Jump to content

Padishar

Members
  • Posts

    2,669
  • Joined

  • Last visited

Posts posted by Padishar

  1. 29 minutes ago, kball said:

    Would love to be able to try this out on my mac laptop. Has anyone figured out a configuration to enable the hotkeys?

    What is the problem?  You can change the key (used with the standard KSP "modifier" key, which I believe is LeftCommand on a Mac) for each of the functions in the GameData/MemGraph/PluginData/MemGraph/settings.cfg file.  Just use whatever values from this page that you want (if you set something, such as the keyRunTests one which is only really useful for developers, to None then it will be disabled):

    https://docs.unity3d.com/ScriptReference/KeyCode.html

  2. 10 minutes ago, tg626 said:

    Yes, I switched vessels several times both by "[" keys and thru clicking "Switch" on the map.

    It froze, after a fashion windows popped up a dialog box stating "KSP_64.exe has stopped responding", etc.

    IMPORTANT UPDATE:

    I goofed, I *assumed* that it was a dead kerbal, as one had just died.  I forgot about the 2 outside back on Kerbin.  

    So what REALLY happens is, I'm controlling a ship orbiting Eeloo, and I choose a kerbal on Kerbin as my redezvous target, and then BAM crash.

    Ahhh, the symptoms sound very much like what I was describing in this post:

    However, I've only seen that happen when actually switching vessels (by any means, KER or stock) with the rendezvous panel open.  Are you certain it happened when you selected the target or did you click the button to switch (possibly a double click, the button may appear under the pointer when you select the target)?  I did revert part of a change I made to fix a problem with setting a target because it was causing the panel to not resize itself properly so that could also be an issue.  It sounds like I'll have to instrument this bit of code quite thoroughly and release a dev version to try to gather some more useful data...

  3. 4 hours ago, tg626 said:

    BUG:  I had an EVA Kerbal die, later on, I wanted to select another kerbal as a rendezvous target and saw the dead one's name in the list.  For kicks I clicked the dead kerbals name and the game crashed

    Strange, if the kerbal died then I don't think his "Vessel" should've existed anymore.  Did you switch vessel or otherwise change scene between him dying and you clicking him in the target selector?  When you say the game crashed, what do you mean?  Did KSP just close or do you mean something else?

    4 hours ago, tg626 said:

    You'll also see a load of NREs related to maneuver nodes, which it wouldn't allow me to set or remove any more - and wheel colliders which I think was because of a lander crashing and exploding (landing legs destroyed)

    Yes, there are a lot of these being spammed:

    [WRN 21:23:56.204] KK: StaticDatabase.updateCache(): vPlayerPos is still v3.zero

    ...and also quite a few of these:

    [EXC 21:24:00.059] ArgumentException: GUILayout: Mismatched LayoutGroup.Repaint
    	UnityEngine.GUILayoutUtility.BeginLayoutGroup (UnityEngine.GUIStyle style, UnityEngine.GUILayoutOption[] options, System.Type layoutType)
    	UnityEngine.GUILayout.BeginHorizontal (UnityEngine.GUIContent content, UnityEngine.GUIStyle style, UnityEngine.GUILayoutOption[] options)
    	UnityEngine.GUILayout.BeginHorizontal (UnityEngine.GUILayoutOption[] options)
    	MuMech.MechJebModuleManeuverPlanner.WindowGUI (Int32 windowID)
    	MuMech.DisplayModule.ProfiledWindowGUI (Int32 windowID)
    	UnityEngine.GUILayout+LayoutedWindow.DoWindow (Int32 windowID)
    	UnityEngine.GUI.CallWindowDelegate (UnityEngine.WindowFunction func, Int32 id, UnityEngine.GUISkin _skin, Int32 forceRect, Single width, Single height, UnityEngine.GUIStyle style)
    

    These:

    [LOG 21:57:55.609] dT is NaN! tA: NaN, E: NaN, M: NaN, T: NaN
       at System.Environment.get_StackTrace()
       at Orbit.GetDTforTrueAnomaly(Double tA, Double wrapAfterSeconds)
       at Trajectory.UpdateFromOrbit(.Orbit orbit, Int32 sampleCount)
       at PatchRendering.UpdatePR()
       at PatchedConicRenderer.LateUpdate()

    Lots of these spamming:

    [LOG 22:02:17.419] [MechJeb2] Warppause : lastAskedIndex=5 CurrentRateIndex=4 WarpMode=HIGH MaxCurrentRate=5

    Quite a few of these:

    [ERR 22:27:23.939] [DestructibleBuilding]: ID for this building is not defined correctly. Cannot unregister.

