The International Simutrans Forum

 

Author Topic: Railway Signals [formerly Long Block Signal]  (Read 27402 times)

0 Members and 1 Guest are viewing this topic.

knightly

  • Guest
Railway Signals [formerly Long Block Signal]
« on: December 28, 2008, 06:03:47 AM »
Hi all,

I am new to Simutrans and have read the start-up guide already. However, I still cannot figure out how to use the long blocked signal.  :-\

The guide says it overrides the station's signal so that it will let a train pass only if the track ahead up to the next signal (ignoring any station signal in between) is not occupied by another train. I have tried but failed.  ???

I have attached my save data and would appreciate it very much if anyone can explain to me the proper way to set up so that the middle, common portion of the track (where "Hillden Old Castle Station" lies) can only be entered by 1 train at any given time.

Alternatively, you may load my saved game, make the necessary changes and post it back, if you think it's rather difficult to explain in words.

A thousand thanks in advance!  ;)
« Last Edit: January 15, 2009, 08:14:06 PM by knightly »

Offline KrazyJay

  • *
  • Posts: 172
  • Infrastructure & Transportation Junk
Re: Long Blocked Signal
« Reply #1 on: December 28, 2008, 08:07:55 AM »
please inform us of the version and pakset you use so we can try to help you further.

knightly

  • Guest
Re: Long Blocked Signal
« Reply #2 on: December 28, 2008, 09:02:43 AM »
I am using Simutrans 100.0 with Pak64 Basis. Many thanks!

Offline KrazyJay

  • *
  • Posts: 172
  • Infrastructure & Transportation Junk
Re: Long Blocked Signal
« Reply #3 on: December 28, 2008, 09:18:47 AM »
I tried with the nightly from dec 26th, it works just fine, if I understand your question well. No lock-ups found.

Offline DirrrtyDirk

  • Devotees (Inactive)
  • *
  • Posts: 1253
  • JR 700 Series Shinkansen
  • Languages: EN,DE
Re: Long Blocked Signal
« Reply #4 on: December 28, 2008, 12:24:49 PM »
Yes, seems ok just the way you did it.

knightly

  • Guest
Re: Long Blocked Signal
« Reply #5 on: December 29, 2008, 06:30:25 AM »
Thanks Krazyjay and DirrtyDirk for trying out my saved game.  :)

With respect to the long block signals, there are 2 particular moments in the saved game which I would like to draw your attention to :

1) When 1 train on the middle common track portion is heading Medieval Castle Station , and another train is heading Old Castle Station in the opposite direction but being stopped at the upper-right long block signal

2) When 2 trains in the direction of Branch Station both running on the middle common track portion : 1 train just went past the Old Castle Station and the other will stop at Old Castle Station very soon

In the first case, you may think that the upper-right long block signal has been working as intended to stop the second train from entering the middle common track portion while the first train is there. However, if you load and run the saved game which I attached with this posting, you will see that I have replaced both long block signals with standard signals (other things unchanged), and the second train will still stop at the upper-right standard signal where the original long block signal stands. In other words, it is not due to the unique function of long block signal that prevents the second train from entering the middle common track portion. I suspect that the blocking may be caused by track reservation for the first train as the game reserves the track ahead in its ongoing direction, and the upper-right signal (whether a standard or long block one) detects such track reservation and blocks the second train.  :-\

As for the second case, it is obvious that the long block signal cannot prevent 2 trains from simultaneously entering the middle common track portion as stated in the Starter Guide. Although such case will not arise in a deadlock, but then what is the unique function of long block signal that differentiates it from a standard signal?  ??? In fact, you will find that the behaviour of the trains in my previously attached save data is exactly the same as that in my currently attached save data with regard to the second case as well.

Although this should be irrelevant, I once tried removing the Old Castle Station stops from the schedules of ALL 3 trains in BOTH save data (previous one and current one) and found that the trains' behaviour are also exactly the same. This is obvious as signals from stations which are not on the train schedule are ignored.

So then, what exactly is the unique function of a long block signal? I look foward to receiving your opinions again. Besides, if you believe you have a scenario which can really demonstrate the unique function of a long block signal, please kindly post it up for me to download so that I can try it out myself. Many thanks once again!  :D



Offline DirrrtyDirk

  • Devotees (Inactive)
  • *
  • Posts: 1253
  • JR 700 Series Shinkansen
  • Languages: EN,DE
Re: Long Blocked Signal
« Reply #6 on: December 29, 2008, 11:37:15 AM »
You are right: in this particular setup (I only downloaded the first file) a normal signal is actually sufficient (as it is in 99% of all cases). But longblock signals do have a unique purpose - and it becomes clear if the setup is changed a little bit.

