But at least it didnt return 0

Do we have to go the full nuclear solver

Honestly

Just write the solver in C at this point

[thin mauve]

I'm leaning toward this in no time

I didn't think it was possible to link C code directly

[tepid dust]

I kinda suggested it earlier

When I said write it in C++

[thin mauve]

You're right, but I thought it would be a looong task

[tepid dust]

Burst would be preferable but I have to look into that

So just native for now

[thin mauve]

I have thought about that

but to Burst compile jobs (but I'm not sure I understood correctly) you need to compile it inside Unity, and there is the issue that you cannot reference the KSP Assembly in unity

At least this is what I understood

Ok, so. The discriminant is fixed by FMA now, but I'm getting -5.2802735331476427E+32 which is.. huge. E+32.
I'm seeing i'm calculating the c with s.x * s.x + s.y * s.y + s.z * s.z, so squaring the body coordinates. This is a big no-no, same FP issue. Will try to recenter all coords on s so we have to deal only with relative positions

[tepid dust]

This is why we need to make an equivalent of KSPBurst

[thin mauve]

Omg

sounds really cool

[tepid dust]

Yeahh

https://github.com/KSPModdingLibs/KSPBurst/tree/master

is what I'm referring to

[tepid dust]

this is exactly what thunderkit is for by the way

referencing the KSP assembly in unity

building it is the hard part

for now just write it in C or C++

translating it later will be easy

[thin mauve]

Sounds good

by the way.. nothing

[tepid dust]

from anything?

[thin mauve]

Discriminant is now in the -E+32 range

so I suppose it could be a different issue

[tepid dust]

Yeah

[thin mauve]

it could be the local positions (they are very far from KSC), but still doesn't make sense. We are talking about E+10/E+11 , and this should cover a meter

even with 1000meters error, it should be seen as occluded

since the radius is 500k

[verbal leaf]

Am I reading it right that in order for users to be able to use it, they have to separately download and install Mono?

[tepid dust]

ah wait

it does?

[verbal leaf]

If it's the case, then it's definitely better for the modder to just pre-compile it in Unity

[thin mauve]

What a nice bug

Ah one thing @tepid dust @verbal leaf : I'm getting this error while loading the Native dll:

[Error  :SpaceWarp.Preload] An error occurred while processing C:\Program Files (x86)\Steam\steamapps\common\Kerbal Space Program 2\BepInEx\plugins\CommNext\swinfo.json:
System.BadImageFormatException: Format of the executable (.exe) or library (.dll) is invalid.
  at Mono.Cecil.PE.ImageReader.ReadOptionalHeaders (System.UInt16& subsystem, System.UInt16& dll_characteristics) [0x000bb] in <6034b380a22b41a596c9dc29d282c0a9>:0 
  at Mono.Cecil.PE.ImageReader.ReadImage () [0x0008b] in <6034b380a22b41a596c9dc29d282c0a9>:0 
  at Mono.Cecil.PE.ImageReader.ReadImage (Mono.Disposable`1[T] stream, System.String file_name) [0x00007] in <6034b380a22b41a596c9dc29d282c0a9>:0 
  at Mono.Cecil.ModuleDefinition.ReadModule (Mono.Disposable`1[T] stream, System.String fileName, Mono.Cecil.ReaderParameters parameters) [0x00006] in <6034b380a22b41a596c9dc29d282c0a9>:0 
  at Mono.Cecil.ModuleDefinition.ReadModule (System.String fileName, Mono.Cecil.ReaderParameters parameters) [0x0006c] in <6034b380a22b41a596c9dc29d282c0a9>:0 
  at Mono.Cecil.ModuleDefinition.ReadModule (System.String fileName) [0x00007] in <6034b380a22b41a596c9dc29d282c0a9>:0 
  at Mono.Cecil.AssemblyDefinition.ReadAssembly (System.String fileName) [0x00000] in <6034b380a22b41a596c9dc29d282c0a9>:0 
  at SpaceWarp.Preload.Entrypoint.AddPatchersFromDLL (System.String dllPath) [0x00000] in <8cf50e6a121d4dc6832239a4a791d944>:0 
  at SpaceWarp.Preload.Entrypoint.AddEnabledPatchers () [0x00084] in <8cf50e6a121d4dc6832239a4a791d944>:0 

i suppose it's scanning my native dll and doesn't find anything readable, is there something I could do that you know?

[tepid dust]

how are you compiling it

wait

wtf

its trying to load the swinfo.json as a dll

that's the error

ah

no I'm dumb

[verbal leaf]

it's looking for the preload API uses in all DLLs

right?

[thin mauve]

I suppose so

[tepid dust]

yeah

[verbal leaf]

we should just try/catch that tbh

[tepid dust]

this might require a sw hotfix

if you wanna do that rq munix

[thin mauve]

Don't worry

[verbal leaf]

I'm off to bed rn

[thin mauve]

I'm nowhere near a solution for this bug

[tepid dust]

same basicallyyy

[thin mauve]

I'd like to sleep

but

this

[verbal leaf]

it can wait until tomorrow

[thin mauve]

bug

[thin mauve]

Well, another unresolving night

We'll see!

[safe lichen]

Here's my idea on a switching algorithm: for short distances, I propose to keep your current algo, but for long distances we can make an assumption that makes the computational much more robust, when we assume that all rays from the orbit to the target are parallel. Now, it requires a threshold when to switch, but I'm sure we can find a reasonable one.

[safe lichen]

(and feel free to ignore me, I would implement it directly but I've never programmed in C# nor in unity so I feel a bit useless, but maybe I can help on the math side)

[shadow gyro]

Don't know what's happening. Everything was working ok, then I launched a new vessel and now none of my vessel have a connection. Understandable for those on Duna and Jool with lower gain antennas, but this one in GKO should have a normal connection. Can I do something more to troubleshoot this?

Also, setting different bands, does that do anything right now or you haven't implemented it yet?

Oh wait, of course, KSC went to the dark side of my relay.... aaaah, nice 🙂

Hmm, but still... I just launched an interplanetary ship with an RA-100 antenna, and I checked - KSC rotated in a way that it's looking straight at my ship, no occlusion. Shouldn't a connection be possible in this case? Or does KSC have a small gain antenna so it isn't reaching my ship, so a relay in LKO is needed? Sorry, I wasn't following this thread so I'm out of loop a bit 🙂

[thin mauve]

Hi @shadow gyro ! Definitely I need to add more ways to debug the commNet status 😄 .. anyway, KSC has a low 2G antenna, so it is possible that it doesn't reach a RA-100 even if the RA-100 sphere "includes" KSC

[thin mauve]

This happens all the times and I still get that "..oh." moment 😄

[thin mauve]

Thanks for your effort and your suggestions! I didn't understand what you meant last time, now it is a bit clearer; the thing I'm missing here is, where is the vessel2 vector? We need to check if the body is between two vessels

(I just don't get the math in the drawing, so I'm asking you for clarifications)

[shadow gyro]

Ah so the KSC has a 2G antenna, that explains it then. But could you buff it up somehow - can it be changed with Patch Manager? It doesn't make sense that this thing has the same gain as a measly RA-2 antenna 😆

KSP1 has (or is it RSS/RO? I forget...) the DSN. 3 points on the planet that can connect to anything in the solar system, given enough gain on the receiving side. Maybe a 3 point DSN is for another mod, but it would be nice if just the KSC one could be buffed, at least to the equivalence of the RA-100

[safe lichen]

\vec{t} is pointing to the target (e.g. vessel 2). The key idea here is that when two objects are very far away (e.g. vessel1 in orbit around laythe, and vessel2 in orbit around kerbin) then their position on their respective orbits has little influence on the direction vector between them. So we can compute occlusion by assuming the direction vector (\vec{t}) the same no matter where we are on the orbit (or no matter where the target vessel is on it's orbit). This simplifies the computations a lot and removes the need to solve an ill-conditioned quadratic equation.

Now, when the vessels are close, say orbit around the Mun and orbit around Kerbin, this no longer works and you need to take their arguments of periapsis into account. I think a reasonable cutoff threshold would be something like if (distance / R_orbit >> 1)

[shadow gyro]

We have a DSN in place 🙂

[tepid dust]

A custom DSN

[shadow gyro]

But still, the DSN and my vessel in Dres's orbit aren't connecting. In the screen: orange circle = DSN I just launched, red circle = my vessel in Dres orbit. They aren't connected. There's no occlusion I'm aware of, they're looking at each other without anything between them.

[lusty pollen]

just saw this in CKAN and came here to say thankyouthankyouthankyouthankyou!!! that's all

[lusty pollen]

i started a new campaign because of this but for some reason the first mission isn't being completed - only new mod was this one...is there some way it's preventing that mission from being accomplished?

(reach 10k feet with a rocket from kerbin)

[storm rune]

@thin mauve Hey! It could be the new Patch Manager release. I literally had the same problem after I updated it. On the new version of patch manger (0.10) I found our that relays do not show bands configuration anymore. I reverted back to Patch Manager v 0.9.4, invalidated all the cache and everything is working flawlessly.

[tepid dust]

That message was before the PM release?

