Jump to content

[1.4.3 <= KSP <= 1.12.5] KSP Recall - 0.5.0.2- 2024-0521


Lisias

Recommended Posts

ANNOUNCE

KSP Recall 0.3.0.1 is on the Wild, featuring:

  • Fixes a huge brain fart of mine from 0.2.2.4 :blush:
  • Closes issues:
    • #52 REPORT: Diagnosing the GoAHead Issue
    • #49 ArgumentOutOfRangeException causes saving to break and makes my save file unplayable

— — — — — 


This Release will be published using the following Schedule:

  • GitHub, reaching manual installers and users of KSP-AVC first. Right Now.
  • CurseForge. Right Now.
  • SpaceDock (and CKAN users). Right Now.

—— — — — 

I found a workaround for upgrading the SubAssemblies!!!!

Use Craft Manager to load the SubAssemblies as they were Crafts, then drag and drop the thing back as a new SubAssembly! 

157281034-ded72f5a-e653-46a4-bcc2-a410cc

Edited by Lisias
All your Distribution Channels are belong to us! :)
Link to comment
Share on other sites

12 hours ago, Lisias said:

Good news.

Umm. Well, maybe not. I played my big career save (25 mb file) more today and there's some shenanigans. From flight I couldn't go to the space center or map, so I cleared input locks and it momentarily helped but now I still can't go to tracking station or space center from the ESC menu. Getting argument out of range or null ref when I click menu buttons.  I'll be back.

Update: OH... I didn't refresh the forum. I'm still on 0.2.2.4 with the MH file commented. 

Edited by Krazy1
Link to comment
Share on other sites

1 minute ago, Krazy1 said:

.Update: OH... I didn't refresh the forum. I'm still on 0.2.2.4 with the MH file commented. 

Try the 0.3.0.1 currently on github. It fixes my side of the ordeal. :)

Link to comment
Share on other sites

19 hours ago, Lisias said:

Try the 0.3.0.1 currently on github. It fixes my side of the ordeal. :)

Thank you so much. I've been hunting for days trying to track down all of my individual mods throwing errors before I finally made my way to the attach on edit error.  I'm good to go now.

Link to comment
Share on other sites

32 minutes ago, jellymanisme said:

Thank you so much. I've been hunting for days trying to track down all of my individual mods throwing errors before I finally made my way to the attach on edit error.  I'm good to go now.

0.3.0.1 is now available on CurseForge, SpaceDock and CKAN too. You should be fine by now.

Anyway, anything weird kick me here, posting your KSP.log on dropbox or something - not everything is my fault :P , and I need to check the log to tell you what's going - when that weird MH problem happens, innocent add'ons bork relentlessly.

It happened that, on this specific case, Recall was borking relentlessly due the MH problem, but also due a major brain fart of mine. I took some time until I could finally understand I was dealing with two different problems striking at the same time! :D 

Link to comment
Share on other sites

22 hours ago, Lisias said:

Try the 0.3.0.1 currently on github. It fixes my side of the ordeal

Welp, I upgraded it from CKAN and I restored the MH filename to normal... first try it pushed KSC about 50 m below ground... only the top of VAB and SPH were visible. Relaunched KSP... now everything's good. :blink:

Spoiler

full disclosure, I have been drinking, but not THAT much

 Thanks

Link to comment
Share on other sites

42 minutes ago, Krazy1 said:

Welp, I upgraded it from CKAN and I restored the MH filename to normal... first try it pushed KSC about 50 m below ground... only the top of VAB and SPH were visible. Relaunched KSP... now everything's good. :blink:

Ha! Typical behaviour on a race condition problem! :)

I need to explain something: I didn't "fixed" KSP by removing the Recall and TweakScale's from the clash with MH at Main Menu. I drove myself out of the way, moving that code to some place that, besides not ideal, it's good enough and are far from the mess MH introduced.