Have a look at this savegame (ST 100.0, pak64):

EDIT: Sorry, a little mistake in the file I'll re-upload it soon.
Use this one here: http://www.simutrans-germany.com/files/upload/LongSignalNEW.sve
« Last Edit: December 29, 2008, 11:44:43 AM by DirrrtyDirk »

knightly

  • Guest
Re: Long Blocked Signal
« Reply #7 on: December 29, 2008, 08:28:31 PM »
@DirrrtyDirk : Thanks a lot for your sample save data.

I have tried it. Although the railway configuration is somewhat different from mine, but with regards to the Long Block Signals and Standard Signals arrangement, I can't see any significant difference (maybe I am stupid) that make yours work differently from mine. It's still two Long Block Signal + Standard Signal pairs with stations in between. If I really need to point out any difference, I can only say that all your Long Block Signals are on the straight rail line, while one of my Long Block Signal is on the branch route. But that should not make any difference, right?

Besides, I found that when a train A is blocked before the lower left Long Block Signal to wait for a train B in the middel track to leave, A starts running again immediately after B turns right at the next junction but before B reaches or passes the next Standard Signal. (Pls see attached image.) However, the Starter Guide says Long Block Signal will prevent simultaneous entry of rail track between it and the next Standard Signal. If that is the case, train A has started to move somehow sooner than expected. (You will see more clearly by pressing "b" to see track reservation).

I must say I am getting confused now. Hope someone can enlighten me on this subject. Thanks once again!  :D

[attachment deleted by admin]

Offline DirrrtyDirk

  • Devotees (Inactive)
  • *
  • Posts: 1253
  • JR 700 Series Shinkansen
  • Languages: EN,DE
Re: Long Blocked Signal
« Reply #8 on: December 30, 2008, 12:45:56 AM »
Don't worry - you're not the first (and probably not the last) getting confused by signalling. ;)

A very useful tool is the "b" key. Now don't bother too much with the tool this represents, but have a look at the view it toggles. You can now see the reservations the trains make, displayed in red. This usually helps in understanding how things work.

The important difference is that in my example there are two stations on a continous piece of single track instead of just one. Maybe the effect becomes even clearer when you use 3 or 4 ... If you don't use long block signals in this setup, a deadlock can happen. Maybe it won't really happen in the simple example I posted, since the trains are always lined up as on a pearl necklace, but it is possible.

Let me try to explain:
Let's assume you use just normal signals on the screenshot you posted. The trains on the left has reserved the way until "Hilden branch station" and is already on its way there. Now just imagine the train on the right would ot just leaving the single track, but trying to enter it! With normal signals it could (and would) reserve the path to "Hillden Old Castle station" and commence going there. Both trains would reach their station - and both would not be able to leave because they're in each others way. With longblock signals, this cannot happen as the second train would be stopped at the long block signal until the first train has not only reached both "Hilden branch station" (and "Hillden Old Castle station" afterwards) but also has left the single track altogether.

I hope this helps a little better.

knightly

  • Guest
Re: Long Blocked Signal
« Reply #9 on: January 10, 2009, 02:09:55 PM »
<obsolete>

Quote

@ DirrrtyDirk

Really thanks a lot for your reply. I have found some time to test out the long block signal, and have finally discovered its behaviour.

Let's assume the following loop route set-up :



++++++++++++++++++++++++++++
+                                                                        +
+                                                                        +
+                                                                        +
+++
L+++A++B+++C++++++++++S+++



+ represents railway, L represents Long Block Signal, S represents Standard Signal, and A, B, C represents 3 railway stations.

Given this set-up, plus 2 trains now moving one after the other closely in anti-clockwise direction towards L (assumming both trains have the same schedule), I found out after repeated testing that :

1> If none of the stations is included in the train schedule, the 2nd train can't proceed beyond L until after the 1st train has reached and passed S. This is just the same as what the starter guide says -- the whole portion of track between L and S is blocked from entering if another train is moving inside it. Similar behaviour is exhibited if there are no intervening stations between the signals.

