openBVE 1.5.0 RC1 - BUGS

Delsin wrote:

leezer3 wrote:
The issue here is their train.

Ah, okay, I wanted to make my own version of the train for it from scratch anyway 

Here's the route+train with the [Include] problem I mentioned https://www.dropbox.com/s/qo5wm7voxdw1u1d/M2%20Taksim%20-%204.Levent%20%2B%20Train%20Alstom%201999.zip?dl=0

In that case i will fix it. I've done a entire new exterior model from scratch. Just need a few texture replacement of my new exterior and it's done.

There is no way NYC Community understand OpenBVE need to work on power of two texture size. I leave like more than a year a topic about "OpenBVE modelling guidelines for Modelers!" but they are careless. Users are experimenting White textures by their fault.

The [Include] problem is caused by some textures loaded as part of the station .b3d files.

These look like this:

Code:

LoadTexture,..//Objetos/barandil.PNG,

It's interpreting the double-backslash as a comment marker & hence failing.

Not sure whether these are the fault of the HmmSim object converter which seems to have been used to create these files initially or whether this is something that's been edited manually later.

Either way, the current build contains a workaround.

Whilst doing some route testing of NWM this morning, I noticed that OpenBve 1.5.0.xx does not recognise a final station stop command correctly.

62475,
.Beacon 23;0;0;1,
.sta Hammerwich 05.56;05.5600;T;;1;1;;;1;0,

The T in the .sta command denotes a terminus and there should be no departure bell/buzzer signal from the guard, irrespective how long you wait. After a 30sec wait there is one.

Tony

First a note: Your posted .Sta command is malformed
I'm not sure if this is straight from the routefile or a typo, but it wouldn't be helping.

Code:

Track.Sta Name; ArrivalTime; DepartureTime; PassAlarm; Doors; ForcedRedSignal; System; ArrivalSound; StopDuration; PassengerRatio; DepartureSound; TimetableIndex

The T must be the *second* parameter (departure time), and you've got it as the third (PassAlarm)

T is definitely not valid for use in PassAlarm.



I haven't managed to test this properly at the minute, but will try and do so tomorrow.
Having said that, at the minimum this report has led me to a very obscure signalling related bug present in 1.4.3 also, so that's something useful!

The .sta name has a time in it so that it shows up on the timetable. The arrival time is shown correctly as 05.5600. The departure time is shown as T and the passalarm is blank. This is no different to all of NWM routes on Open Bve 1.4.x and they work correctly.

Tony

Ah, you're right, missed that

https://vps.bvecornwall.co.uk/OpenBVE/Builds/OpenBVE-2017-07-02.zip

That should hopefully work correctly now.
I've also fixed a third related bug, which in some cases caused the reverser to bounce on and off at terminal stations. (Specifically anything using a .Net plugin with the AI driver enabled, e.g. the 323 & UKTrainsys)

Let me know and assuming this works, I'll push a replacement stable build.

That's another bug squashed, Leezer.  It works fine now. Well done.

Cheers,
Tony

1.5.1.4 is now live on the main site

One further station related bug squashed too:
Rolling backwards when departing from a station on a steep hill no longer causes the stop to re-trigger.

Further testing of NWM route diagrams with OpenBVE ver 1.5.x, I have been experiencing the occasional crash. (This is previous to the latest ver 1.5.1.4 incidentally). I don't know when this started to occur, so I cannot give you any clues about version levels of 1.5.x.

Here is the crash log of the latest crash:

03/07/2017 21:38:02
OpenBVE 1.5.521.1130-travis Crash Log
Program is running on the Windows backend
Current screen resolution is: Full-screen 1920px X 1080px 32bit color-mode
Current routefile is: D:\Program Files (x86)\OpenBve\Railway\Route\NWM_Open\Shifted 2001-2005\07.21 [220] 1V16 Maybank TMD-Ludgate Hill via loop XP 4car.csv
Current train is: D:\Program Files (x86)\OpenBve\Train\Voyager_openbve
Current train plugin is: UKSpt.dll
The current routefile caused the following exception: 
System.IndexOutOfRangeException: Index was outside the bounds of the array.
   at OpenBve.Renderer.RenderFace(MeshMaterial& Material, Vertex[] Vertices, OpenGlTextureWrapMode wrap, MeshFace& Face, Double CameraX, Double CameraY, Double CameraZ)
   at OpenBve.Renderer.RenderFace(ObjectFace& Face, Double CameraX, Double CameraY, Double CameraZ)
   at OpenBve.Renderer.RenderScene(Double TimeElapsed)
   at OpenBve.OpenBVEGame.OnRenderFrame(FrameEventArgs e)
   at OpenTK.GameWindow.RaiseRenderFrame(Double elapsed, Double& timestamp)
   at OpenTK.GameWindow.DispatchUpdateAndRenderFrame(Object sender, EventArgs e)
   at OpenTK.GameWindow.Run(Double updates_per_second, Double frames_per_second)
   at OpenBve.MainLoop.StartLoopEx(MainDialogResult result)
   at OpenBve.Program.Main(String[] args)
Current program memory usage: 578.4mb
System memory free: 1944mb

I was wondering if the crashes are due to comments that I often have in the route files? Like so:-

12795,
.SigF 5;1;-2.2;4.2, ;;Banner Repeater
.Freeobj 0;177,

Although, the crashes do not occur everytime in the same place, the one common thing that I read in the crash report is:

System.IndexOutOfRangeException: Index was outside the bounds of the array.

The same routefiles run under OpenBve 1.4.x never crash.

Tony

Hmm, that looks like something is wrong with an object somewhere.
I can't think of anything thats immediately been changed in there, but will try and take a look tomorrow.

My immediate supicion would be that its something in one of your random modules which is only being loaded on some runs

Do you have any reasonably reliable routefiles which crash often with the current public NWM version?
I can always turn on the AI & wack up the time acceleration factor & see what happens

The shorter diagrams rarely crash, so you may be right about some random rogue object. The 10.32 freightliner Maybank-Wellingham is the one that will usually crash somewhere along the way it seems. And I suppose any other diagram that works the same route, although I haven't got that far with testing. I have never suffered with system crashes using OpenBve 1.4.x, and I appreciate that a lot of randomness has been built into NWM, so finding the causes of a crash might be difficult.

I will be unavailable most of the day, but I will be back this evening.

Cheers

Tony

At the minute, I can't get anything to crash at all, this is with a copy of NWM 2014 from Trainsimstuff, which I then dropped the most recent version over the top.
I've tried running various routefiles on AI mode with a mix of trains, time acceleration from anything from 5x to 50x.


Your crash log seems a little odd, as it thinks that the crash occurred during the loading process, but from what you're saying, it's happening as you're driving?
I don't suppose you have any logs which have a track position in them do you (Those generated whilst driving should show the current track position, viewing distance etc.)?

I can't see anything immediately wrong in the function which crashed, which is the main drawing function used by everything, so doesn't really tell me a lot.


The best answer I can currently give you is that whatever's causing this is going to very probably be located at the following position:
* Take the track position you crashed at.
* Move to the next 25m block boundary. (e.g. Crashed at 105, therefore the boundary is 125)
* Add on the viewing distance. (e.g. 600m viewing distance gives ==> 725)
* The object causing issues is *likely* to be somewhere between 725 and 750.


If I can reproduce, I'll fix, but I'm shooting blind at the minute

Edit:
Have also tried a completely clean copy of the current version, no crashes there either.

I wonder, how does an "index out of array" occur? I mean, the array is created by the program itself, so shouldn't it know beforehand whether a specific field does exist at all? Sorry if it's a noob question, my programming skills are more of a tinkering and on-the-fly-learning approach...

hi i have downloaded the latest ver which comes in a windows installer and when i try to run it it has a virus according to panda,= which gets rid straight away, can anything be done about this.  by the way it was from the main site.thanks jckhinks.

Quork wrote:I wonder, how does an "index out of array" occur? I mean, the array is created by the program itself, so shouldn't it know beforehand whether a specific field does exist at all? Sorry if it's a noob question, my programming skills are more of a tinkering and on-the-fly-learning approach...

