I have managed to reproduce it using vitaly's method. Thank you for this report and reproduction. It's a bit mysterious, something to do with accessing memory after it was freed. But I'll track it down as top priority.

EDIT: Now I know the cause of it.

EDIT2: It seems to me this bug is not new, and has been there ever since shadows could be cast on layout objects, I think when the concrete objects were introduced.

EDIT3: Now I've found a seemingly more reliable way to get the crash, and in this case it takes place in the new code that selects the correct draw lists depending on your height above ground. In this method, you do not wait for the car to go out of physics. Just use in car view and pause the game before entering SHIFT+U mode. Then delete the object the car is sitting on. You should get a crash when you exit SHIFT+U mode.

EDIT4: Fixed in my version. There are a few quick things on my list that have come up so there should be a test patch out sooner than expected.

Last edited by Scawen, Wed 17 Feb 2016, 21:01.

Quote from Pose1don :

Okay but there was handbrake on MRT for years in LFS. Why the change now?

Thanks for explanation.

There was previously a handbrake on all the single seater cars in LFS. But there are no handbrakes on single seaters in reality. Some people were using the handbrake to gain an advantage in some places while hotlapping. This forces other people to use the same unrealistic trick to remain competitive and there was a request on the test patch thread to remove this exploit while we were on some other hotlapping improvements.

Hi sinanju, thanks for the test.

First of all I see I made a mistake with the folder name. It's actually data\training and not data\lesson. So I've edited the first post with that correction. Hopefully that should solve the problem you found and you should see those files, ready to load.

I see what you mean that my description could be quite confusing. I've changed the "How a training lesson is constructed" section to be more informative.

It's interesting the way you have written your version, in a step by step way that is probably easier to understand. Anyway for now at least I have added a bit more info. I know it needs more, like a description of the format for the text file.

About your last question, yes, if you loaded the layout and image file into your lesson, then when you click "save" at the bottom of the screen, the saved file is a .lsn file, representing the lesson itself and that would be something you could add to the list.

Test Patch 0.6K23 is available with many improvements!
https://www.lfs.net/forum/thread/88821

K23 is available on the public test patch thread!

Thank you for the testing, design and suggestions!

https://www.lfs.net/forum/thread/88821

THE PLAN!

I plan to release an incompatible test patch tomorrow evening (Thursday).

Some of you may wish to try the current incompatible version, K21, on the programmer thread. Tomorrow's will be similar, and with the Rockingham with the sunset weather, and I'll see if I can squeeze in another update or two before the test.

I don't recommend that patch for everyone, but if you have an interest in the layout editor or InSim, then it might be good to have a look.

The change log would be a good place to start.
https://www.lfs.net/forum/thread/88999

After Thursday's Test Patch, the plan is to do a full release on Saturday.

K14 is now available in the first post : https://www.lfs.net/forum/thread/88999

It doesn't seem like a big update but in fact it is. The best solution for the duplication bug, when moving InSim checkpoints quickly and repeatedly in multiplayer mode, was to change all the marshall code to support multiple object selection and use similar code to the other types of objects (and share some code). They can't share all the code because these circles internally are really a completely different type of object.

It turns out this was a 12 hour job, because it is complicated how this multiplayer editor works, but I think it works now. I'm not fully confident about that so I'd be grateful if you could have a go and see if it stands up to a test.

Layout editor :

Multiple object selection is now available in marshall mode

Fixes :

Quick repeat moving checkpoints in multiplayer caused duplication

The incompatible updates are listed or discussed on the programmer test thread and to me they are interesting! Yes, most of the changes need InSim to use them but the changes offer some interesting possibilities for online racing, and that's why I'm working with server admins and InSim programmers on that thread. It's all about helping to support good online racing.

Thank you for the test.

I can't think of a reason why Airio would think you are on multiple times. I've checked the code in LFS where it sends IS_NPL and don't see any change there (if join requests are not enabled - and they are not enabled with version 6). It should be compatible.

You can receive another IS_NPL if you are already in the race but that was always the case. Subsequent NPL for the same player use the same PLID (if he has not spectated) so the external program knows you are leaving the pits rather than joining.

I don't know if you have any time for a boring test but maybe you could try reproducing it by entering / leaving pits and also spectating. If you get the Airio bug, you could try the same with a public version and see if the same sequence of joins results in the same bug.

Thanks, got it and the fix will be in K12.