2> If all 3 stations are included in the train schedule, the 2nd train will start to move past L once the 1st train has completely left the station platform of C.

3> If only A and B are on train schedule, the 2nd train will start to move past L once the 1st train has completely left the station platform of B

4> If only A is on train schedule, the 2nd train will start to move past L once the 1st train has completely left the station platform of A

5> If only B is on train schedule, the 2nd train will start to move past L once the 1st train has completely left the station platform of B

6> If only C is on train schedule, the 2nd train will start to move past L once the 1st train has completely left the station platform of C

7> If only A and C are on train schedule, the 2nd train will start to move past L once the 1st train has completely left the station platform of C

8> If only B and C are on train schedule, the 2nd train will start to move past L once the 1st train has completely left the station platform of C


So, to summarise, the long block signal either

(1) blocks other trains from proceeding if another train is running on the portion of track from long block signal up to and including the next standard signal, in case where all intervening stations are not on the train schedule or where there is no intervening station

(2) blocks other trains from proceeding if another train is running on the portion of track from long block signal up to and including the last tile of the station platform of the last intervening station which is included on train schedule, in case where there are intervening station(s) and at least one of them is included in train schedule.

Thus, if anyone believes in the straightforward concept that a train is blocked from proceeding as long as another train is running inside the track portion between a long block signal and the ensuing standard signal, he or she will very likely be surprised to find out that the blocked train starts moving past the long block signal earlier than expected.

Besides, if the 1st train has all 3 stations on its schedule, but the 2nd train only has B on its schedule, the 2nd train can start moving past L once the 1st train has completely left the station platform of B. That means, the train schedule which I have mentioned frequently above is the train schedule of the blocked train (2nd train), not the schedule of the 1st train running inside the signal-bound track portion. As the train-blocking behaviour of a long block signal depends on the schedule of the blocked train, the same set-up of signals will behave differently as trains with different schedules pass by.

I don't know whether the starter guide is incomplete in describing the behaviour of long block signal, or the long block signal is not behaving as it should in the way intended by the developers. Anyway, due care must be exercised to attain the desired train blocking, track reservation or train pacing.


</obsolete>




Another small observation :

If you place a long block signal after a standard signal and have a train X running between the 2 signals, when another train Y arrives at the standard signal, the standard signal cannot block this second train from proceeding as it usually does.

+++++++Y+S++++++X+++++++L++++++++++++>>



EDIT 1 : Below and green words above

Having said so much about long block signal, I think I must also digress a little bit to talk about the behaviour of a choose signal for railway, as there is something similar between it and a long block signal.

It is important to understand the mechanics of a choose signal :

1) In essence, it actually does not work with the next upcoming station ahead on the track. Rather, it works with the next stop on the schedule of the passing train (waypoint also counts as a stop). This aspect is similar to the long block signal.

2) It not only lets the train choose from the available platforms in the next stop, the choice of platform will also be made immediately and the portion of track from the choose signal up to and including the chosen platform will be reserved. If such track reservation cannot be made successfully due to the presence of another train in that track section, the train will wait at the choose signal until that portion of track is clear.

3) Any intervening standard signal between it and the next stop on the train schedule cannot stop or cut short the portion of track reserved.

Usually, when we use a choose signal, we will place it before a station with multiple platforms so that incoming trains can choose from among them. If all trains passing through the choose signal also have that multi-platform station on their schedule, there should not be any problem.

However, as we continue to expand our railway networks, very often we will reuse existing railways. New trains are introduced but they may not be scheduled to stop at all stations along the way. If they drive past a choose signal but do not have the upcoming station on schedule, they may reserve a really long portion of track from the choose signal all the way up to and including the next stop's chosen platform.

To illustrate, let's assume the following set-up :

++++++++C+X+S++++++++++S+++++++Y++++++++S++++++++++++++++Z++++++++++>>


A train is driving into this track section from left to right. The train only has Z on its schedule. X is a multi-platform station and thus a choose signal C is placed in front of it.

When the train drives past C, it will look ahead and choose from any available platform from Z. If it has found and chosen an available platform of Z, the train will try to reserve the portion of track from the choose signal up to and including the chosen platform of Z.