Nothing changed in the latest release that shouldve broke anything here

[storm rune]

My CKAN shows that PM released today at 14:54 CET. Anyway, with the PM update (v0.10) 70% of my vessels lost connections and did not want to reconnect.🤷‍♂️

[verbal leaf]

yeah, almost an hour after Falki's message

[tepid dust]

Alright, maybe, but I doubt it because this mod uses none of the syntax I did change

[storm rune]

Well, let's see what @shadow gyro replies. If he had the same issue and my suggestion helped him than its fantastic.

[tepid dust]

Falki coulrnt have been using PM 0.10.0 is the thing

[storm rune]

@tepid dust @verbal leaf have you tried PM 0.10.0 with CommNext 0.4.0? Maybe I need to reinstall and empty the cache for PM 0.10.0 or I just have a weird bug🤣

[tepid dust]

I can test ... not today tho, Im not home for the rest of the day

[storm rune]

Guys, thank you so much for developing mods and making games better❤️

[lusty pollen]

might be the PM thing. I also reverted to 0.9.4 and now the mission works 🤷‍♂️

[tepid dust]

Alright all this means is falki has a different bug

I know the pm bug

[tepid dust]

New release coming should at least fix this bug

[verbal leaf]

and done

should be fixed in 0.10.1

though Falki's bug is not related to this

[shadow gyro]

I was on Patch Manager 0.9.4. and all the bands were available.

[brazen oasis]

FYI I don't think bands are implemented, just a placeholder for them to choose them on a part. @thin mauve was going to implement them soon but then got sidetracked by the bug with the occlusion check. But maybe you were just mentioning bands not being available because it implied that the CommNext patch scripts weren't running.

With the occlusion check having trouble (possibly because of massive coordinates and float precision issues) that might be @shadow gyro's bug too. I'd wait for the dust to settle on that before reporting any issues that could be due to a faulty occlusion check.

[fading fulcrum]

I have the same problem with the new Patch Manager, relays don't have band configuration anymore.

[verbal leaf]

you need to update to 0.10.1 and delete cache

Kerbal Space Program/BepInEx/plugins/PatchManager/cache

[fading fulcrum]

give me a second, updating right now

thx alot, my kerbals are online

[shadow gyro]

When I left Dres's orbit, I got a connection. So some occlusion was detected while I was still in Dres's SOI

[brazen oasis]

Hmmmm... I wonder if there is another, separate issue from the occlusion issues having to do with science being able to connect. I seem to remember hearing about a bug in vanilla that sometimes you could use probe control via a relay, but couldn't upload science via relays. It never caused me grief before because I would always have a big enough antenna with the Communotron 88-88 to call home directly to KSC. However, now with this mod we're forced to use relays because of the nerfed KSC and occlusion. Does anyone know if this bug is still a thing in vanilla? If so, it's a show stopper for this mod (unless we can some how fix it)

[brazen oasis]

Hmmmm... I think the problem has been fixed in vanilla but it is a problem with the mod, and I don't think it is due to the occlusion bug. I've only had the occlusion bug when close to a moon such as Laythe in the Jool system, but not when far out where my higher powered relays are. However, even after teleporting my vessel right next to one of the relays I can't upload my science. Here's my mk2 vessel which has a Com-88-88 on board but it isn't open. It also has integrated transmitters in the Pod and a probe core, so it obviously should be able to reach the relay right next to it.

You can see the relay is connected via map view, and both vessels look like they are connected with a green icon below the compass

However, with the mod, I can't upload science. See "No Connection", both before and after I open the Com-88-88:

Removing the mod allows uploading science without using the Com-88-88:

So something seems broke in using relays for science specific to this mod

Still without the mod, I teleport to low Laythe orbit. Now I can't upload the science which is expected because I'm too far from the relays even with vanilla ranges:

but opening the Com-88-88 at this point let's me upload

So... with vanilla, uploading science works through relays. With CommNext currently, it doesn't appear so. And I think this is probably separate from the occlusion issues, because it happened while right next to a relay that wasn't experiencing this issue. Save game below

I hope this was useful, and I'm sorry to be the bearer of bad news. This is hard stuff!

[thin mauve]

Hi everybody! An hard day at job, plus I got the flu, what a wonderful combination; I'll try to make some progress but a bit slower these days probably. Now, let me reply to all these great feedbacks!

[thin mauve]

I didn't implement it with Patch Manager, but for sure I can expand it; I'm thinking about 50G (the equivalent of DSN2 in KSP1), but I'm open to feedback

[verbal leaf]

I don't expect it's something that can be done with Patch Manager

[thin mauve]

Well this is really not true.. if I'm behind the body, orbiting it, even if other vessel is faaar away, I'm still going to be occluded every time I go on the dark side; plus, the direction vector when you're on the "light" side, is technically the same as when you are in the dark side, but the difference is exactly that!

-- [S1] --- ( Laythe ) --------------------- [Kerbin]
---------- ( Laythe) -- [S1] --------------- [Kerbin]

Am I missing something else?

[thin mauve]

This could be an issue from the low altitude. I'm actually using the body radius which should be lower than every possible altitude iirc, but maybe the algorithm is being a little picky here and it thinks you are "inside" Kerbin 😄 . Could you send me the file? Should be easy to fix (4 days later...)

[thin mauve]

Oh, interesting; I've seen there were some things checked in a "custom" way, dedicated to Science Tranmission; I'll double check side effects with the new algorithm / new relays

[thin mauve]

Thanks for your words and the help you've been giving in this thread. Hard stuff for sure, but I'm happy for all the support from the community!

[obsidian solstice]

Do you have any plans for this mod if CommNet makes it into vanilla? Just curious

[thin mauve]

Hi @obsidian solstice ! Technically CommNet is already in vanilla; I'm patching the CommNetManager for example 😄 ; I heard that devs do not want to implement occlusion or to nerf ranges, so this is the reason why I started the mod, but maybe I am misunderstanding your words?

[obsidian solstice]

Im referring to a proper occlusion and relay system

[tepid dust]

I think the devs qualified that saying without the proper tools or viz

[brazen oasis]

Did they ever do that in KSP1?

[obsidian solstice]

They didn't say they were not going to implement it, they said that it wasn't a priority as only a select part of the userbase will use it

KSP 1 does have occlusion yes

[thin mauve]

Oh, well, if they include it in stock I think I'll just keep the "delta" or whatever is missing from this implementation; but given how many things they have to do, this is a nice thing to leave on the corner

no?

[brazen oasis]

definitely!

[thin mauve]

Plus.. I wanted it now! 😄

[thin mauve]

Yep exactly that!

[obsidian solstice]

I still don't really like how they responded to this, its very weirdly worded to me, but at least we have it as a mod now

[shadow gyro]

Here's one of the saves from a bit later after I wrote that. I launched a new DSN because that first one was a bit embarrassing to look at. It's landed on top of the VAB, with direct line of sight to Dres. Vessel "Eye of Destiny" is in orbit around Dres and also with a direct line of sight to Kerbin. Kerbin's DSN shows in orange, but Dres's vessel is in red, no connection.

[thin mauve]

I'm not really worried about this; mods in KSP1 often improved base game! Think about robotics, EVA construction, etc.

[thin mauve]

I'll debug it as soon as I finish with the damned occlusion bug, thanks @shadow gyro !

[obsidian solstice]

They're somewhat implying that KSP 1's system was so forgiving as to be pointless, but my experience with it was very contrary to that

[shadow gyro]

They said they're thinking about revisiting occlusion sometimes later. And I agree, I don't think this mod will be obsolete ever.

[obsidian solstice]

Likely not yeah

[thin mauve]

I share this. "forgiving" is something a really seasoned player could say, but still a nice planning part

Sooo, general announcement: I'm tracking all your reports here, since I'm worried I can lose something in the meantime: https://github.com/orgs/Kerbalight/projects/5/views/1?layout=board

I've written there some parts of your message and the reporter name, so I can remember better, I hope this is not a problem for you and your privacy; if it is, let me know and i'll remove your username from the GitHub Issue

Plus I hope having a tracker could help give you all a rapid summary of what's going on!

[safe lichen]

Exactly, the "dark" and "light" side are defined by the direction vector between vessel1 and vessel2, and the radius of each vessel and it's orbit. The difference with this faaaar away algo is that this direction vector is considered constant for a given orbit around the CB. As such the occlusion determination becomes a simple projection of the vessel vector (defined from the CB origin) to the plane defined by the direction vector (t). The idea is similar to why we consider sunlight to be parallel on earth, and the earth terminator to be it's diameter, but instead of the sun, you consider vessel2. I'll take some time to write it out in more detail.

