Patch: another pbs for r11923

[wolfc]

Version 0.2 has two opposite terminals crash free, although there is a deadlock sometimes when trains can't find clear path to their destination. Tunnels and bridges remain broken though.

[Expresso]

Well, got a nice bug in relation to pre-signals and depots. I'm aware the use of pre-signals is reduced a lot with this patch, but I deliberately deviated from the expected behavior to expose bugs. I hope this is of some use to you.

[pavel1269]

Atm, if you are making patch which is changing signal states, better and better functionalyti in most cases result in more and more bugs in presignals

[wolfc]

Well, got a nice bug in relation to pre-signals and depots. I'm aware the use of pre-signals is reduced a lot with this patch, but I deliberately deviated from the expected behavior to expose bugs. I hope this is of some use to you.

I didn't disable the old signal behavior yet of waiting for any exit signal to be green. (Another example that YAPF is not finished. Without PBS YAPF is not consulted for a train move. So the train is spit out and left to it's devices (/signals). This leads to trains heading every way but the right way.)

[planetmaker]

Well, got a nice bug in relation to pre-signals and depots. I'm aware the use of pre-signals is reduced a lot with this patch, but I deliberately deviated from the expected behavior to expose bugs. I hope this is of some use to you.

IIRC this works exactly this way with the default behaviour I also experience in the latest beta or nightly.

[CMircea]

Holy...I can't wait to compile this baby and start a game, bugs or no bugs, first real PBS system in OTTD!

[OoffTopic]Is it just me, or OpenTTD has almost caught up with the patch? NewInd. was a big step, PBS is yet another, what's left?[/OffTopic]

[FooBar]

first real PBS system in OTTD!

Was the one in the MiniIN not real enough? It worked for me back then.

[OoffTopic]Is it just me, or OpenTTD has almost caught up with the patch? NewInd. was a big step, PBS is yet another, what's left?[/OffTopic]

Well, there's still a lot of stuff I'd like to use in Open:
- Snow in temperate
- Custom bridgeheads
- Enhanced tunnel entrances
- And of course PBS, but that's heading in the right direction now!

[richk67]

Snow in Temperate will be part of a new All-Climate* World that will be developed this spring. It will allow all climates in one map (yes, even Toyland).

*which climates are included at generation will be selectable. So you can choose Arctic+Tropical for your Kilimanjaro scenario if you want.

[blitzkrieg]

Holy...I can't wait to compile this baby and start a game, bugs or no bugs, first real PBS system in OTTD!

[OoffTopic]Is it just me, or OpenTTD has almost caught up with the patch? NewInd. was a big step, PBS is yet another, what's left?[/OffTopic]

Well the recently released(though still in development) restricted signals are being developed for ottd too!!!

[DaleStan]

[OoffTopic]Is it just me, or OpenTTD has almost caught up with the patch? NewInd. was a big step, PBS is yet another, what's left?[/OffTopic]

You should know better than that!

Well, there's still a lot of stuff I'd like to use in Open:
- Snow in temperate
- Custom bridgeheads
- Enhanced tunnel entrances
- And of course PBS, but that's heading in the right direction now!

Off the top of my head:
- Programmable signals
- Routing restrictions
- More configurability
- Reversible money cheat (so I can repay the "loan")
- Cht:Renew

[CMircea]

first real PBS system in OTTD!

Was the one in the MiniIN not real enough? It worked for me back then.

By 'real' I mean how the signal configuration actually is in RL, and the new proposed PSB system suggests that. Its great to see a patch that actually implements it.

[OoffTopic]Is it just me, or OpenTTD has almost caught up with the patch? NewInd. was a big step, PBS is yet another, what's left?[/OffTopic]

Well, there's still a lot of stuff I'd like to use in Open:
- Snow in temperate

AFAIK, that's already possible, not sure if a patch is needed.

- Custom bridgeheads

The was a branch, but it needs to be resurrected.

- Enhanced tunnel entrances

If that's gonna enter OTTD, I want the track on the tunnel to be the same as normal track, not like in TTDP where it's kinda stone-like.

- And of course PBS, but that's heading in the right direction now!

Yep, I love it's working with YAPF. Will you excuse me now, I've got to compile it

[pshemko]

Anyone has a version that works with latest svn (11814)?

[wolfc]

I got another version, but this one I don't like at all. Depots just won't work, because of a lot of ugly things. YAPF doesn't tackle it at all. It leaves it to the signal code.
The main problem is that there are multiple trains on one track going in different directions. Combine this with the fact that a depot is not (really) a signal and you have got a nice problem.
So at some point a lot of trains will stare at a piece of reserved track that can't be handled anymore.

[wolfc]

0.4 brings more fixes on depot handling. There are still some cases where a train can't reach the depot it has chosen for servicing. This brings the lot to a stand still. I'm not sure how to proceed here. I want trains without reachable destinations to stop (this makes issues with single track and double-way track easier).
I first want to be able to run tracks crash free (trains that is ) with all the basic functionality in place before I start on the tunnels and bridges.
So try to create save games with crashing trains.

[wolfc]

...

Bugrit, you're save game causes some nice train crashes.

The trains won't go over the bridge, because that's currently an invalid path.
And apparently I've miscoded the hack for trains coming out of a depot into an unsignalled route (the terminal). So it's trying to fill platform 2 up with loco's. Somehow this doesn't fit.

[pshemko]

I'm not sure if leaving the depot should work in version 0.4 but clearly - it doesn't

Train A slows down before the signal in front of the station:

img4-1.PNG (48.95 KiB) Viewed 2501 times

and gets a green light:

img4-2.PNG (49.31 KiB) Viewed 2500 times

the result is pretty obvious

save game (no grfs) attached

[pshemko]

The bridges and tunnels still don't work. I've also noticed one more thing - trains seems to prefer a route through a depot instead of a straight one; both train A:

img4-3.PNG (48.94 KiB) Viewed 2480 times

and train B:

img4-4.PNG (59.77 KiB) Viewed 2476 times

save game attached.

[wolfc]

I'm not sure if leaving the depot should work in version 0.4 but clearly - it doesn't

Hehe, like I said in the previous post. Leaving a depot into an unsignaled dead-end doesn't work. Either make the signal before the terminal a double on or put a signal right after the depot, before it starts crossing other tracks.
The second is the more realistic approach (albeit realism is a bit far fetched with the current depot mechanism ), but it completely makes old maps worthless. So I want something that is compatible with the way people are used to signal their tracks. I could leave it as it is now, because your example isn't really the standard (/old) way and the new way should always leave a block-end (/signal) available for trains to stop at.
Let's for the moment say that a depot has all the functions of a main-signal, so I've to fix this problem. (I can't just put a flag on the depot that says green, because the route only applies to a particular train, so I'll do some more hacking.)

Train A slows down before the signal in front of the station:

Yes, quite annoying isn't it? Especially if a train wants to move down a route with a lot of reversed signals.
I think I missed a slowing flag somewhere.

the result is pretty obvious

save game (no grfs) attached

Thanks, keep them coming.

[pshemko]

I don't think trains should be allowed to crash into each other if there is a signal between them, even if the layout is wrong. Trains simply should not be able to get into already occupied section of the track.

[wolfc]

I don't think trains should be allowed to crash into each other if there is a signal between them, even if the layout is wrong. Trains simply should not be able to get into already occupied section of the track.

I partially agree. Trains should not exit the depot when it might cause a crash.
But if a train is heading towards a signal facing the opposite direction (because of a force proceed for example) it should not stop (or even slow down) at the signal. It should ignore the signal altogether (just as in real life).

For a later version I might implement something like 'proceed on sight'.