I, literally, took my car out of the dangerous lane - letting anyone behind me risking crashing in the pile-up in my place.

So, anyone being instantiated on Main Menu are still prone to have some problems - and given the description of the mishap, I'm betting Kopernicus is now the one taking the heat.

You have a pretty decent computer, but your GPU is not top notch anymore - only 3GB of VRAM. And, frankly, KSP became a VRAM guzzler. I'm reasonably sure that you need to trim down your textures to Half-Quality, or even Quarter-Quality depending on how much add'ons you have installed.

I can advance to you that on my rig, with ~1.5GB VRAM, KSP 1.12 behave pretty nicely trimming down the Textures Quality to 8th. The Stock textures were still acceptable (they really scaled up things!), but the Add'Ons' ones became unbearable. This is the main reason I'm doing this research at first place, as I need to trim down the StockTextures to match up the ones on older Add'Ons, so I can balance the overall visual experience.

Since you have twice the VRAM I have, it's reasonable that you should be aiming to 4th Quality (again, your mileage will vary due the add'ons you had installed).

And why this happens? 

Because when you blow up your VRAM's capacity, the GPU driver starts to "steal" memory from the CPU to store the textures, and so the GPU and the CPU starts to get contention while accessing that shared memory. Contention is the bus injecting wait-states on one device while the other is accessing the shared memory, and obviously, with the GPU trying to cram more FPS, the CPU starts to starve a bit, leading to a slower processing, leading to MH being delayed its startup on Main Menu, leading to that problem GoAHead experienced - but now that I'm not on that lane anymore, the crash is hitting someone else.

Making things even worse, everything and the kitchen's sink are using VRAM nowadays, including your browser. So the pressure on the VRAM is higher on recent Operating Systems - users still using Windows 7 should experience way less performance issues using weaker GPUs. Closing your browser will surely help (I can't open Facebook while playing KSP 1.12, otherwise the performance tanks!!).

 

1 hour ago, Krazy1 said:
  Reveal hidden contents

full disclosure, I have been drinking, but not THAT much

Apparently, you didn't drank enough! :sticktongue:

 

1 hour ago, Krazy1 said:

 Thanks

If you ever get some weirdness like that again, promptly quit KSP and send me the KSP.log, out_put_log, the whole shebang. By looking on the order of the events (and the probable exceptions), we can figure out who is the one being bashed this time, and then we can try to figure out a workaround for the problem like I did on Recall and TweakScale.

Not everybody being loaded on Main Menu is being affected by MH - I suspect that only the ones trying to read and/or write to the GameDatabase are susceptible to this crapness.

Cheers!

Link to comment
Share on other sites

40 minutes ago, damerell said:

I'm on Windows 7 (yes, I know; no, it isn't) and don't have MH. Truly I am blessed.

It was exactly what I was using on my beefier x5500 server turned into gamer machine - until the GPU gone kaput and I had to rely back to MacOS since them! :)

(boy, running KSP on a dual Xeon 5570 is absolutely fantastic - until the next power bill…)

Link to comment
Share on other sites

14 hours ago, Lisias said:

only 3GB of VRAM

So insufficient VRAM could cause KSC to sink into the ground. :confused: 

I have a theory Lisias... you just made up this wild story and told me to drink more so I would buy a new video card and then donate my old card to you! :P  Just kidding.

So far, I only saw this behavior once the first time after upgrading Recall to 0.3.0.1. If it happens again I'll try to save files.  

Performance ramblings...

Spoiler

I do have Scatterer and Parallax and the settings cranked up. It still shows some free VRAM though at 1920x1200. I'll try lower texture setting to see if it matters. 

q1vLjWm.png

It still seems that part count is the most significant performance hit. I have 296 parts for this snapshot. Single-process physics engine just sucks.  Those numbers are so aggravating... 12 FPS because it's CPU limited at 15%. 