[thin mauve]

Oh, now that you say it that way it completely made sense

[thin mauve]

So it would be something like that? Sorry for v_p, I don't really remember the vector to plane decomposition formula

(A, B are the sats)

where v_p is AE anyway

[safe lichen]

yes, instead of projecting along the line BD to find occlusion (requiring the quadratic solver) you project along the green dotted line. If the bodies (A and C) are far enough away the difference is tiny, but the computation much more robust (and faster)

[thin mauve]

Ok, I need to bang my head on the math but seems feasible

[thin mauve]

I tried to debug it further, so I drawed (without scaling nor positioning, so don't look at Kerbin: just look at the white sphere which is Laythe and the purple line, which is the two vessels connection line seen from the algorithm): the main issue I see is that.. it's way off! I get it could be FP errors but, plus this is drawn at a very reduced scale, so not representative, but.. so much delta?

sometimes it gets back onto the white sphere

oh my god

time

the problem.. I think.. is that the planet moves between the moment I store the planet position

and the moment the game stores the vessel position

in fact if you time warp it gets worse

lol, this sounds like a Science Lab phrase

Ok, not the only issue. But surely one of the issues

[thin mauve]

Ha-ha! LateUpdate for the win, ladies and gentleman!

[brazen oasis]

woohoo!

[thin mauve]

OH GOD YES

Look at this beauty!

So smooth!

I'm really excited

[brazen oasis]

Yeah, sometimes the hardest tasks feel the best when you overcome them

How did you fix it?

[thin mauve]

I realized the Connections compute job was started in the unity Update function

and all the vessels positions were updated in another object Update function too

But for Unity Execution Order, it is not guaranteed their order; KSP2 has some utilities to handle the ordering, but in the end the simplest thing was using the LateUpdate method exposed by Unity, which is called only after all other Update methods have been computed; this way we are sure all bodies positions are calculated at the right moment

https://docs.unity3d.com/Manual/ExecutionOrder.html#UpdateOrder

In the end the issue was that in that small timestep, even without timewarp, bodies position changed by a huge margin (obviously.. space! 😄 )

[brazen oasis]

very cool! I'm surprised that it made it so off because things had been flashing so much and hadn't moved much during the flashing, but obviously

ah

yeah, these planets are moving very fast, and just the relative positions were moving slowly

[thin mauve]

Exactly!

This was the factor I didn't see until now

[brazen oasis]

so if you got Laythe's pos off by a few ms it would be a really big deal

very cool, glad you figured it out!

[thin mauve]

I'm happy too, this was starting to be a real pain 😄

[brazen oasis]

yeah, not so much fun when things aren't working. But now, it should really pick up!

[thin mauve]

I'm releasing it with this fix

That I feel was one of the worst

[tepid dust]

Are you still doing calculations in C

[thin mauve]

Yes

I'm wondering

if it's better to omit that

[tepid dust]

Profiling is what you need here

[thin mauve]

I love the added precision, but side effects could be devastating..

[tepid dust]

What side effects?

[thin mauve]

I compiled it only for 64bit Windows

[tepid dust]

I doubt you are really doing any pointer math no?

[thin mauve]

No no

nothing like that

more about distribution

[tepid dust]

Ksp2 only runs on 64 bit windows for the moment

Anyone playing on linux uses wine/proton

[thin mauve]

Sounds like the exact answer I was looking for

[thin mauve]

To check if the extern call is hurting performance, right?

[tepid dust]

Yes

Tho an extern call wouldnt be

Rather if the performance gain is that much

[thin mauve]

It's more about the ability to do FMA

[tepid dust]

Ah fair

[thin mauve]

it's just that I didn't find a way to do it inside C# without the extern dll

[tepid dust]

The extern call wouldnt be that bad

It wouldnt hurt to do the entire solving in c tho

[thin mauve]

Plus I should say that now that relay & ranges are in use

I had many ways to bail out calculations as early as possible

Execute took 0ms (nodes=13, numBodyOcclusions=593, numIntersections=8,connected=12/13,relays=12)

In this case even if the possible occlusions where 600+

only 8 has been calculated

[tepid dust]

I see

[thin mauve]

no wait, the discriminant has been calculated 600 times

[tepid dust]

Side note: Not that I care too much about rust
I wonder how hard c#/rust interop is

[thin mauve]

I love rust a lot

as much as I hate it

[tepid dust]

I like/dislike it at the same time

I no longer completely dislike it

[thin mauve]

haha

https://samrambles.com/guides/window-hacking-with-rust/creating-a-dll-with-rust/index.html#dllmain

Seems doable anyway

[tepid dust]

Rewrite all your code in rust then :3

[thin mauve]

hahah

not tonight!

[tepid dust]

Even your ui code

[thin mauve]

Even the github action

And then all KSP2

[tepid dust]

Even the dot net compiler

[thin mauve]

All praise the RUST LORD

[tepid dust]

Rewrite git in rust

[thin mauve]

And then rewrite rust in rust

(iirc it's already in rust, but doesn't matter)

Finally some decent logs

Execute took 0.0847ms

[tepid dust]

Fasterrrr

[thin mauve]

We have a baseline now!

I left it in ms

Until now

(I wasn't so optimist..)

[thin mauve]

v0.4.1 "Rocket Science" is out!

It includes this fix for the body occlusion bug, which deserved its own patch
https://spacedock.info/mod/3560/CommNext#changelog

I know I need to make the forum topic & the mod news thread.. time..

[safe lichen]

Very cool, so the problem was somewhere else, how typical. Visualization to the rescue. I'm wondering if @bronze anvil has similar issues in FP: that the stored positions are wrong cos they are at different moments in time.

[shadow gyro]

Ok, so for my Dres occlusion bug, it DOES work now (yay!), however with a weird behaviour. When you first load a save I sent you, it won't be connect, same as before. Then when you EVA a kerbal it then connects. When he reenters again, it's still connected. So it's as if something gets triggered when he first leaves the vessel. Overall nice update!

[bronze anvil]

I was wondering the same thing. When a Hohmann transfer results in an impact encounter at Mun that’s one thing, but when it results in falling short of reaching Minmus’s orbit at the Ap that’s another. Still, it’s worth considering

[thin mauve]

I'm a very visual guy, so the representation is always a fundamental thing for me 😄

Yes very different from first discoveries, a very intriguing one

[thin mauve]

Oh, good to know; I'll test it later, but to understand: does it keep connected even after save & reload? or it's something just for the current session?
Does it reset to unconnected when you switch to a far vessel?

[raw parcel]

Small feedback: tooltips for the map view buttons would be great to make it clearer what they do (unless I missed those)

[left obsidian]

What are all the Bands ? X S K Ka ? @thin mauve

[thin mauve]

Great Idea! I can bring them back from WW

thanks 😄

[shadow gyro]

Maybe it's best you take a look at yourself; I'm using the same save I sent you.
I'm testing it now and it seems that Eye of Destiny (Dres orbit) successfully connects when a kerbal leaves the vessel. It stays connected when he reenters. But if you quick save and reload, it will again be disconnected if the kerbal is inside. If he's outside when you quick save and load, it stays connected 🤷 . It doesn't make sense to me that it has anything to do with kerbals being aboard, but I can consistently reproduce this.

[thin mauve]

Actually, just a stub for customized connections: https://github.com/orgs/Kerbalight/projects/5/views/1?pane=issue&itemId=53173679

[thin mauve]

It's clearly a case of a Sus Kerbal

[raw parcel]

Will the different bands have varying characteristics? Like a range vs bandwidth compromise?

[thin mauve]

I'd love to think about something like that, but bandwidth right now is relevant only for science, so not sure if this could have an impact on gameplay

@shadow gyro 😆

do you think @raw parcel it could be still relevant?

[raw parcel]

Hmmm a more relevant gameplay feature would be making bands with higher ranges require more EC for transmission

[thin mauve]

this is interesting!

..we'll need a lot of planning tools, I fear..

so they could act like a base range multipler

[raw parcel]

Or even crazier idea, making some antennas directional with band having an effect on FOV

[shadow gyro]

You could unlock higher frequency bands with newer tech. That's how Real Antennas (KSP1) work afaik. I don't know if I'd be for that (might be too complicated to players), so I'm just stating it as a possibility.

[raw parcel]

Yeah you'd need to decide the level of realism / difficulty you want for this mod

It can go very far or stay pretty simple

[left obsidian]

that's a great idea

Without it everything would connect to each other without any sense

imagin your lunar lander signal comming back to the ksc while passing through your surface scanning sat or some random antena-equiped space debris

And talking about surface scanning can falki's orbital survey mod's antennas scan while transmit ? Caus i feel like if you use some end-of-techtree antenna to scan kerbin , it would comme with great range, and I feel like it shouldn't be able to act as a relay transmit your comm signal comming from jool.

[soft fiber]

If anything, I feel this could be a plugin/dependent on this mod, as I think directionality is just a difficulty grade too far. (or an option, but I think having it as a separate thing would be better)

[left obsidian]

why cant even lock rotation during time warp yet .

[thin mauve]

I'd like to reach a "medium" compromise. Nothing that you can't understand unless you study it too much; I'd be glad to offer plugins / API / compat with "Plugin" mods like @soft fiber says, but not for the base mod experience; I still feel this is KSP not Orbiter 😆

[thin mauve]

Bands could easily solve that

I'm leaning toward a more Band and channels approach

since Band is a nice gameplay feature on its own

while channels would be more appropriate for customization

I'm still a bit worried, just because I don't want to make it hard to setup your commnet

so probably at the beginning you can have only the "Default" X Band channel, "Default" UHF channel etc.