If such track reservation is successful, it will proceed forward. If such reservation is unsuccessful due to presence of another train or simply because there is no unoccupied platform at Z to select at the moment, the train will stop and wait at the choose signal. Note that any intervening standard signal has no effect on such potentially aggressive reservation of lengthy track section. Needless to say, reserving long portion of track will increase the likeliness of deadlocks as well as the waiting time of a train.

So, once again, the behaviour of a choose signal depends on the individual schedules of the passing trains. One has to be careful in planning where to set up a choose signal, especially if the track portion is shared by more than one railway line.

Edit 2 : Obsolete section above regarding Long Block Signal
« Last Edit: April 08, 2009, 04:41:03 PM by Knightly »

Offline VS

  • Senior Plumber (Devotee)
  • Devotees (Inactive)
  • *
  • Posts: 4856
  • Vladimír Slávik
    • VS's Simutrans site
  • Languages: CS,EN
Re: Long Blocked Signal
« Reply #10 on: January 13, 2009, 10:06:57 PM »
Wow, you managed to accumulate more text than anyone wants to read ;) But it seems highly useful! Thanks!


Offline emaxectranspoorte

  • *
  • Posts: 241
  • GOD BLESS AUSTRALIA
Re: Long Blocked Signal
« Reply #11 on: January 13, 2009, 10:40:46 PM »
When I had version 99.17 pak64 a long time ago, I had similar problems with long block signal. Then, I had just thought, that a long block signal stops a second following train coming into a rail block, before the first, finishes its route through the track block (with many stations). I understand now this is not the case. :) Maybe someone should invent a Advanced Long Block Signal. ;) :D Or something similar to that, IMHO. :) Is there any other solution? Am I missing and/or forgotten something?

As for the Choosesignal, as small suggestion from my experince, IMHO: Place a choosesignal before Z and a normal signal after Z, that way you have two signals on which to choose between platforms: at X and Z.


knightly

  • Guest
Re: Long Blocked Signal
« Reply #12 on: January 13, 2009, 10:44:27 PM »
@ VS

Thanks a lot for both your discouragement (regarding post length)  :P as well as encouragement (regarding content)  :D

Just hope that my lengthy post above is useful to other Simutrans players.


@ emaxectranspoorte

Thanks for your comments and suggestions. With regard to my choose signal example, station Z is just a single platform station. So, the train actually has no need to do a platform search; it's just because the train happens to drive past a choose signal that platform searching and track reservation is triggered. Putting another choose signal before Z cannot prevent such aggressive track reservation behaviour.

IMHO, I think it may be better not to depend the behaviour of long block signal and choose signal on the passing train's schedule -- it will be better if their behaviour is dependent only on track layout and respective signal placement. As you can see now, the 2 signals' behaviour is not consistent from time to time, which is somewhat undesirable and not easily understandable to new players (including me  :P ).
« Last Edit: January 13, 2009, 11:19:31 PM by knightly »

Offline VS

  • Senior Plumber (Devotee)
  • Devotees (Inactive)
  • *
  • Posts: 4856
  • Vladimír Slávik
    • VS's Simutrans site
  • Languages: CS,EN
Re: Long Blocked Signal
« Reply #13 on: January 13, 2009, 10:52:09 PM »
Well, when I will have to use long signal, I know where to go for information - or where should I send those who ask. That is a big plus. It does not matter that there is too much text. If the behaviour depends on too much, it inevitably ends so. You can safely drop the discouragement and keep the rest :)

Offline emaxectranspoorte

  • *
  • Posts: 241
  • GOD BLESS AUSTRALIA
Re: Long Blocked Signal
« Reply #14 on: January 13, 2009, 11:51:29 PM »
IMHO, I think it may be better not to depend the behaviour of long block signal and choose signal on the passing train's schedule -- it will be better if their behaviour is dependent only on track layout and respective signal placement. As you can see now, the 2 signals' behaviour is not consistent from time to time, which is somewhat undesirable and not easily understandable to new players (including me  :P ).

I think, IMHO, part of what you are talking about in bold behaviour used to be in older versions of Simutrans like 88.10.5 stable, for example. If I remember right, anyway. I think the current behaviour is because it's sometimes too cumbersome to place signals everywhere as it might in a very complex network become, at times, a problem. So that's why a station can also act as a normal signal. :) :-\ It's somewhat realistic, as well. :)
« Last Edit: January 14, 2009, 12:01:50 AM by emaxectranspoorte »

knightly

  • Guest
Re: Long Block Signal
« Reply #15 on: January 15, 2009, 03:52:01 AM »
Please pardon me for digressing again, this time on pre-signal and standard signal, for I have related observations to share.

