[Rseding91] 0.14.1 - Tanks handle slowly

[mikeydunn69]

This should be easy to reproduce.
Enter a tank and use the steering keys and I have found it is extremely slow to turn, making the tank unusable.

[Klonan]

Thanks for the report,

Can you check if the turn speed is the same as in 0.13?

[Klonan]

Just checked, its exactly the same in 0.12 and 0.14...

[_alphaBeta_]

I'm not checking values, but the turning speed of the tank in 0.14 is quite slower than 0.13. It's a profound difference that I noticed immediately. Was the increase in maneuverability a bug in 0.13? I've checked the changelog, but I'm not understanding why you were concerned with comparing the 0.14 values to 0.12 and not 0.13, unless 0.13 was bugged.

[Loewchen]

I'm not checking values, but the turning speed of the tank in 0.14 is quite slower than 0.13. It's a profound difference that I noticed immediately. Was the increase in maneuverability a bug in 0.13? I've checked the changelog, but I'm not understanding why you were concerned with comparing the 0.14 values to 0.12 and not 0.13, unless 0.13 was bugged.

13.20 on the left, 14.5 on the right:

FacTankStats.JPG (89.76 KiB) Viewed 6843 times

Could theoretically be, that it was changed and then changed back within .13 but that still would not make it a bug. You could mod it to your liking btw.

[_alphaBeta_]

I'm certainly believe the numbers are the same, but I still think there's another problem here.

Just now, I carried out the following experiment for both 0.13.20 and 0.14.5

1. Create a new sandbox game.
2. Click the left button on all the prompts essentially giving you starting items, all research, and being in cheat mode.
3. Construct a tank.
4. Place three assembler machines in a row nearby.
5. Enter the tank.
6. While maintaining full throttle, attempt to circle the three assembler machines by holding either left or right continuously to steer in a circle.

They're not even close. On 0.13.20, you can hold a tight turn and basically circle those three assembler machines very closely. The turning radius on 0.14.5 is more than double this. I selected assembler machines randomly just to establish scale between the two experiments, but it could be anything.

The values may be the same, but something else is wrong. As pure speculation, it "feels" like not all the turn user input is being registered every frame. Even with no forward motion (turning on a dime), the circling motion is jerky on 0.14.5 and smooth on 0.13.20.

Can anyone else reproduce this?

[Adil]

Well, the prototype is unchanged but numbers are definitely handled differently internally now.

I've conducted the following test: I've hopped into the tank and used the following command

Code: Select all

m=0; l=p.vehicle.orientation script.on_event(defines.events.on_tick,function()  local q=p.vehicle.orientation-l; l=(q+l); if q>m then m=q print(m) end if m>0.99 then m=0 l=0 end end)

I believe the above function prints the maximum change of the angle between two consecutive ticks.
Then I've pressed 'd' to start rotating the tank.
The numbers printed in 0.12 version were 0.35000something, pretty much what prototype says.
The version 0.14, however, outputs twice as small turning speeds of 0.00175something.
The car in 0.14 demonstrates turning speeds of 0.009 instead of 0.015 in prototype. (It depends on the speed of the vehicle however.)

If anything, I believe that these modifiers to prototype values should be explicit.

[_alphaBeta_]

Well, the prototype is unchanged but numbers are definitely handled differently internally now.

Thanks for the much more quantitative confirmation than my experiment. Though in this case the difference/problem is noticeable enough to be identified from casual observation alone.

Can we get this moved out of "Not a Bug" subforum? This affects driving either vehicle quite substantially, so I think it warrants dev attention.

[Adil]

It's now 0.14.07 and earlier observations are reproduced.

[_alphaBeta_]

Issue persists as of 0.14.12. I'm trying to get this thread moved back into the Bug reports board instead of opening a new report there. All the info is here.

[Loewchen]

I move this back for revaluation...

[_alphaBeta_]

I move this back for revaluation...

Thanks Loewchen!

My concern was that the developers simply noted the lack of change in the prototype and moved on. I just want to make sure they're aware that the prototype numbers are handled quite differently. Looking at how a tank spins on a dime (with no forward motion) between a 0.13 and 0.14 build really has me thinking this was unintentional (there's considerable hesitation as the tank passes through certain orientations which affects the turning radius while moving also), or it at least should have been explicitly mentioned in the changelog that the maneuverability was being decreased. Attacking bases in tight spots is really not a fun experience at the moment, but if the developers did indeed intentionally make the change, this discussion belongs elsewhere at least.

[Xeteth]

This spiked my interest so I decided to take a look myself.

I ran two turning-circle tests on version 0.12 and 0.14 (don't have a copy of 0.13) and found that the maneuverability is indeed wonky in 0.14

Here is a comparison side by side; note that the version 0.12 tank is able to turn in a fairly tight manner whilst the 0.14 tank is slow to turn. Unfortunately you cannot see due to the shit FPS of the .gif (someone get Klonan to give me his ShareX settings ) but it appears as though the 0.14 is unable to turn properly when facing certain directions.

Version 0.12 tank on left, 0.14 on right

Download the original gifs here if you want a closer look:
https://www.dropbox.com/s/33g9g3uiuzfnm ... s.rar?dl=0

[Rseding91]

Fixed for 0.14.13.