And then later you unlock multiple channels

or multiple bands

I'm a bit undecided: I would like it to be simple to use and understand, but I wouldn't want to limit players in building more advanced networks.

[raw parcel]

My main opinion on all of this is just: the main question to ask is "what is the incentive for players to use different bands and subchannels?"

[thin mauve]

Sounds like the right question

[thin mauve]

About channels: you don't want spaghetti networks, it has been a pretty common issue in KSP1, and different solutions emerged like Constellation or Remote Tech itself with cones

[raw parcel]

What's the main issue with spaghetti networks excepted aesthetics? (Not that aesthetics aren't important)

[thin mauve]

I'd say performance, plus "debuggability" sometimes you just miss the situation and don't know how to plan your next mission

[raw parcel]

Oh that's fair yeah

[thin mauve]

@shadow gyro obviously I cannot reproduce the Sus Kerbal bug ;_;

this is right after loading

After a bit it loose connection

But KSC is on the other side

am I missing some RA-100?

[brazen oasis]

So I am loving the release! It's working great! Here's my current Jool system:

[thin mauve]

Working on some UI goodies.. time to provide analysis tools

[brazen oasis]

This gets at what I am looking for in the bands or channels

[brazen oasis]

I thought I'd provide my thoughts because I know you're thinking about what to implement there now.

[thin mauve]

I'm all ears!

[tepid dust]

A biblically accurate angel but with ears instead of eyes

[thin mauve]

lol

[brazen oasis]

What I'd like to implement is to have a tier-A set of sats that are connecting planets together. Those are my polar sats. Then I'd like a tier-B set of sats that distribute coverage within a planet's SOI. Those are the sats orbiting in eq orbits ouside the moons here. Finally I have tier-C sats that orbit low on each moon.

I'd like the tier-B sats to only get coverage from each other or tier-A sats

And I'd like tier-C sats to either get coverage from tier-B, or from tier-C orbiting the same body, for example Lathe

[thin mauve]

so you'd like to allow direct A-C connections?

[brazen oasis]

I think there is a direction here

so I would like each tier to get its service from either the tier closer to KSP or its peers

but I don't want a tier B to get service from tier C or tier A to get service from tier B

[thin mauve]

ahn, directional

[brazen oasis]

yeah, I think so

[thin mauve]

directional sounds harder but doable, it's an interesting idea

[brazen oasis]

I'm trying to think if I could do it without directions. What I don't want is something like the bottom sat in the picture below to get service with so many hops away from KSC:

That's a tier-A sat, it can connect home by itself, but it's routing through a tier-C.

I might not need directional if we minimize hops

[tepid dust]

Hmm
Could a comm network count as a directed acyclic graph?

[brazen oasis]

or minimize total distance

does't comm-net already have a direction? You even shade it showing the direction

Am I understanding it correctly that everything is centered around KSC? I wonder if that might change when they have interstellar... maybe we'll get other alien races' version of KSC. That's what some of the lore makes me think

But for now, it seems very directional with KSC at the root of a tree, and I'm trying to keep what I intend to be last-leg distribution nodes to be leaves of the tree and not trunks

Maybe if an antenna can have an uplink and downlink channel / band?

[thin mauve]

definitely

at least, it's how I'm processing it 😄

weighted DA

DAG* with dijkstra

[thin mauve]

Yup, both things: it's directional, and everything starts from KSC

[brazen oasis]

or maybe when you are connecting to bands, you try to connect to the highest one first. Maybe it is a priority thing

[thin mauve]

but technically I always find all the isolated sub-graphs, so I could know if there is another satellite marked as "CommNet" source, if there are other nodes connected to it

in fact, in first release, this was a bug that made other relays connected even if there were no direct connectiont o KSC 😄

[brazen oasis]

maybe we just need a priority you can set for a relay and you always try to connect to the highest priority item you can connect to, and the closest if multiple for that priority

just throwing different ideas out

[thin mauve]

priority would work to choose, and it's not a problem since I'm always choosing the "best" path; it's just a matter of defining the criteria, which one is the best

but the main issue I see is choosing if it's allowed or not

[brazen oasis]

exactly, maybe I don't mind if it is allowed as long as you pick the higher priority over the closest

[thin mauve]

well, not too much of a problem: it could be that lower bands can receive signal from higher bands, but not the inverse

[brazen oasis]

I like this

[thin mauve]

The only thing is that now that I think about it

it's not a DAG

[tepid dust]

Ahh
I was gonna suggest looking into it for possible optimization lol

[thin mauve]

it's just a weighted graph

direction doesn't matter

mh

no well this is not possible

direction matters

[brazen oasis]

yeah, direction matters because there is away and towards KSC

KSC is the root

The thing I am seeing googling DAGs that isn't in comm-net is the multiple paths to the same node

I'd say it is a tree with a root, because we don't really have a concept of any kind of point to point transmission that doesn't involve KSC rn

[thin mauve]

this is the update I'm working on right now in reality

[brazen oasis]

If we had a use case where vessel-A had to send a message to vessel-B, we might have more of a network without direction with multiple paths

[thin mauve]

I'm tracking all possible connections

(in range)

between two vessels

while right now I show only the choosen connection

[brazen oasis]

I'm actually fine with whatever happens under the covers. But when I see the network with the green lines, it would make me feel good to see it using things for their intended purpose, and it would be nicer to see the connection of that bottom sat going either to the top one or back towards Kerbin

[brazen oasis]

This would get me what I'm looking for, I think.

or else some kind of configuration that allows you to choose to minimize # of hops?

[thin mauve]

Right now there is the option to choose between "nearest relay"

or "shortest distance to KSC"

but hops could be definitely doable

[brazen oasis]

there is? where is that config?

[thin mauve]

in Settings > Mods

under "CommNext"

[brazen oasis]

I need to toggle it and see what happens

[thin mauve]

it's cool!

try it

even if I find nearest relay a better default

[brazen oasis]

I think I like it better. Here's what I see after toggling that and reloading my game.

That's fair, because my local relay has a bigger range than I need and can make it up to the polar relay

I might need to nerf the RA-2's even more 🙂

I think this is great! We'll see what you come up with bands, but I think I have enough with this to bend things to my will 🙂

[thin mauve]

with bands it would be cooler, I think I'll release at least a basic version soon

since you can choose colors and which connects together

even if you'd need multiple antennas

[brazen oasis]

multiple antennas wouldn't be a problem. I'd enjoy that!

[thin mauve]

good to know!

new tools! Plus, this definitely helped me find a bug.. all connections except for KSC one just die

or.. not

it's right, technically it doesn't have anymore outbound connections

since KSC is offline

well I may add them anyway, but this should help debug the connections a lot!

well, it's enough for today, have a good night my friends

.. or day.

timezones.

😄

[unborn basalt]

Say, will signal strenght be a thing? Like where you get 50% science with 50% signal? Or it does not have to be linear... like get 100% science up to 75% signal and it drops off below 75%.
Is this something that is considered?

[tepid dust]

I don't think that accurately models how digital communications are done

it's either you get the information or you don't

[unborn basalt]

If you transmit for long time because of low bandwith, there can be loss... but i get what you mean.

I was just thinking it was a good gameplay tradeoff in KSP1...

I mean what is more realistic, have some potential loss, or never ever have any loss?

[tepid dust]

longer transmission times

[unborn basalt]

I loved what kerbalism did, with the data, and the storage drives and all that... id love to have that back 😄

Anyways, i cant contribute anything helpful here, i just wanted to give that input.

[thin mauve]

Hi @unborn basalt thanks for the input anyway; I'll add something down the line related to power consumption, data transmission, etc, but I still didn't dig down the science part

[unborn basalt]

´s all good. The project is shaping up to be great, you get there when you get there 😉

[hidden nexus]

Not sure if this is a game bug or related to CommNext.

Got three satellites in 700km Kerbin orbit, each sat consisting of an empty FL-R25 tank with an Octo-2, RA-2 and solar panels attached.

CommNext is set to require power, and since the Octo-2 has only 5 EC the sats (should) run out of power when in Kerbin shadow and (should) stop to relay.

But all sats will continuously relay, even the one in the shadow with supposedly no power, if you are not in control of the shadowed sat.

Properly losing power and relay when in shadow and controlled.

Oh and when you do control a sat and it has lost power in the shadow, if you then switch to another vessel, the previously controlled sat (lower right corner) will stay "powered down" even once it gets back into the sunlight.

[obsidian solstice]

KSP2 does not check EC of background vessels

This was part of a performance optimization due to complaints about poor background vessel performance

It will just save whatever EC it was last at and use that to determine whether its powered up or not

[hidden nexus]

I see, thx. So one better has to make sure to never leave a vessel out of juice, until a less awkward "optimization" has been implemented?

[obsidian solstice]

It uses a variety of logic to try to figure out whether it should check background vessel resources, such as if the vessel is currently doing a burn

That logic will likely be expanded on with colonies

Mods can also override this logic and force background resource checks if needed, which is what KLSS does

So CommNext could do so as well if that was desired

Perhaps relays that are responsible for the connection to the active vessel could have their resources checked?

