New OpenBVE Build- Testers Please

I did some testing on those route you tried and I get a lot of bouncing:

Graymac's new 25/ Cheltenham- I get a lot of bouncing and I loaded it a few times and every time it bounced with at least one carriage but usually more. I get less bouncing (but not always) if I don't limit CPU usage.

Class 104/ Plymouth route- I did not bounce and I tried another few times and it wobble quite badly with 2 out of three 3 carriages derailed but also only 1 carriage other time I loaded it. When I don't limit CPU usage all the carriages wobbled and then reloading it and 2 out of 2 wobbled badly.

323 with 3D cab/ Birmingham X-City- No wobbling but details in the 3d cab were not rendered properly with bits missing and holes in the cab which you can see through to the outside.

2700 with BWR 0920 Ballyfeckin-Ballinamo- No wobbling and displays correctly.

Scratches head....
I know it'll be something overflowing somewhere in the timings for the physics, but I'll need to catch it doing it first
Can you please try jumping between stations a few times and seeing if it happens then?
Would be useful to know if it's only on first load, or whenever the train gets repositioned


I'll try on the laptop tomorrow and see whether that gives any clues.

Cheers

Chris Lees

http://www.bvecornwall.co.uk

It does not happen when jumping between the stations which I tried a few times if the train is not wobbling when I jump stations. However that it not always the case with it wobbling at the new station even if has stopped wobbling at the first station. But if it is wobbling when I jump stations it will continue to wobble at the jumped to station.

Hmm, this is a blind attempt at a fix:
**Newer build further down**

It adds a small change to always force a simulation update before the first frame is rendered.
Now in theory this shouldn't make any difference, but we'll see what happens.

Still trying to reproduce, but no luck.
I've tried with and without CPU limiting enabled
323's 3D cab looks OK to me here, but I'll investigate further if we can fix the bouncing.

Cheers

Chris Lees

http://www.bvecornwall.co.uk

Sir Chris Lees
A train plugin ats_os1.dll can not work in your new version both 1440d and the 1440e
This ats_os1 plugin does work for trains in my version old 1430

And also the version 1440e does have a train coach [car1] behind locomotif derail at loading before any move.  It is strange.

Good day and night for you
Hijau

I tried 1.4.4.0.e and I still get the bouncing and it is about the same as before. I would not want to be a passenger on that train!

I think it might be related to how detailed the route is with it not as bad or does happen on a more simple route compared to a more detailed route. I still get the issues with the 3d cab but no wobbling with the X City line.

It's a pity you can't reproduce these issues. I have not had these issues in the past and I have driven many routes and trains.

OS_ATS1.dll is working just fine here
I've just fixed a bug with transparencies in 3D cabs under certain lighting positions, which I'm hopeful will fix the glitches Matt was seeing, but I still can't reproduce the bouncing.

**See below**

I've made another blind shot at preventing it with a hack in the toppling function, but I'm not hopeful with this one.

Edit:
May sound like a stupid question I know, but can you check if it goes away when you turn off toppling and derailments please?
That *may* shed a little light on things, as to what the sim thinks is going on.

If this doesn't work, I'll try posting a build with the entire shaking code pulled out, which I think is what's causing it; If you load a route with 1.4.3, you'll note a slight bounce of the cars at the start of the route, which seems to have got out of hand here.
 If that works, we can start some deeper digging, but not being able to reproduce it here is a massive pain

Cheers

Chris Lees

http://www.bvecornwall.co.uk

Un setting toppling and derailments makes no difference, I still get the bouncing with 1.4.4.0.e.
I then tried 1.4.4.0.f and I did not get the bouncing at all. 3 out of 3 tests with Class 25 and Cheltenham and 3 out 3 tests with Class 104 and Plymouth loaded without the wobbling. The 3d cab rendered correctly on the X City line.