EDIT:

K12 is now available.

It's another incompatible "sub-test" patch in the programmer section. The test patch is mainly for programmers until Thursday when I plan to do a 'proper' incompatible test patch here.

InSim programmers should definitely have a look. Server admins might like to have a look at the new grid manipulation buttons.

Just remember it's not compatible with what most people have. https://www.lfs.net/forum/thread/88999

Last edited by Scawen, Wed 3 Feb 2016, 17:55.

Another incompatible "sub-test" patch K11 on the programmers thread. It's mainly for programmers until Thursday when I plan to do a 'proper' incompatible test patch here.

But InSim programmers should definitely have a look. Server admins might like to have a look at the new grid manipulation buttons too.

Just remember it's not compatible with what most people have. https://www.lfs.net/forum/thread/88999

Well, that is the "Borderless window system" referred to in the first post. It was a major focus on my side of this series of test patches, several days work on that.

Borderless window system :

New SHIFT+key functions are displayed on right of Screen Options
Borderless modes allow ALT+TAB to other programs without minimising
Multiple monitor support is greatly improved by a borderless window
New command /window min/max/monitor/virtual (no parameter=restore)

Watch out though, borderless windows have worse performance than exclusive mode full screen.

Yeah, sorry about the RO hotlap confusion during these test patches. Just inevitable as some parts of the track surface were fixed. We plan to release the full version on Saturday and that should be stable for a while.

Quote from sicotange :

Thanks for adding this new packet Scawen. It works like a charm (I haven't tested all features sufficiently though).

Thanks for the test. Please let me know if it seems there is enough to do your ideas like multiple races in one environment, or if something else would help. Also the behaviour when a player gets /ujoin on his username while he is in the garage. How would this affect things in your case?

Quote from sicotange :

One thing I noticed. The ground level check is on by default. If it is an easy addition, would it make sense to add the no ground level check option to StartPos? Or would this go too much in the direction of "surrealism"?

To help me understand, what would this be used for? I think LFS always does some kind of ground check to avoid dropping the car too far. But as far as I know, the current system should work on things like concrete objects and doesn't have to be the actual ground.

Programmers and other people with a related interest, please head over to the programmer forum where there is a special test available.

NOT COMPATIBLE! Only for specific testing! https://www.lfs.net/forum/thread/88999

New join request system if ISF_REQ_JOIN is set - see InSim.txt
New system to reset a car at a location with or without repair
New admin commands /ujoin username and /uai username

It needs some good testing before we go public. More features should be added but let's discuss it on the other thread.

This is an incompatible version intended for programmers.

It needs testing and I am not advertising this for normal users.
I expect to add more incompatible features as discussed.
The download contains LFS.exe / a DCon.exe / InSim.txt


Changes from 0.6K21 to 0.6K22 :

Layout editor :

Temporary start lights (in layout editor) middle light is now amber

Interface :

Message "Connection has no car selected" info adding remote driver

InSim :

New packet IS_SLC reports a connection's currently selected car
Packet TINY_SLC to request an IS_SLC for all connections

Fixes :

Racing car spawned outside pit lane had speed limiter enabled


Changes from 0.6K20 to 0.6K21 :

InSim :

Changed TINY_SEL to more useful TTC_SEL to get a connection's selection
Similarly IS_AXM with PMO_SELECTION can set a connection's selection
Overridden start lights states are now sent to joining players


Changes from 0.6K19 to 0.6K20 :

Layout editor :

You can now set an identifier for a start lights object

InSim :

IS_SSH documentation updated as it is no longer only for bmp files
New packet IS_OCO can be used to override some or all start lights


Changes from 0.6K17 to 0.6K19 :

Layout editor :

New button "place on ground" to restore ground check to objects
You can now adjust width / diameter of multiple marshall objects

InSim :

A car moved by an IS_JRR packet now has the handbrake applied
Added TINY_SEL to request an IS_AXM with layout editor selection
New value PMO_REQUESTED renamed to PMO_TINY_AXM for consistency
New IS_AXM option PMO_SELECTION to set the current editor selection


Changes from 0.6K12 to 0.6K17 :

Layout editor :

Multiple object selection is now available in marshall mode

InSim :

Added TINY_AXM to request IS_AXM packets for the entire layout
InSim checkpoints and circles can be placed in the autocross editor
New packet IS_UCO sends info about InSim checkpoints and circles
Zbyte added to IS_OBH so the layout object can be identified