[hidden nexus]

Yeah, while that wouldn't help with a situation like in the last screen, it would of course at least prevent the loss of critical connection(s).

[raw parcel]

Oh yeah that reminds me I don't need it anymore for supplies consumption 'cause I'm not using the ResourceFlow system anymore. I only need it for recyclers now

[obsidian solstice]

Darn, my example ruined

[warm stratus]

Interesting that nobody using 45 degree inclinations.. usually 3 satellites with 45 degree inclination are fully suffice for a constant connection

[left obsidian]

why 45?

[obsidian solstice]

3 satellites on the same inclination at geostationary orbit will always have two blind spots right? The inclination just determines where those blind spots are

[left obsidian]

so when the inclinatuon is 0 aren't the blindspots at the poles ?

[obsidian solstice]

I believe so

[left obsidian]

ohhh so wiyh 45 deg it will change over time

[obsidian solstice]

No it wont

It'll just be offset to two circular areas

that are not the poles

[left obsidian]

but if the planet is rotating faster or slower than the satelites the spots have to move

[obsidian solstice]

Thats why I specified geostationary orbit

[left obsidian]

oh

[obsidian solstice]

if its not geostationary orbit then yeah the spots will move

[tepid dust]

Yes
This is why the USSR used tundra and molnjir orbits iirc

[shadow gyro]

Well yeah, there will be blind spots, but only for surface vessels or really low satellites. Most of the time it's fine.

Molniya orbits are cool

[tepid dust]

Ahh i knew i spelled that wrong

[bronze anvil]

I belive molnjir is the Norse spelling for the same thing, but they're using the bifrost...

[shadow gyro]

Yeah, Asgardians don't need Molniya orbits, they have more advanced tech

[thin mauve]

Well, all ready for some Wayfarer's Wings vibes! In the end I built a vessel report window, which I think could be very useful for anyone; @shadow gyro it seems like the vessels get occluded by Dres first and Kerbin later, see this video:

[tepid dust]

ooooooo

[soft fiber]

Oooh. New wing.

Flying in a spaceship with no connection to KSC.

"ET no phone home." (And it could also connect to commnext!)

[thin mauve]

Ohhh awesome

[soft fiber]

(I still really like the idea of Easter egg or silly wings)

[thin mauve]

I love them too

[tepid dust]

"Interplanetary Game Of Telephone"
Have at least 10 hops home

[soft fiber]

Oh that's also good.

[thin mauve]

I've heard there were a messagebus implemented in SpaceWarp to do inter-mod communication

am I right @verbal leaf @tepid dust ?

[verbal leaf]

yup

[thin mauve]

sounds perfect

I need to make the Orbital Survey wings too

[verbal leaf]

we really need some up to date SW documentation

😭

[thin mauve]

but that should be easier with experiments

[tepid dust]

Just ping @CheeseGPT she totally exists and knows how to do this stuff

[verbal leaf]

we need an API key to your Neuralink implant

[thin mauve]

😄

[shadow gyro]

That looks really, really nice!

[thin mauve]

Now working on displaying this in the map view

that should be definitive

[thin mauve]

ShaderGraph is cooler than I expected

[tepid dust]

Interesting music lol

[thin mauve]

lol

[verbal leaf]

does it work well with KSP2? I know @crisp plover had some issues because KSP2 doesn't use URP

[thin mauve]

I was just saying god why windows captures audio from the damned screen capture tool

it's a screen capture tool

not a hey what's in my headphone right now capture tool

[crisp plover]

shadergraph can use the built-in shaders, that's what ive been using, it wont even let you select URP/HDRP shaders because you would have to install them first, and that would break a lot of things in the editor- dont ask how i know

[verbal leaf]

tbh I find it very useful

[thin mauve]

I'm not having issues, the actual Shaders in CommNext I released are all SG based

but hey I'm not an expert

[verbal leaf]

in that I don't need to whip up something more complex like OBS for short clips

including stuff like game sound

[thin mauve]

Haha I was just curious about the default choice

the feature by itself is very nice

[thin mauve]

I'd like some feedback here, what do you think?

[verbal leaf]

I really need to try to look into implementing LFO shaders with SG

because if that works, it will change everything

[thin mauve]

Plus, they're animated!

[verbal leaf]

oh that's so cool

[raw parcel]

I'd put the vessel name below the list of relays and add "Filter" and "Sort by" labels above the dropdowns

Oh and for the comm lines, maybe a bit thinner would look even nicer?

[thin mauve]

The moving ones or the normal ones?

(these are shown only when you open the vessel report)

[raw parcel]

The arrow moving ones

[tepid dust]

holy shit thats the coolest commnet stuff ive seen in ksp2

[raw parcel]

Ohhh

[thin mauve]

let me show you

[thin mauve]

🎉

[raw parcel]

Ohh that's so cool looking, yeah the arrow size is perfect for this type of window
How are the arrow colors chosen?

[thin mauve]

It's Purple for outbound, Blue for inbound, Red for disconnected

(I tried to match the window UI)

But i'm not so convinced

[tepid dust]

you should not scroll the disconnected one?

maybe that will make it obvious that it's stopped

[thin mauve]

Seems a nice idea

[thin mauve]

Done!

I should say it's very precise with FMA

[tepid dust]

Well we did basically just say screw you to floating point error in the discriminant

[thin mauve]

it's really awesome

[tepid dust]

As I said, you should really just move your entire root solver into C

if only to make it more consistent in case you need to do more magic

[thin mauve]

plus it has been very fun to find out that there isn't any bug here, but just a real occlusion

[thin mauve]

I'm using a lot of NativeArray black magic by unity right now, would bit a bit worried to move that part there

[tepid dust]

Use C++ and do Array<N>

[thin mauve]

is it threadsafe?

I don't technically need it to be, but you know..

[tepid dust]

Oh I see

Is native array thread safe?

[verbal leaf]

yup https://docs.unity3d.com/Manual/JobSystemNativeContainer.html

[tepid dust]

I mean Array<T,N> in C++ is basically T[N]

Hmm

Are you accessing the same index from multiple threads @thin mauve

[thin mauve]

I had some checks to avoid it

but in one case it was easier to just do it

since it shouldn't never happen

but if it happens, it's threadsafe

[tepid dust]

Then Array<T,N> should work

[thin mauve]

Well right now it's not my main issue, but if I take the hard path I'll do this refactor too

[tepid dust]

True, setting up the job may be a pain like that, but you can use the native pointers from NativeArrays as well

[thin mauve]

I think this would be a cool post 1.0 thing to do, very exciting

[raw parcel]

I wonder if an Array<shared_mutex<T>, N> would work well 🤔

[thin mauve]

or we could just try to Burst compile it..

[tepid dust]

Apparently that might like require mono on the modders computer

[thin mauve]

ahn, forgot about that part of the conversation

Well building a native one in CPP for sure would not be easier

amazing, for sure, but not easy

[verbal leaf]

what's the output of it if you compile it in Unity?

does it like give you a native DLL or something?

because that would be way easier to do than forcing the users to have Mono

yeah seems like it

[thin mauve]

It emits another DLL yes

[verbal leaf]

then there's no need for runtime compilation

just pre-compile it with Unity

[thin mauve]

Ah, I thought the issue was for the modders (which I tought was reasonable), not for the players; yeah if the players should download Mono that could quickly become unmanageable 😄

[thin mauve]

damned floating precision.. here we go again

(the blue connection is from KSC, so I think in the shaders floats are losing precision)

[verbal leaf]

lmao

it's just packet loss over that huge distance

[thin mauve]

hahaha

sounds like a perfect "it's not a bug, it's a feature!"

[thin mauve]

It's coming out really nice I'd say

[tepid dust]

when your solver has more precision than the shaders

[thin mauve]

hahaha

I just made them bigger and who cares, now they doubles as distance indicator 😄

[tepid dust]

That feels odd to me

Idk why

[thin mauve]

it's not beautiful, I share that

[verbal leaf]

it does kinda feel like they're squished unintentionally

[thin mauve]

I'm so sad

[tepid dust]

Hmm

Do you need it rendering the whole distance

[thin mauve]

Well it depends on the angle, if you see it from the side yes

like that

[tepid dust]

Even then

It goes out of view due to being too thin before it reaches the end

[thin mauve]

yup, it's just that handling the camera position is.. something I don't know how to use

[verbal leaf]

I'm more so wondering if it could be instead done so that it always renders each arrow with the same size, independently of distance and zoom

[thin mauve]

The main issue

is that the MapView when you "zoom"

scales the world

[verbal leaf]

[thin mauve]

for example

after a bit of zoom out, it goes back to normal

since now scaled space is ok

I'm pretty amazed how they did manage to get the MapView working

it's absurd they are resizing the world every time you zoom

I was a bit confused the first time I was debugging the positions

[tepid dust]

Ah yeah

[verbal leaf]

conclusion: let's just rewrite the whole game using UITK and ECS

[tepid dust]

I found this interestinf

But thats basically what zooming is in camera space anyways

But why scale the world

[thin mauve]

a floating camera zoom basically

I don't get it

I red a bit of code but it's still a mistery to me

there's sure a really interesting idea behind, but I've got no clue

@verbal leaf I thought I was going to bed too late in these months

but gosh you are beating me

[verbal leaf]

I mean you're still up

[thin mauve]

lmao

[tepid dust]

Both you could just go to bed yknow

[thin mauve]

I.. can't.. shaders...