    This followed by lots of the NREs:

    [ERR 23:02:19.632] Can't remove Rigidbody because KSPWheelController (Script) depends on it
    
    [EXC 23:02:19.633] NullReferenceException
    	VPWheelCollider.UpdateVisualWheel ()
    	VehiclePhysics.VehicleBase.LateUpdate ()
    [EXC 23:02:19.642] NullReferenceException
    	VPWheelCollider.GetGroundHit (UnityEngine.WheelHit& hit)
    	VehiclePhysics.VehicleBase.CalculateSuspensionTravel (VehiclePhysics.WheelState wheel)
    	VehiclePhysics.VehicleBase.FixedUpdate ()

    Then there was a final vessel change and the last thing that happened was an autosave.  Basically, it looks like KSP got seriously upset by something which upset both MJ and KER.  This bad state may have been why the kerbal didn't get deleted properly or it may have nothing to do with it.  Either way, it may be possible for KER to detect this bad state and avoid listing vessels that could cause a problem but it will be awkward as there is no indication of the click to set target at all so it's tricky to determine what could've been wrong with it...

  4. Just now, fr33soul said:

    so i can extimate the distance covered with the rocket in order to know if it can reach a designed planet.

    That has nothing to do with what Scott Manley is calculating...  I suggest you go and read/watch some general "rocket science" tutorial threads/videos as one of the first things you should learn is that deltaV is (by far) the most important value that can be calculated and is the only number that is really important when deciding if you can reach a particular body...

  5. 21 minutes ago, fr33soul said:

    I'll suggest to add the "Distance covered" measure by a rocket in the KER - VAB stats,

    Why?

    22 minutes ago, fr33soul said:

    in order to know how much km the rocket will cover.

    What possible use is that?  It just tells you how far the rocket goes while the engine is burning, which isn't exactly high on the list of important numbers where rocket performance is concerned...

  6. 9 minutes ago, NeuroticGamer said:

    Is there any value in me installing (I have a legit license via my employer but never had reason to use it) JetBrains' dotMemory profiler?  https://www.jetbrains.com/dotmemory/features/

    Would it help in identifying the heavy hitters in terms of garbage generation?

    No, KSP doesn't use .NET, it uses an old version of Mono built in to the KSP executable.  However, the Unity profiler can be used to do the job instead, see this thread for details on how to set it up...

     

  7. 5 hours ago, Jebs_SY said:

    I just debug my 100-mod 1.2.2 setup, cause allocation rates around 30-40MB/s are not nice... and I wonder, would it be possible/useful for memgraph to show/calculate the average allocation rate between the 2nd last and last garbage collection? Maybe only when the memgraph-GUI is shown on screen. Do you think this is useful?

    Well, it wouldn't be difficult to total up the memory allocated as it goes along so it needn't be slow.  However, because of the inaccuracy inherent in the readings that MemGraph takes due to how the Mono heap and the padding mechanism works, I'm not sure it would be useful enough to justify the effort.  The real allocation rate is actually what the graph displays in the time immediately before a collection, the average for the whole interval would often come out a lot lower than the real rate.

    5 hours ago, Jebs_SY said:

    EDIT: Another interesting thing to compare different setups would be the interval. However, interval alone doesn't help, cause the heap can be different big. So maybe Heap-Delta (Max-Min) would be useful, too. Or maybe even Interval / Heap-Delta. This should result in a comparable performance number, or?

    Again, all the values can be rather inaccurate and combining two such values will make the result even worse, so I can't see it being much more useful than just making a judgement from the graph and values that are currently displayed.

  8. 1 hour ago, Argelle said:

    Where am I calculating wrongly? The 4 BACCs do not contain 30700 fuel?

    Exactly.  The 30700 is the mass of all the parts that get decoupled at the end of that stage so it includes the "dry" mass of the SRBs and the mass of the decouplers.  If you want to know the mass at burnout then you can either do a "verbose simulation" in the settings and dig the info out of the log file or you can simply tweak the fuel in the SRBs to 0 and see what KER gives for the overall mass (you'll probably need to turn on "All Stages" as that stage won't generate any deltaV and will be hidden)...

    Edit: also, the suggestion to add the actual fuel mass to the display has already been made, it just isn't easy to squeeze it in without needing to make all sorts of other adjustments to the window...

    Edit 2:  Each BACC is 1500kg when empty and each decoupler is 25kg.  This makes the mass at burnout 26195kg which gives the TWR of 3.89.  Note, you should really use 9.80665 for g as that is what the game (and KER) uses, it can sometimes affect the second decimal of the TWR...

  9. On 12/9/2016 at 7:36 PM, RentedTritium said:

    In this version my game locks up every 6th second.

    This is almost certainly a garbage collection induced stutter.

    On 12/9/2016 at 7:36 PM, RentedTritium said:

    Probably it's a mod checking something inappropriately though since it seems like I'm the only one.

    Stock installs also suffer from garbage collection induced stutter.  You certainly aren't the only one, try searching the forum for the word "stutter" and then go look at the thread linked in my sig...