Right now video card prices are insane. Who pays >$1K for a video card? I'll wait 6 months, hoping the chip shortage will be better. I think I'll skip 2K and get a 4K. 

 

Link to comment
Share on other sites

6 hours ago, Krazy1 said:

So insufficient VRAM could cause KSC to sink into the ground. :confused: 

Well… To tell the whole history, insufficient VRAM leads to using RAM, than leads to contention, that leads to terrible frame rates due both processors (CPU and GPU) having to fight for a piece of RAM that are accessed under a slower bus by the GPU (to makes things even worse).

However, on Unity, drops on the frame rates implies in some key code from your Application being called less times. So things that could had been run in 2 seconds on a 60 fps rig ends up taking 4 seconds on a 30 fps rig - or even 8 on a 15 fps rig. Tying your CPU to the frame rate was a less than ideal idea.

On KSP, however, things got way worse because people just ignores concurrency, no one gives a rat's SAS :P about it. And so most of the code on KSP just tries to "do things fast enough" in the hopes nothing bad happens.

But once you clog up your VRAM, that then lowers your framerate, that then makes things that usually runs in less then a second to start to take tens of a seconds (or worse) to happen, this castle of cards collapse.

Recall and TweakScale were the ones being thrown under the bus when this happens, but I moved them into a place where (until now at least) there's no concurrency happening (the Scene transition from LOADING to MAINMENU). Apparently Kopernicus had the bad luck of taking my place on this mess.

If Kopernicus finds his way out of the mess too, then someone else will end up being thrown under the bus in our place - and so something else will eventually happen.

 

6 hours ago, Krazy1 said:

I have a theory Lisias... you just made up this wild story and told me to drink more so I would buy a new video card and then donate my old card to you! :P  Just kidding.

So it didn't worked? :sticktongue:

 

6 hours ago, Krazy1 said:

So far, I only saw this behavior once the first time after upgrading Recall to 0.3.0.1. If it happens again I'll try to save files.  

Yep, because on 0.3.0.0 and previous, Recall was the one caught on the cross-fire and, so, it started to misbehave due badly initiated data before Kopernicus could had a chance to run. Once Recall withdrew itself from the dangerous lane, Kopernicus (apparently) took the hit instead, and so it was the one misbehaving due badly initialised data.

This came to my mind because the VAB is slightly higher than the ground level where it is - i.e., the floor of the VAB is at ~70 meters above the sea leval, and the VAB have about ~80 meters - so when the whole KSC position failed to be initialised, the default for variables unitialised (0) should have told Kopernicus to draw KSC at high 0, i.e. under the ground, and so only the parts of the buildings higher than 70 meters were displayed to you.

Edited by Lisias
tyop! Surprised?
Link to comment
Share on other sites

31 minutes ago, Lisias said:

Kopernicus (apparently) took the hit instead

There's a long stutter from Scatterer when it first paints the clouds on Kerbin at the main menu. And also KerbalX connects... so there's a server connection involved. Imagine the timing variation there.  

 

Link to comment
Share on other sites

15 hours ago, Krazy1 said:

There's a long stutter from Scatterer when it first paints the clouds on Kerbin at the main menu. And also KerbalX connects... so there's a server connection involved. Imagine the timing variation there.  

Curiously, I never got any problems with KerbalX.   Nice catch.

Just remember that these are not problems, not even the triggers. They are inocente bystanders stepping on a landmine.

The problem is uncontrolled concurrency.

Edited by Lisias
tyop! Surprised?
Link to comment
Share on other sites

  • 2 months later...

ANNOUNCE

KSP Recall 0.3.0.2 KSP Recall 0.3.0.3 on the Wild, featuring:

  • After some weeks of denying, it was finally realised that AttachedOnEditor is needed downto KSP 1.4.3 - on the very first ModulePartVariant implementation.
  • Fixes a small brain fart of mine on the INSTALL.md file.
  • Updates KSPe.Light to the latest release (2.4.2.1 at this time)
  • Closes issues:
    • #55 The KSP Editor is screwing things since 1.4.3
    • #50 Not sure how to install 0.3.0.0
  • On 0.3.0.3, an annoying warning was fixed - the AttachedOnEditor thing is now intended to be installed downto 1.4.3, the warning was bogus.