[verbal leaf]

and I can't... condition editor

[tepid dust]

aaaaaa

And here I am laying in bed

[thin mauve]

@tepid dust what time is it there

[verbal leaf]

https://tenor.com/view/what-is-sleep-patrick-no-sleep-gif-15421799

[thin mauve]

lol

[tepid dust]

11

[thin mauve]

god

if it's time to go to bed in US

it's definitely too late

someone could argue it's early right now

[verbal leaf]

I have a very fucked up schedule in that some days I go to bed at 7am, and some days at 10pm

[tepid dust]

Your sleeping habits are questionable

[thin mauve]

mine for sure 😄

well let's not exceed it

have a good night my friends!

[verbal leaf]

good night, I still want to make this work before I go

🥲

[thin mauve]

v0.5.0 is out!

Plus finally I've setup the forum thread & mod news post 😄
#1208437501807566888 message

[raw parcel]

Hmm I don't know if it's a bug but this vessel isn't getting signal even though it has other relays in sight and has power. I might be misunderstanding something here?

[verbal leaf]

@full oriole hey, I just wanted to let you know that I deleted your message in the CommNext thread in #1168292144507793488, since the threads there are only meant to have messages by the mod's author (https://discord.com/channels/1078696971088433153/1168636558228860968)

[raw parcel]

Also some small UI suggestions:

• Add a color / weight change when hovering over the CommNext appbar icons and the Comms Report close button
• I would love some tooltips for all the non-textual buttons to know what they do at a glance
• I would disable the horizontal scrollbars for the dropdowns (see attached)

But all in all thank you for the great mod & update!

[verbal leaf]

it's really dumb that Discord doesn't have better permission management so having a pinned rules thread is the best we can do

[full oriole]

sorry my bad

[verbal leaf]

but yeah it sucks, the visibility is not great

[full oriole]

the threads was a great improvement over old set up but is a bit busy AF

[raw parcel]

Also the Comms Report doesn't update its current active vessel when switching vessel in the map view

[thin mauve]

It's very strange, it's like it doesn't find power in any relays, so for sure a bug; could you send me the savegame?

[full oriole]

do i need to manual delete any cache or anything if i snag it from the server? as opposed to waiting for CKAN to update?

[raw parcel]

(might want to refresh it on VesselChangingMessage)

[verbal leaf]

nope, nothing extra should be needed

[thin mauve]

I would disable the horizontal scrollbars for the dropdowns (see attached)
I'd like this too but it seems it's not possible ;_; Dropdown UI is customized in root PanelSettings, so there should be things done in the main UITK repo for this to work

[verbal leaf]

the native dropdown is a huge pain in the ass

[thin mauve]

I could make it wider, but it always depend on localization

[verbal leaf]

it will get much better once I make a custom implementation of it

[thin mauve]

Add a color / weight change when hovering over the CommNext appbar icons and the Comms Report close button
You're right! Missed that

[raw parcel]

I have these mods installed FYI

[thin mauve]

I would love some tooltips for all the non-textual buttons to know what they do at a glance
I could add this to "Sort" and "Filter" dropdowns so they are clearer too

[thin mauve]

Almost the same, so I think the issue should be in some early setup

I'm missing some edge cases maybe

[tepid dust]

@thin mauve have you tried importing fma from urctbase.dll?
Though that may be more pain than its worth

[thin mauve]

Nope didn't try

But it's working fine right now, have you found any issue?

[tepid dust]

No
I was just wondering if it were possible is all, but in all honesty I think it isnt

[thin mauve]

groan

damned bugs..

Could you try sending me the logs? Maybe there is something helpful there

Ok, reproduced it

I had the "Relays does not require power" on

@raw parcel I don't think that would be easy, in the meantime could you play with the "Relays require power" setting set to No? I think I'll have to dig inside of some of your KLSS code to see how you handled EC 😄

[raw parcel]

Roger that, if you have any question about my code don't hesitate to ask

[brazen oasis]

Hi! I'm loving the new version! Thanks so much! I think the comms report window really gives a lot of great information, and the new connections visible when that window is open look really cool!

I do have a request after thinking about how you had discussed bands would work. If we're going to have multiple antennas so you can connect to multiple bands would it be possible to restrict the antenna or band to just "in" connections? That would give us a lot more control over how connections happen.

I had thought I could get things working the way I wanted with the minimize KSC distance, but although that is closer to what I want, it's still connecting in ways that I would rather it not. I think this functionality is kind of related to the band functionality so it might wait until you implement that. Then for each band, you'd have sources and consumers for connectivity towards KSC.

[thin mauve]

antennas so you can connect to multiple bands would it be possible to restrict the antenna or band to just "in" connections? That would give us a lot more control over how connections happen.

It's something I've been thinking too, the direction; I'm not so sure it would be so easy to implement, but this evening I'd like to start working on the bands thing (after fixing these bugs), so I'll tell you better

[brazen oasis]

Keep in mind what users would care about is the green tree which has direction already and not something more abstract under the covers. If there are invisible connections that are not used when connecting to KSC, those don't matter. It might be tweaking the criteria of creating that path which you already allow the user to choose between closest next hop and min distance to KSC.

If you can't add direction to that check, I would appreciate having a minimize hops option, which is closer to what I want, but having direction control would be optimal

Thanks again for this mod, it already greatly enhanced the game!

[tepid dust]

I need to make a patch manager addon for making wayfarers wings easier

Not that its too hard

[thin mauve]

tooooltips

[tepid dust]

What
Can this be added to sw?

Wait no

This is its own whole bar

Kind of like what @verbal leaf is planning for the app bar library

[verbal leaf]

yep

[thin mauve]

I'd prefer so

it's just a "TooltipWindow" in front of other windows

with a Manipulator to show it at the right place

(to be clear it's not connected to the specific window, one TooltipWindow is ok for all the windows)

[hidden nexus]

Not sure if this has been said but could you make the Report button a toggle, so that a subsequent click would close the window?

[thin mauve]

already done will be in next patch ❤️

[thin mauve]

In the end it was me not updating the Vessels State, it shouldn't have been needed but.. it works now 😄

[shadow gyro]

QoL request: persist app tray position in save game files (or somewhere else). I'd like to move the tray to the top left and not have to reposition it again.

[shadow gyro]

Just wanted to follow-up on https://github.com/Kerbalight/CommNext/issues/5. I'm still getting the same bug as I described earlier. It seems that all vessels appear not to have EC when the save loads. But when a kerbal exits the vessel, state gets refreshed and everything is normal. Since you cannot reproduce it, I assume it's some interaction with another mod that I have that's causing this. I suggest you disregard this for now until (or if?) someone else reports this, cause it might be some weird edge case that might not happen again.

[thin mauve]

Oh thanks @shadow gyro

Didn't try with EC consumption; I'm fixing some bugs with it right now

I'll try again right now

[thin mauve]

@brazen oasis @shadow gyro I've released v0.5.1 which should fix this EC bug; it should be the same affecting you @raw parcel . Plus it should handle the Science Relay issue, but I did only some simple tests so I'd be glad to have some feedback from you on this side @brazen oasis .

The main issue I've found is that the game, by itself, does not update comm range when a vessel is undocked or a part decouples, so I had to trigger internal refresh in the mod

[brazen oasis]

I'll check the science thing later today. Thanks so much!

[thin mauve]

.. And v0.5.2 with position saving & an hotfix to toolbar icons. Have a good night!

[brazen oasis]

position saving?

ah, position saving of the toolbar

[thin mauve]

yup

[brazen oasis]

I still don't see 0.5.2 in CKAN (it takes a bit) so I'm going to test the science thing using 0.5.1

science works with repeaters in 0.5.1. Thanks!!!

[hasty bloom]

Feature request: I'd like to be able to disable the relay component of a comms part, because due to Orbital Survey I have relay antennae on the region survey satellites, but I do not want them to be valid relay nodes.

[shadow gyro]

Works beautifully!

[thin mauve]

Good news! Thanks for the feedback 🙏

[thin mauve]

Hi, yes we’ve got some features planned to avoid this, like bands and channel switching

[brazen oasis]

Yes, one of those should have a "none" setting eventually. That way you can dedicate an RA-2 or RA-15 to just Orbital Survey.

@thin mauve I've also play tested loosing contact with the commsat network and I correctly was not able to upload science, so I think your fix is working perfectly.

[thin mauve]

The bands thing appears to be harder than I thought 😄

I had to support multiple ranges for each vessel (based on bands selected on each antenna)

plus I think it's really hard right now to debug it visually

[wanton oar]

Because all the range bubbles are overlapping?

[thin mauve]

the range bubbles are an issue on their own, the Bands issue is that now you have to understand that two sats may be in range but cannot comunicate anyway since they have different bands

or that they are in range, same bands, but the bands are set on the RA15 antennas while the RA100 is set to a different band which doesn’t allow communication

[wanton oar]

How difficult would it be to make a UI switch to only show the range bubbles for antennas active in a given band?