Sir Chris Lees
Yes here good news also.
My errors.  My PC did loose its brain with plugin was not working any version
PC in need of reboot, now ATS_OS1 working in all versions new and old

I did also download and run your 144f with the good news
Train did not wild sway left and right and derail after loading
Train sway is very little. I do not like swaying when train stopped of any version and do wish it did not.  I wish for option menu to disable sway off

I thank your for your version 144f, it is good for me
Good day and night for you
Hijau

That's great

Really, I'd prefer to fix the root cause of the problem, but that's not happening at the minute as I simply can't reproduce it, and I think the current hack this will probably do for the meantime.

This is a slightly newer build, which adds null checks to a couple of places in the texture API:
**See below**

It also adds progress text back to the loading screen.
Nothing really interesting unfortunately, but this cures a couple of very obscure errors I've run across whilst running the debugger. It also means that it actually loads in GEDebugger now
(Passing null texture handles to OpenGL is a recipe for trouble....)


Cheers

Chris Lees

http://www.bvecornwall.co.uk

With 1.4.4.0.g the F3 camera radius, and pitch values are not working correctly. The cant is displaying the values as travel along the route but the radius does not change from 0 and pitch value is not shown.

Thanks for your efforts with this, so far it looks good though I have not driven a whole route just testing the loading and driving a few minutes of the route.

Sir Chris Lees

I do have the question of your new datatype CurveRadius[CarIndex]
?Does this value equal in meters?
Does this value equal infinite meters in not curve but tangent straight away
or does this value equal to zero with a tangent straight away?
I do ask this question as my equation for bogies turning it does explode to error with infinite meters.

Good day and night for you
Hijau

If the line is straight, the value is zero.
On a curve of 500m radius, the value is 500, curve of 150m radius, the value is 150

To be specific, it returns the value set by the .curve command for the block in which the specified axle is.


The F3 follower hasn't yet been fixed, it's on my list of things to sort out- I spent far too much time last night chasing the OpenGL error above.

Cheers

Chris Lees

http://www.bvecornwall.co.uk

OK, so I haven't yet fixed the F3 follower (Sorry, forgot....)

This build however enables joystick support:
http://www.bvecornwall.co.uk/downloads/beta/OpenBVE_1440h.7z

There's one issue-
Only the first attached joystick is recognized correctly, even though the dialog implies that more than one is possible.
(OpenTK appears to have a bug/ 'feature' so that when an invalid joystick is enumerated, it returns the first connected joystick)

It also includes a newer self-compiled version of OpenTK, as joystick hats are broken in the current release version. This may be very slightly faster than the previous builds, but really there shouldn't be anything in it.

Testing much appreciated

Edit:
For good measure, here's my source code to date:
https://github.com/leezer3/OpenBVE

I haven't been tracking the changes until this point, so it's a fresh lump of code.

Cheers

Chris Lees

http://www.bvecornwall.co.uk

Sir Chris Lees
I do thank you for your version 1440h 
I did download and did a first test run with the good news
I find no differences in running from the version 1440g
My PC does not possess the joystick so I can not speak of this

As in the version 1440h before, my PC has 80 fps interior and the 40 fps exterior
with the stacatto when using keypad numbers and exterior roaming

It is good for me and will now use new version replacing old 1430
Good day and night for you
Hijau

1.4.4.0.h works fine for me, but sorry I can't test with a joystick. I did unzip this version over the top of the old version which froze the program when I pressed a key. So I deleted the files from the folder and unzipped this version to the same place and it did work correctly.

Chris, I took OpenBVE 1.4.4.0h for a testrun and seems to be doing just fine. I'm glad I can resize the Window while playing and also that I don't have to go and download SDL.

Just a minor thing, not important in any ways, it'd be nice to see the OpenBVE icon in the top left corner of the app window.

I'm glad this appears to be working reasonably well for most people!

I will note that there are still rather a lot of things I haven't got as far as debugging properly

Current known issues on my list to fix:

• Signals not appearing on Linux (Path resolution problem means that the objects aren't loading correctly)
  
• Window icon
  
• Only the first attached joystick works correctly
  
• F3 curve radius values wrong
  
• *Not an issue as such*- Homepage/ forum links in the sim/ source are currently dead.
  