It's unlikely that this release will be of use for too much people, I don't thing there's many people using KSP as old as 1.4.3 as me - but a fix is a fix, and I lost some time chasing my tail on this krap anyway, so published it was.

While trying to diagnose a regression on TweakScale, I tracked down the problem downto KSP 1.4.3 . Since it was current knowledge that KSP 1.9.x was the one that screwed the Editor, I convinced my self that I had somehow screwed the craft load on TweakScale somehow. Only after exhausting all possibilities that a certain aspect of the misbehaviour was TweakScale's fault that I realised that the problem at hands may be, in true, two different problems - being one of them our old friend KSP Editor screwing up crafts on load.

So, yeah, I was wrong - KSP 1.9.x did something new on Editor que at that time had screwed again TweakScale, but the origin of the problem was KSP 1.4.3 (or 1.4.2, the Brief?), with the very first implementation of the ModulePartVariants. I'm not exactly thrilled but the discovery, but it explains why TweakScale was, until recently, potentially screwing up 3rd parties the same way it as being screwed by KSP Editor: I was playing tug-of-war with KSP Editor since 1.4.3 on TS's code.

As result, KSP-Recall is needed downto KSP 1.4.3 now.

Oh, well… Better late than never - they say.

— — — — — 


This Release will be published using the following Schedule:

  • GitHub, reaching manual installers and users of KSP-AVC first. Right Now.
  • CurseForge. Right Now.
  • SpaceDock (and CKAN users). Right Now.

Being a minor, bug fixing release, all Distribution Channels were updated at once.

Edited by Lisias
As result, KSP-Recall is needed downto KSP 1.4.3 now.
Link to comment
Share on other sites

  • 1 month later...

NOTAM

KSP 1.12.4 is on the wild! :)

But Steam wasn't updated yet, or at least it's not available to me at this moment. So it will take some more time until I download the thing and check KSP-Recall on it.

In the mean time, KSP-Recall will warn that it wasn't tested yet on 1.12.4 if you update to it before I release a new update - as usual, usually Recall will behave fine, but since I didn't check it, I can't guarantee it.

Backup everything before updating to 1.12.4 just to be on the safe side of the Force! (and you can manually disable any fix that eventually misbehave).

RUD safe!!! :sticktongue:

Edited by Lisias
Link to comment
Share on other sites

NOTAM

TweakScale (and some other mods) are borking due PDLauncher changing how things are started up on KSP.

On reddit, it was found the copying everything (but Squad and SquadExpansion) from <KSPROOT>/GameData to <KSPROOT>PDLauncher/GameData will "fix" the problem.

Problem: what happens with the settings.cfg? The savegames? KSP.log surelly is there now.

I will have an answer to it soon.

In a way or another, using PDLauncher is highly not advisable with or without Mods. 

NOTAM

Given that I finally got a Windows machine (blergh) to test things, I could check things myself and understood what happened about PD Launcher

The whole ordeal was a mess. PD Launcher is not a problem. POINT. TweakScale was the one reporting something wrong, not the victim nor the cause of the problem. POINT.

Some dude tried to accomplish something on Steam Launcher, made a mistake, and then a lot of people started to shoot the messenger (TweakScale) and I got involved on the shooting without a gun.

Yes, I'm eating my words on this one.

— — 

I just got notice of an even better solution!

 

Edited by Lisias
post edit
Link to comment
Share on other sites

To follow up on that, my KSSL launcher is windows only right now and requires the .net core, but the good news is that core (what I use) is cross platform and I should be able to support other KSP platforms/OSes (this is why I didn't just go with straight .net framework).  If anyone wants to get me the executable names of the launcher and the main game on the other platforms, I can probably support them too.