Short answer is that something's been overlooked somewhere

Without getting horribly technical, the best guess I've currently got is that there's an object containing a mesh which is malformed in a unusual way.

TLDR:
Just for example:
In order to make up a mesh, we've actually got several distinct arrays.
One contains the positions of all verticies.
One contains the faces, texture references and associated texture coordinates.
The final one contains the textures themselves.

In order to render something, we itinerate through the meshes's list of faces, pulling the appropriate verticies and texture-coordinates as we go.

Something in one of these arrays is broken, and there are 100s of ways this could have happened (Loading error, positional update gone wrong, wrong animated object state etc. etc......)
Just abandoning the face in question when we encounter the error doesn't work either, as our openGL state will then be broken, which causes all manner of fun when we try to render the next face.

I'm still trying to reproduce, but nothing at the minute.
Will try the laptop when I get back from work, but I can't see a reason for this to be system dependent.

jckhinks wrote:hi i have downloaded the latest ver which comes in a windows installer and when i try to run it it has a virus according to panda,= which gets rid straight away, can anything be done about this.  by the way it was from the main site.thanks jckhinks.

https://www.virustotal.com/en/file/627952a8f4f06a07a33773c86cdec111d6dbab38b78a7dc32d14f88b962ac5f3/analysis/1499172663/

This would appear to be a heuristic detection.

Not a lot I can do with these; Panda may well decide it's safe after a few more users have seen it, otherwise you'll have to unblock it manually.

hi thanks for that .

Hey y'all.

Sorry for bothering, but what's keeping OpenBve dependant on .NET / Mono? I'm curious and I kinda want to port it to Android or Haiku OS...  Plus, a graphics based interface would look much better than WindowsForms in my opinion!

Cheers.

Edit: I do understand this is not a bug, but I felt it wasn't worth creating a thread just for it.

Glory! koshikii wrote:Hey y'all.

Sorry for bothering, but what's keeping OpenBve dependant on .NET / Mono? I'm curious and I kinda want to port it to Android or Haiku OS...  Plus, a graphics based interface would look much better than WindowsForms in my opinion!

Cheers.

Edit: I do understand this is not a bug, but I felt it wasn't worth creating a thread just for it.

Just the fact that you're asking the question means you don't understand it (And I don't mean that in a nasty way either.....)

.Net / Mono is available on Android, and the main form interface is actually a rather small part of things.
The trouble with porting to anything other than Windows and x86 compatibles is rather the fact that the entire graphics engine is written around intermediate mode openGL, and to change it essentially requires re-writing most of the program.

If your platform supplies a software intermediate mode openGL wrapper (e.g. Pi3), it'd probably be possible to get low-end BVE2 routes running, but further than that requires the engine re-writing.

Just found out that Reversed = True in any [bogieN] in extensions.cfg file crashes BVE in startup, even with the latest build, in the same time it works okay in [carN]

Delsin wrote:Just found out that Reversed = True in any [bogieN] in extensions.cfg file crashes BVE in startup, even with the latest build, in the same time it works okay in [carN]

Can't reproduce this one.

Can I see a copy of the train please?
My initial guess would be that it's something to do with the specific object (or object type) you're using, but it's far easier to find the bug with the broken example to hand

https://www.dropbox.com/s/eislxcflzhgmx6g/train.zip?dl=0 the train, don't mind the simplicity of the exterior, it's a WIP from like a year ago    when I "disable" Reversed = True for bogies, it loads properly

Thanks, found and fixed.
Have also spotted a bug where the axle locations of the front bogie were being used on the rear in one place.


I've also made a *lot* of changes to the exterior load/ update process to make the TrainManager class a lot smaller, and generally improve a bunch of somewhat iffy code....

I don't think anything new is broken, but please keep half an eye on things

https://vps.bvecornwall.co.uk/OpenBVE/Builds/OpenBVE-2017-08-13.zip

Thanks!
Also, is lever position display in bottom left corner intended to become white for power notches instead of previous blue? Appeared only with the 11 Aug build (maybe sometime earlier)