As you'll see from the couple of commits I've made today, I've started on the process of working through the existing code and eliminating logic problems, and generally trying to tidy the code up.
Whilst none of this should really affect anything from the end-users point of view, it helps immensely when trying to find odd issues.

At this point, I think it's also very much worth pointing out (again) that this is a work in progress.
Whilst I've got a reasonable idea of what's going on inside the engine, I don't claim to be an expert, and I'm *certain* there are issues that I haven't yet found other than those listed above.

Use for production purposes at your own peril

Cheers

Chris Lees

http://www.bvecornwall.co.uk

Another build; This is specifically to fix bugs on non-Windows platforms:
http://www.bvecornwall.co.uk/downloads/beta/OpenBVE_1440i.7z

This should fix the following issues:

• Signals not appearing on Linux.
  
• Main menu crashes when selecting an item using the mouse on Linux.
  
• Previous builds didn't work on Linux at all, as they didn't like the timer function- Implemented a Linux specific alternative
  
• PNG files recompressed to avoid libgdiplus crashing on certain Linux versions.
  

This has been tested on Ubuntu 14.04.

Cheers

Chris Lees

http://www.bvecornwall.co.uk

Oh, that wonderful libgdiplus bug we had for years! Will not miss it.

I'll take this for a test on Ubuntu 15.10 and report back.

So here is my experience with OpenBVE 1.4.4.0i. First of all, my hardware is:

• Quad core i3 CPU
  
• 8 GB RAM
  
• Nvidia graphics card with open source driver
  
• Ubuntu 15.10 64 bit live cd
  

I installed OpenBVE from the Software Center to get the dependencies. Then I installed OpenBVE 1.4.4.0i and ran it via mono. The start screen loaded fine but the Browse section for both routes and trains were grey. This has been such for a long time. Once I went to the Latest tab and back it was OK, routes were visible.



Then I started a route and played it. Everything was working fine. Then I resized the window while playing. When I resized the window towards the right edge of the screen that part became black. When I resized it towards the left edge of the screen the resizing worked well.



Then I closed OpenBVE and started again. The start page loaded fine. Then I clicked on the previously loaded route again and OpenBVE start screen crashed and disappeared. This is what the Terminal printed for me:

http://pastebin.com/Y4Cvm2Fp

Now that's typical, I thought I'd actually got that crash nailed down  
It does it too on the last official build if that's any consolation.

Selecting a previously used route using the Tab key will stop it crashing, but that's the best I can really suggest at the minute. It's never been 100% repeatable here, and it doesn't crash under Windows (Mono or Microsoft runtimes), and assuming it's the same as it's always been, adding a breakpoint under Mono-Debug also stops the crash, which means that it's nearly impossible to debug. (Specifically, I think there's some sort of race condition somewhere...)

Will do a little more digging and see what I can come up with.

Cheers

Chris Lees

http://www.bvecornwall.co.uk

Update- I can't reproduce the crash at all on 14.10 (32-bit)
1.4.3.0 crashes reliably under the same circumstances, so the problem is at least partially diminished.

Dependencies wise, the only things not installed on a default Linux build should be:
libopenal1
Mono

I'm in the process of rolling a 15.10 64-bit VM and will see what I can find.

Edit:
Any chance of a copy of the file UserData\Settings\options.cfg after a crash as well please?

Cheers

Chris Lees

http://www.bvecornwall.co.uk

Here is my Options.cfg, again on Pastebin:

http://pastebin.com/UU1RGUyB

The app crashes reliably, no chance of running the same route again. This is on a live DVD, don't know if there had been any fixes related to this but I highly doubt it.

The terminal output talks nicely about Windows Forms:

http://pastebin.com/3rgVxcug

Extra info: after I run another route, I'm able to play the formerly crashing route. The point is that I can't play the same route twice after each other.

What is more, I cannot play any route files from the same directory after each other. I mean, I played Nagyirtás-Nagybörzsöny. I closed OpenBVE. I tried playing the other direction of the route, Nagybörzsöny-Nagyirtás and it crashed too.

Later any route that I selected from the latest tab crashed OpenBVE.