[thin mauve]

it’s a good idea and something i’d like to implement, not sure if sufficient but could be a great start

[left obsidian]

It seems like the distance sphere isnt shown for small antenas

and seems like sometimes things aren't connected when they should be

And it shows the current vessel in the vessels list , i dont think it should be shown

Also the Range shown in the upper right corner is wrong

It always shows 200Mm

And i dont even have 1 antena that does 200Mm

These 3 Antenas dont show distance sphere #1203886735394734130 message

[thin mauve]

@left obsidian yes currently Range is shown only for relays

but I suppose we could toggle them for any vessel, maybe with a different color

could be related to the range bug

[thin mauve]

Like which one? Could you show the arrows drawn in map view? It's pretty much impossible to say you more without that

it could be you're using a non-relaying antenna?

I should definitely highlight that in the UI

[left obsidian]

oh there's a difference in relays and non-relays ? i didnt notice that that was even a thing.

So that's why my constalation couldn't work

The wrong distance shown is defently a bug thought

[left obsidian]

And could you chsnge this?

[thin mauve]

I will look at it

@left obsidian Relays are described in the OAB anyway, just pressing Shift you can see which ones are providing the signal relay feature:

[tepid dust]

Ohhhh

How do you do the heading part?

I want to know for V-SwiFT to make it look better

Unless it already does it and im dumb

No it doesnt

It just says "statistics"

[thin mauve]

I didn't touch the "## Statistics" part 😄

I only added the # Signal Relay bit

I tried to check the left column but it's pretty much hardcoded

[tepid dust]

But how

Is what I'm asking

[thin mauve]

Oh, it's a Module

wait

[tepid dust]

Because my module only shows with a # statistics on the right

[thin mauve]

PAMModuleVisualsOverride +: [
        {
            PartComponentModuleName: "PartComponentModule_NextRelay",
            ModuleDisplayName: "PartModules/NextRelay/Name",
            ShowHeader: true,
            ShowFooter: true
        }
    ];

[tepid dust]

Ahhh

[thin mauve]

Are you using Shift?

(I know

[tepid dust]

Yes

[thin mauve]

dumb question

sorry)

[tepid dust]

Its ModuleDisplayName I was missing

[thin mauve]

Even if

I'm not sure it works correctly

let me see if I find the previous discussion

I don't find it, anyway I had some issues and iirc:

• In the OAB select, you can pass anything you want in ModuleDisplayName
• In OAB/Flight PAM, it will alway look for PartModules/<Module name without "Module_">/DisplayName

[brazen oasis]

Hey, heads up there is a minor usability thing that may trip up new users. I think this was introduced when you put in code to save the location of the toolbar. I started a new save, and this is where I think it put it:

I couldn't find it / thought it was missing at first

but I can drag it to a good spot.

My thought is you need to give a non 0,0 default value for it for new saves, maybe 200,200 so people see it and can drag it where it makes sense for their resolution?

(or if this doesn't sound right I could have just accidentally moved it there somehow)

[thin mauve]

Oh strange, I tried to keep the default position if it's not set; I'll try with a new save, thanks @brazen oasis

[brazen oasis]

well like I said it could have been user error on my part. Wouldn't be the first time

[thin mauve]

Default position is always a problem so it could easily be a bug

[thin mauve]

Some progress on the Bands update.. new UI in the PAM with band picker, range bubbles visualization dedicated for each band with color coding, vessel report details in case band is not matching

Still bulkier than what I hoped for

[full oriole]

Does the Current commnext include the visual graphic? i assume this it totally unrelated to com lines bc there isnt a dependency? I do not have the visuals and am not very good at all this yet. The settings menu for mods is smooshed so i cant really tell if there is a hot key for the menu. it doesnt appear on my app tray.

[thin mauve]

yup it should be enabled and there should be a custom app tray saying commnext in the map view; would you mind sharing a screen?

[verbal leaf]

#1203886735394734130 message (this custom app bar should be on your map screen)

[thin mauve]

yes it has no dependency on CommLines anyway

[verbal leaf]

this is an unrelated mod

https://discord.com/channels/1078696971088433153/1163071637214724096

[left obsidian]

XD wrong channel

sorry

[full oriole]

I had to pick my mother in law up from PT.

I reinstalled it right before I left. The band option is appearing in parts manager.

About 30min I'll be back.

[full oriole]

only the 2 on the outer most band would qualify for relay. other 4 are direct sats that were using com lines before i uninstalled it

[verbal leaf]

you won't find the mod in the app bar tray with other mods

it adds its custom floating app bar

which looks like this: #1203886735394734130 message

[full oriole]

its in the farthest reaches of the upper left corner without any visible choices upon mouse over.

Is there a way to move the window

if default settings include visible lines, something else is likely at play as well (Maybe? I wouldnt know)

[verbal leaf]

you should be able to just drag it around

[full oriole]

i tried without success i will try diffrent areas of visible frame

yea

Thank you

little bastard was hiding behind the KSP menu bar. Im so happy.

[thin mauve]

Sorry, this is a bug in latest release which I should fix. Happy you’ve found it anyway!

[full oriole]

🙂 I love the mod

[left obsidian]

Seems like after changing the name of the vessels it didnt syncronise with the Commnext vessel connection report

[brazen oasis]

me too!

[next galleon]

I started a new save specifically to use this mod and it's so fun at the start before having any relays

waiting for the KSC to line up or using K2-D2 to basically "send commands" in advance

[full oriole]

I want to play the first bit of WMCC with this

[left obsidian]

We shouldn't be able to have commnet when rentering , could it be add to the mod so that when there's the plasma around the craft we loose connection

[full oriole]

ooo... awww... fireball block signal?

[thin mauve]

I should say I didn't even test the tech tree progression 😄.. sorry if it has been too hard

[thin mauve]

sounds cool, will think about how to implement this!

[next galleon]

Oh no, you do have relay capable antennas pretty early, I just haven't gotten around to it yet

[left obsidian]

Nice

[manic halo]

Hey, just wondering why the landed probe on Eve has no connection to the KSC?
Both are in range of each other, the orbiter has a relay antenna, and the lander an antenna in range of the orbiter; but I have no connection for science
Might be that I misunderstood something

[bronze anvil]

Eve's atmosphere is highly attenuating IIRC. It's all the methane, it tends to absorb the RF... 😉

[thin mauve]

Hi @manic halo , so you can control the vessel but science connection is not working?

you could send me the save game, so I can check it; we've had some issues with science, but I thought they were fixed with latest update

[manic halo]

Exactly, had the issue re-appear with a Duna rover between the screenshot and now. What's weirder is that I have another lander on Eve that has no problem
Sending the save file in dms

[thin mauve]

Thanks, which is the vessel name in question?

[thin mauve]

Well, bands update is near! I've got all the new features in the UI with some cleanup:

• PAM allows you to select bands. Highter level antennas allow you to choose multiple bands.
• Vessel comms report now displays in a compact way which band is used for the connection
• Plus, in the bottom part, you see each band and its corresponding range for the active vessel
• Clicking on the connection focus the vessel on the map. With the "Control" button you can switch to that vessel in flight mode.

Now I just need to fix some bugs in the Sphere Rulers render and it should be ready

[obsidian solstice]

Are bands an optional feature or do you need to interact with them?

[thin mauve]

By default it gets the X band everywhere

if you don't change them, you get everything in that band

So basically you don't need to think about it

It's a completely opt-in gameplay thing

[left obsidian]

Are sphere's shown for every antena now ?

[obsidian solstice]

Ahhh got it

Good design

[left obsidian]

And will there be a way to disable some antenas to work as relays ? (like the ones used for environmental survey)

[thin mauve]

Nope, I still have to tackle that

[left obsidian]

maybe with V-SwiFT

[thin mauve]

Yes, it's already in the screenshot "Enable Relay"

[left obsidian]

one relay mode , one orbital survey mode

[left obsidian]

that could be cool

[thin mauve]

isn't this enough? Plus, you can switch behaviour in flight anytime, seems better 😄

[obsidian solstice]

I’m loving how this mod seems to be coming along. Definitely excited to play with it

[thin mauve]

Thanks @obsidian solstice ! Hope it will withstand the hype.. 😄

[obsidian solstice]

If it doesn’t then you’ll be up on the chopping block. Only the best modders survive.

[next galleon]

I'm doing my first relay network and I'm so giddy lol

Started a new save and I did a few missions around Kerbin and Mun without any relays, just going with god

it bothers me I can't make the perfect KEO orbit tho

I'm always a couple tens of meters and a couple tens of arcminutes away

[thin mauve]

Yep we need a "fix orbits" thing

I could add that CommNext one day

just to fix relays

[next galleon]

Flight Plan is perfect for that

[thin mauve]

Oh, perfect so

[next galleon]

but I like using MP thrusters to put the probes into the final orbit so it shows up at 0 deltaV so I can't make nodes

[obsidian solstice]

I reported that bug awhile ago

But they put it into long term backlog

[thin mauve]

gotcha

[thin mauve]

this is unfortunate

[next galleon]

yeah that's an idea for another mode. Call it "trust me, I have more deltaV"

let us make nodes without checking the deltaV requirements

[obsidian solstice]

It’s not possible

I don’t remember exactly why, it has to do with ksp 2 nodes being calculated in a different but much more accurate way than ksp 1

But how much deltaV you have is a required variable

[next galleon]

ah yeah cause it actually shows you for how much of the trajectory you'll be burning

[obsidian solstice]

Right

[next galleon]

I never checked but MP engines do show up as deltaV yeah?