This is provided of course the launcher is on other platforms.  If it isn't, I'll probably just port the code to .net framework 4.5 (we all have that) and forget about cross platform.

Edited by R-T-B
Link to comment
Share on other sites

18 minutes ago, R-T-B said:

This is provided of course the launcher is on other platforms.  If it isn't, I'll probably just port the code to .net framework 4.5 (we all have that) and forget about cross platform.

As far as I know, only Windows got the PD Launcher configured on the Steam Launcher, besides it's being downloaded and installed on MacOS (and probably Linux). I don't know why, as by the Krakens I could not manage to run it - at least on MacOS, I will try Linux eventually.

Edited by Lisias
yeah, typo.
Link to comment
Share on other sites

  • 3 weeks later...

ANNOUNCE

KSP Recall 0.3.0.4 KSP Recall 0.3.0.5 on the Wild, featuring:

  • Adds a check for the pwd problem that started to happen when people tried to get rid of the PD-Launcher in an unfortunate way.
    • Check this for the whole history.
    • Additional links (do not try them, they cause this problem!)
  • Better diagnosing logs.
  • Updates Module Manager Watch Dog to the latest (1.1.0.3 at this time)
  • Updates to the latest KSPe.Light. (2.4.2.4 at this time)

KSPe (and KSPe.Light, the spinoff used on KSP-Recall) was reworked to allow running KSP with the pwd set to the wrong place without borking. This will not fix KSP, it will still shove files on the current pwd and Kraken know what ones. I think things are now less secure, but since there is a consensus to be followed on KSP, so be it.

So, KSP-Recall from now on will pesky the user about the problem, but without preventing him from running the game. I will not support KSP nor my add'ons if this warning is being shown to you on start-up. When (if ever) KSP is updated to properly cope with being run with the pwd different than Application Root, I will remove this.

EDIT: 0.3.0.4 was ditched, as I thought I had found a bug on it but ended up being a MacOS specific behaviour. Since I had already created some new logs, why not push them into the wild?

In time: MacOS is imune to the `pwd` problem! :) 

— — — — — 


This Release will be published using the following Schedule:

  • GitHub, reaching manual installers and users of KSP-AVC first. Right Now.
  • CurseForge. Right Now.
  • SpaceDock (and CKAN users). Right Now.

I stoled some hours from day-job to get this done, I need to finish the day before properly test it and then publish it on CF and SD.

Edited by Lisias
0.3.0.5 is on the Wild. All your Distribution Channels are belong to us.
Link to comment
Share on other sites

On 11/22/2022 at 2:39 PM, Lisias said:

ANNOUNCE

KSP Recall 0.3.0.4 KSP Recall 0.3.0.5 on the Wild, featuring:

  • Adds a check for the pwd problem that started to happen when people tried to get rid of the PD-Launcher in an unfortunate way.
    • Check this for the whole history.
    • Additional links (do not try them, they cause this problem!)
  • Better diagnosing logs.
  • Updates Module Manager Watch Dog to the latest (1.1.0.3 at this time)
  • Updates to the latest KSPe.Light. (2.4.2.4 at this time)

KSPe (and KSPe.Light, the spinoff used on KSP-Recall) was reworked to allow running KSP with the pwd set to the wrong place without borking. This will not fix KSP, it will still shove files on the current pwd and Kraken know what ones. I think things are now less secure, but since there is a consensus to be followed on KSP, so be it.

So, KSP-Recall from now on will pesky the user about the problem, but without preventing him from running the game. I will not support KSP nor my add'ons if this warning is being shown to you on start-up. When (if ever) KSP is updated to properly cope with being run with the pwd different than Application Root, I will remove this.

EDIT: 0.3.0.4 was ditched, as I thought I had found a bug on it but ended up being a MacOS specific behaviour. Since I had already created some new logs, why not push them into the wild?