Misc :

Maximum number of autocross circles increased from 150 to 180


Changes from 0.6K11 to 0.6K12 :

InSim :

INSIM_VERSION increased to 7 to support new incompatible packets
Backward compatibility system - send INSIM_VERSION in the IS_ISI
Added TINY_ALC and SMALL_ALC to get and set allowed cars

Misc :

No longer stored in message history : /press /ctrl /shift /alt

Fixes :

FF was not initialised when "Input when window is inactive" was set


Changes from 0.6K10 to 0.6K11 :

InSim :

IS_MSO / IS_III / IS_ACR message out packets now have variable size
IS_BFN can now be used to delete a range of buttons with one packet
In game spectate then SHIFT+P and type /join - message is now shown

Start grid controls :

Start grid clear button is now available in multiplayer for admins
Add to grid buttons for admins beside names in list of connections
New arrows to move grid positions - removed "swap position" button

Fixes :

Player with car in game leaving pits no longer sends a join request
In 0.6K10 the IS_MTC Msg To Connection with UCID 255 did not work


Changes from 0.6K9 to 0.6K10 :

InSim :

New join request system if ISF_REQ_JOIN is set - see InSim.txt
New system to reset a car at a location with or without repair
New admin commands /ujoin username and /uai username


Download : Test Patch 0.6K9 must already be installed!

http://www.lfs.net/file_lfs.php?name=LFS_INSIM_TEST_6K22.zip

Last edited by Scawen, Thu 11 Feb 2016, 20:01. Reason : K22

It's possible this may have been fixed in the latest test patch: https://www.lfs.net/forum/thread/88821

The startup is a bit cleaner. I can't say for sure if the bug is fixed as this one bug I have never managed to reproduce or test.

Test Patch K9 - Another Rockingham update with fixes for the reported bugs.
https://www.lfs.net/forum/thread/88821

Test Patch K9 - Another Rockingham update with fixes for the reported bugs.
https://www.lfs.net/forum/thread/88821

I left it compatible for now. That means the new sunset lighting is still excluded.

Quote from MandulAA :

Again, I just test it, I don't have any real goals with these two monitors together for LFS.

Thanks for the test. That seems to work correctly.

It is on my list to look at the UI pixels going onto the wrong screen.

Quote from yaper :

As you can see on first screen-shot, LFS is incorrectly (or maybe it is the Windows itself) calculating screen object positions. Interesting thing is that when I press Win button and start menu appears, calculation of screent object positions is becaming correct. >:|

Thanks for the test.

I think that, as it appears in the first image, is a Windows bug. LFS doesn't do any calculations for which part of the image should appear on which monitor. LFS only asks Windows for the position and size of a rectangle for the virtual desktop, then draws a single rectangular image. Windows takes care of positioning.

After you open the start menu, that seems to be as I would expect it. It is possible to use interface scaling options to make sure all buttons are visible.

Quote from MandulAA :

So, I have a secondary screen on the left (1440x900, connected on VGA to the notebook) and the built-in, primary screen on the right (1920x1080). When I choose the borderless window (multiple monitors) option, it looks like this:

I think this is OK. It looks as if your setup is very wide so LFS defaulted to the "Wide" settings and by default that assumes 3 monitors. I expect you can change it in view options to have 1 monitor on one side and 0 monitors on the other side. Then by setting the correct number of pixels for each screen, the interface should be correctly aligned.

But still there would be a scaling issue as Whiskey suggests if the pixels on one monitor are a different size from the pixels on the other monitor.

I've done something about this now.

From the next test patch, if you select TCP for position packets, then your LFS will also send position packets to the host using TCP. So it will be TCP in both directions.

NOTE: if you want to test this in the next test patch (K8) this new feature will only work if you join a K8 server. If you join a K7 or older server, your LFS will still send UDP position packets.

Quote from blackbird04217 :

This probably isn't high priority but it seems to me the KY2 PTH file is outdated.

I see in the track editor, that is how it is. It's just a bug that has always been there. Not usually noticed as our AI drivers can test the physical surface when creating their line. I've made a note to see if I can do some kind of repair job on it.

Test Patch 0.6K7 is now available with an updated Rockingham track!
https://www.lfs.net/forum/thread/88821

Test Patch 0.6K6 is now available! https://www.lfs.net/forum/thread/88821

It includes a new borderless window full screen mode system.