[obsidian solstice]

MP is never calculated for deltaV as far as I’m aware

Maybe it does for the puff engines?

I haven’t checked

[next galleon]

yes it does

[thin mauve]

yep I think the issue is with RCS thrusters only

[verbal leaf]

schlosrat was at one point writing a patch for Community Fixes to include RCS in the calculations but it turned out much more complex than expected and didn't quite work as expected, so it was scrapped

[obsidian solstice]

Interesting

Seems like one area where the devs must improve things

Mods can’t do much

[next galleon]

I guess the problem is exactly that. They're RCS

so their firing direction don't necessarily line up with the vehicle's axis

I guess the easiest thing would be just make a version of the tiny RCS linear thrusters act like engines

[verbal leaf]

Yeah, that would be the best solution for now, probably

@tepid dust when V-SwiFT gets part module switching, this would be a great use

Switching between engine and RCS modules for the one-directional RCS ports

[tepid dust]

Gotcha

I mean thats likely next on my todo list surprisingly

After I test the stuff I'm doing with module editing/scalar editing

[verbal leaf]

Awesome!

[next galleon]

these are my relays. I put the thrusters on 1% thrust to fine tune as much as possible

[thin mauve]

Well here we go: band selection to filter bubbles and easy vessel switching / focusing in map view

[raw parcel]

Do the bands have functional differences or are they for network customization only?

[thin mauve]

Right now only network customization

I wanted to add more things but this release already took me too much time, so I’d prefer to release this with a complete feature, other things will come later

[raw parcel]

Understandable 🫡

[hasty bloom]

I'm very much looking forward to it!

In my quest for good signal strength in the inner Kerbolar system (does that even do anything?), I may have gone a bit far.

Excuse my badly tonemapped HDR -> SDR screenshots.

[tepid dust]

Oh gods

[thin mauve]

This is art

[raw parcel]

https://en.wikipedia.org/wiki/Mandala

[thin mauve]

@manic halo I'm debugging your landers on Eve, and I didn't find the issue; it seems EMPLP-2 is occlued by Eve, EMPLP-1 is connected, while EMPLP-Lander has no power; from your screen everything seems fine to, so probably I didn't get your exact issue?
I'm releasing anyway a new version, so you could try with that and we could get a better picture

(it could be related only to the under sea level-bug I told you, in this case it should be working in this release)

[manic halo]

Will see what is fixed with the new version

[left obsidian]

what does X S K Ka V mean ? anything specific ?

[thin mauve]

Released! v0.6.0 is out

[thin mauve]

@left obsidian https://en.wikipedia.org/wiki/Deep_space_bands

I added V which is another IEEE frequency

but they're just names for now

[left obsidian]

yep just saw, nice

[brazen oasis]

So just trying out the bands for the first time, and I'm still learning. One thing that confused me is that I have a number of polar satellites that are meant to be the back-bone network. These all have 1 RA-100 and 2 RA-15s. I changed the RA-100 on the Jool-Polar-North sat from X Band to V Band and disabled the RA-15s. I then moved to the Jool-Polar-South. However, that sat already had the settings changed. This is actually what I was going to do but I expected to have to do it to each satellite individually.

As I go from controlling each polar sat I see my changes already made. Also from the network perspective it seems like it happens when I control the vessel for the first time.

Still not sure this is a bug -- it might be kind of a default thing the first time you control a vessel with the new version. I'm continuing to play with things...

[brazen oasis]

So how are "secondary bands" used? Are they just a second band or do they imply direction (like secondary bands are only for incoming / towards KSC connections)?

[thin mauve]

Very weird, I’ll give it a try

[tepid dust]

I imagine you can have a planetary relay with a primary band of your interplanetary comms band and a secondary of your intraplanetary comms

[brazen oasis]

ah, the same part seems to only have one spot for the band. So changing an RA-15 on one changes it for all

[thin mauve]

it should be configured for each sat anyway, could you send me some screens?

[brazen oasis]

Here's Jool-Eq-1. It currently has both bands on both RA-15's set to V-Band

I set the top RA-15 to no secondary band. With out me doing it on the second one, the bottom one also gets updated:

[thin mauve]

Ah gotcha, I thought I fixed the symmetry set thing; the problem is that by default state is “shared” between two parts placed in symmetry in OAB

[brazen oasis]

yeah, I think it might be more, also between vessels

(the symmetry thing wouldn't be too big a deal)

Now I go to the polar sat

[thin mauve]

ok this is definitely worse

[brazen oasis]

this has 2 RA-15s (it is a modified version of the other sat) and those also have the setting made

Now I'll add an entry here

I added Ka Band

and back to Jool-Eq-1 and it also has Ka Band selected

[thin mauve]

Ok this is definitely.. strange

what the hell

[brazen oasis]

it's almost like the bands are being stored statically on the part class

[verbal leaf]

I know that the Kerbal Headlamps mod had a similar bug with the custom module

[thin mauve]

Let me double check the code, maybe I messed up something

[thin mauve]

Ha, the one with the Data_Headlamp initialization?

the ??= not working?

[verbal leaf]

this was the issue and fix iirc

[thin mauve]

well, let's do this!

thanks as always @verbal leaf

[verbal leaf]

it might not be related to your issue, but it's the one time I remember seeing the same symptoms

[thin mauve]

as always I cannot reproduce it, but it could be because I've already loaded the savegame with intermediate versions

@brazen oasis could you send me the savegame?

[tepid dust]

The issue here is then the default initialization of the data module

I think what happens is somehow every part module gets the same instance of the data

[verbal leaf]

yeah

definitely seems to be the case

[thin mauve]

Like if they ModuleData is recycled between Modules?

[brazen oasis]

here you go

That was taken immediately before my screenshots I think

[thin mauve]

Perfect, thanks!

[brazen oasis]

Not sure if this makes a difference, but everything that is tied together were copies of each other. For example, the polar and eq sats are based on the same initial sat. (I forget which one started as a clone of the other, but I know it's one way or the other.) If you are having trouble reproducing without my save, try launching multiple versions of the same (or slightly modified) vessel.

It's either that, or every instance of the part anywhere, which should be easier to replicate

[verbal leaf]

oh, that kinda sounds like the data object was created once in the VAB and then the same reference was used for all launched instances

weird

[brazen oasis]

yeah, that's why I said it. Maybe some kind of internal id?

[verbal leaf]

I mean reference specifically as in like an address pointer

in C#

[verbal leaf]

ah and it actually makes sense based on this, I think

[thin mauve]

But I don't get why we don't have this issue with other Modules

[tepid dust]

Because of that fix

[thin mauve]

I'm talking about KSP2 internal modules

[verbal leaf]

the original code on the left sets the this.dataHeadlamp field only once, and then on all subsequent calls, it doesn't do anything, since it was already set once (in the VAB)

and then the fix on the right is simply to create a new instance on each call of AddDataModules

[thin mauve]

For example we have this module too, in Module_Light

[tepid dust]

because each of the KSP2 internal modules has it instantiated on the prefab

[thin mauve]

Oh, so they have a Module_XYZ directly in Unity

this means that technically it's serialized

[verbal leaf]

while PM modules get instantiated in the VAB, I'm assuming

[tepid dust]

They get it instantiated OnPartPrefabLoaded

So the prefab the VAB uses has it instantiated

this uhh also means that most stock modules won't work with PM

and I just didn't realize that

That's going to go in my book of hidden invisible bugs to fix

[thin mauve]

I really can't reproduce it, but I'm testing if it works in my copy without issues and I'll release it

[verbal leaf]

now that is weird

if you can't reproduce it by launching the same vessel multiple times, then it can't be as simple as what we thought

[thin mauve]

oh, with the Add way now I can't use KSPDefinition

[tepid dust]

Each oab prefab is loaded once in the OAB

every time the OAB opens tho

No wait now I'm definitely confused

[verbal leaf]

how could the bug manifest the first time then?

multiple instances of the same part launched on one assembly?

[tepid dust]

idk
how is the same module being carried through to flight?

I don't even see the case where that could be happening

[verbal leaf]

yeah

[manic halo]

Follow up on this: seems to be fixed for the underwater lander (video 1), but not for the one on land (video 2). See how it is connected, full control, but can't beam science back :(

[tepid dust]

And wait the ??= should work anyways

This is what I have for V-Swift

[thin mauve]

I think the issue should be somewhere else, I just launched 4 identical vessels and I can configure different data on each one

And even on parts in symmetry set, so should be something different

[verbal leaf]

plus I can't imagine how this issue could persist through saving and loading

since that serializes the objects

[thin mauve]

Yes

The worst thing is that I cannot reproduce it even with your save @brazen oasis