In time: MacOS is imune to the `pwd` problem! :) 

— — — — — 


This Release will be published using the following Schedule:

  • GitHub, reaching manual installers and users of KSP-AVC first. Right Now.
  • CurseForge. Right Now.
  • SpaceDock (and CKAN users). Right Now.

I stoled some hours from day-job to get this done, I need to finish the day before properly test it and then publish it on CF and SD.

Uhhh I don't really understand what is happening here, I got this warning but I haven't messed with the PD Launcher and don't understand how the pwd is different than the application root. What is the pwd and what is the application root? As far as I know KSP is just installed to the default folder.

4m4UADC.png

Link to comment
Share on other sites

1 hour ago, TurtleMountain said:

Uhhh I don't really understand what is happening here, I got this warning but I haven't messed with the PD Launcher and don't understand how the pwd is different than the application root. What is the pwd and what is the application root? As far as I know KSP is just installed to the default folder.

Send me the KSP.log using dropbox and I will check it.

— — POST EDIT — — 

I just fired up my Windows Test Bed, and other than PD-Launcher downloading an update (still analysing what wold it be), I had no problems on running Stock 1.12.4 KSP and KSP-Recall on a full Steam managed installation.

Please send me your KSP.log - there's something different on your installation for sure.

Edited by Lisias
POST EDIT
Link to comment
Share on other sites

ANNOUNCE

KSP Recall 0.3.0.6 on the Wild, featuring:

  • Better diagnosing messages.

I think I opened a (yet another) can of worms around here. This KSP-Recall release will provide better diagnosing messages, so I can have a better chance of diagnosing the pwd problem without reproducing the user's installation on my punny Windows Test Bed.

— — — — — 


This Release will be published using the following Schedule:

  • GitHub, reaching manual installers and users of KSP-AVC first. Right Now.
  • CurseForge. Right Now.
  • SpaceDock (and CKAN users). Right Now.

Since the change is merely some diagnosing messages, I don't see the need of updating the Distribution Channels progressively.

Link to comment
Share on other sites

Trailing backslash mismatch, perhaps? Or upper/lower case disagreement? I see this is implemented just with String.Equals, which is a rather poor approximation of path equality.

(I'm here because CKAN users brought this false positive message up on our Discord.)

Edited by HebaruSan
Link to comment
Share on other sites

51 minutes ago, HebaruSan said:

Trailing backslash mismatch, perhaps? Or upper/lower case disagreement? I see this is implemented just with String.Equals, which is a rather poor approximation of path equality.

(I'm here because CKAN users brought this false positive message up on our Discord.)

Should not be the trailing backslash, because I'm using IO.Path.DirectorySeparator. (and KSPe path utils always normalise the pathnames to the host operating system before returning to caller - so, nope, in this case the Equals is good enough because the paths were preemptively normalised).

Humm. Yeah, it can be upper/lower case disagreement. I keep forgetting about Windows mess.

I'm trusting KSPUtil.ApplicationRoot is being built at runtime, but if it's hardcoded then indeed we will have mismatches on the comparing if the user renames a directory in the pathname.

I will check this hypothesis  on my Windows Test Bed and come back to you ASAP.

— — POST EDIT — — 

No dice. By renaming any directory in the path where the KSP_x64.exe is on my Windows test bed, the Application Root correctly reflects the change.

The only working theory I have by now is some 3rd party code changing the PWD. But this I can only diagnose with the user's help.

Edited by Lisias
Hit "Save" too soon.
Link to comment
Share on other sites

Join the conversation

You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.

Guest
Reply to this topic...

×   Pasted as rich text.   Paste as plain text instead

  Only 75 emoji are allowed.

×   Your link has been automatically embedded.   Display as a link instead

×   Your previous content has been restored.   Clear editor

×   You cannot paste images directly. Upload or insert images from URL.

×
×
  • Create New...