  10. 2 hours ago, fr33soul said:

    it was on the previous version of this mod:

    Ahhh, by "previous version" you mean the previous major version which was last updated in October 2014 (and the version you show the screenshot of was released about 18 months before that).  Actually, that's probably where I was remembering it from, too... :wink:

    Anyway, I've added a readout (currently in kPa) in my latest pull request...

  11. 1 hour ago, Oshika said:

    If i understand well (because my english is not perfect) i have to use the mod+end function of your memgraph mod to remove the stutter or to reduce it until the devs found a solution/fix that (are they even know this issue ?) ?

    Yes, you understand.  The devs do know about the issue, they have been making significant improvements in the amount of garbage the stock game generates in both the 1.1.x and 1.2.x versions.  Mod developers are also, generally, aware of the problem and most have been trying to make similar improvements.

    1 hour ago, Oshika said:

    I have 16gb of ram, if i put a lot of mods (well, Being careful that the mods are compatible and up to date) can i use the "total" line of padheap.cfg to allocate more ram and be able to remove more stutter ? what value should i use ? 8192 ?

    Yes, you can simply adjust the total line to make the stutters further apart.  On a 16 GB machine with a lot of mods I would recommend not going much above 4096 or so, as many mods also use up memory for textures and models which are kept outside the Mono heap.  4 GB of padding should put the stutters pretty far apart unless your game is generating a really huge amount of garbage.  I would actually recommend starting with 2048 and see what effect it has, then increase it if necessary...

  12. 20 hours ago, Oshika said:

    Edit : Ok so even with a fresh install and only memgraph as mod, i have stuttering, this is insane seriously i don't understand : 

    This is just how the game engine works combined with code that generates too much garbage.  Basically, Unity/Mono keeps a certain amount of free headroom in the heap and once this fills up the garbage collector is triggered which causes a short pause.  The headroom is a certain fraction of the total size of the heap so when less memory in being used there is a lot less headroom and the stutters can happen quite frequently.  The MemGraph heap padding feature both forces a large amount of extra headroom and, due to the increase in total heap size, makes Unity/Mono allocate more headroom itself.

  13. On 12/8/2016 at 4:54 PM, Oshika said:

    I finally re-installed my game, no mods, only the memgraphics, and no more stuttering, then i put only a few mods and the stuttering come again (every minutes) but i want to remove it completly, is it possible ? 

    Sorry for not replying to this sooner, you have a very large list of mods and it's likely that several are each contributing significantly but it's hard to say which ones.  With a mostly stock install it should be possible to use the MemGraph heap padding feature to virtually eliminate the garbage stutters as long as you have enough physical RAM in your machine and the "Last" value at the top of the graph isn't showing a huge number.  A screenshot of the graph window in this new install could be helpful, leave it running in flight for a little while with the stutter and then hit Mod-End, leave it running for some more and then take a screenshot.  If that doesn't improve it enough and you have a reasonable amount of free physical memory while running KSP then you can change the total= line in the padheap.cfg file in the MemGraph/PluginData folder...

  14. 10 hours ago, Aristarchus said:

    Where can I submit a bug report? I noticed the Time to AN was behaving weird, it looks like it is jumping ahead one orbit when the ship crosses periapsis. Time to DN is behaving fine as is Angle to AN.

    Here is usually fine, or you can add an issue in the GitHub repository linked in the first post.  That readout does look wrong, though it appears to be as the ship passes the apoapsis rather than the periapsis.  I'll take a look at the code concerned and see if I can see what's wrong.  There was a change in KSP 1.2 to the range returned from some orbit related functions so this may be the cause if this is a new problem...  It would be very handy if you could copy your save and delete all the other vessels, just leaving the one you've shown with this orbit, and then upload it.  Just in case the issue only happens with certain configuration of orbit and proves awkward to reproduce...

    @Kertech, @kball and anyone else reporting problems, please include screenshots and/or craft files and/or save files and a detailed description of what you think is wrong...

  15. 5 minutes ago, razark said:

    I haven't seen anyone specifically say this is or isn't working in 1.2.2, but there's a user up this page that reports it seems to be working.

    There hasn't been an official release for KSP 1.2.2 yet but KER 1.1.2.8 should work as well as it did in KSP 1.2.1.  The link in my signature is to my test build of the latest fixes I've done (vectored thrust and staged docking ports) built against the KSP 1.2.2 assemblies...

  16. 2 hours ago, Red Iron Crown said:

    Looks like emirkustirika is using a docking port as a decoupler (can't tell if it's in the staging list or not due to cropping). AFAIK KER ignores docking ports in the staging simulation, it would assume the craft on the left won't dump the empty tank on top when it's empty.

    Staging for docking ports is toggleable, perhaps KER could detect if it is enabled and account for them? (I have no idea how easy/difficult that is.)

    Yes, that's what I assumed too.  It turned out to be quite simple to make the simulation stage docking ports sensibly so I have submitted a pull request...

×
×
  • Create New...