First of all, the wiki says that "stations act as 'de-facto' signal points". This is not true in general, as only stations registered on a passing train's schedule behave like a signal point. This concept has already been illustrated by Frank in this thread. The related pitfall can be made clear with an example. Assuming the following set-up with trains running from left to right :

++++++++S++++++++X++++++++++++Y++++++++++++++++Z++++++++S++++++++>>

Originally you have a line with a few trains passing through this track section, stopping at all stations X, Y and Z. Later, you have a new line that reuses the same track portion but its trains don't stop at any of the stations X, Y and Z. The result is, the new line's trains will wait at the 1st S for a relatively long time if the track between the 1st S and the 2nd S is not clear of any train; and once the track is clear and they can start moving, they will reserve a lengthy track portion from where they are up to the 2nd S.


However, that is not the complete picture either. I would like to point out further that, actually all stops on a passing train's schedule will behave like a signal point -- whether the stop is a station or just a way point.

As an illustration for pre-signal, let's assume the following set-up :

++++++++P++++++++W+++++++S++++++++S++++++++>>

2 trains are running consecutively from left to right, one closely following the other. W here means a way point on the 2nd train's schedule. P is a pre-signal and S is a standard signal. When the 1st train drives past P, the 2nd train stops and waits at P; but since W behaves like a signal point, the 2nd train will start to move once the 1st train has driven past the first S.

Similarly, as an illustration for standard signal, let's assume the following set-up :

++++++++S++++++++W+++++++S++++++++S++++++++>>

Again, 2 trains are running consecutively from left to right, one closely following the other. This time, the 2nd train which has been blocked at the 1st S will start moving once the 1st train has driven past W.

IMHO, it is really quite ridiculous that way points also behave like a signal point and affects the behaviour of real signals. Besides, if stations are intended to be de facto signal points, they should be consistently so under all cirumstances, except of course in the case of long block signal.

As you can see, the behaviour of all 4 signals depends on the passing train's schedule, and all of them will potentially exhibit irregular or inconsistent behaviour as trains with different schedules pass through them. The down side is, one has to be clear how the schedule of each line or train interacts with each signal along the track to decide if things work out as intended; otherwise, pitfalls can easily be overlooked. If you are not careful enough, you can accidentally let a train wait for an unnecessarily long time or reserve an unnecessarily long portion of track, potentially increasing the likelihood of deadlocks.

Again, I would say that it will be better if all signals' behaviour is dependent only on track layout and respective signal placement. Anyway, all we can do now is to watch out for the pitfalls and be careful in using the signals.

@emaxectranspoorte : If a railway system is complex, it is quite likely that there will be a few trains with different schedules running on the same track section. In that case, the inconsistent behaviour of the signals will more likely than not make your system difficult to manage.



Another small observation :

If you place a choose signal or pre-signal after a standard signal and have a train X running between the 2 signals, when another train Y arrives at the standard signal, the standard signal cannot block this second train from proceeding as it usually does. This is the same as with long block signal mentioned before.

+++++++Y+S++++++X+++++++C/P++++++++++++>>



« Last Edit: January 15, 2009, 05:43:18 AM by knightly »

Offline DirrrtyDirk

  • Devotees (Inactive)
  • *
  • Posts: 1253
  • JR 700 Series Shinkansen
  • Languages: EN,DE
Re: Long Block Signal
« Reply #16 on: January 15, 2009, 08:11:35 AM »
Please pardon me for digressing again

Actually the only problem with that is, that nobody expects a briefing about the workings of presignals in this thread, so you should probably have opened a new one with a matching title...

Offline jamespetts

  • Simutrans-Extended project coordinator
  • Administrator
  • *
  • Posts: 20684
  • Cake baker
    • Bridgewater-Brunel
  • Languages: EN
Re: Long Block Signal
« Reply #17 on: January 15, 2009, 07:29:34 PM »
This is extremely interesting, and I quite agree with Knightly about the problems, both in gameplay and realism terms, caused by the inconsistent behaviour of signals. The real question is: how easy is this to fix in the code?

knightly

  • Guest
Re: Railway Signals [formerly Long Block Signal]
« Reply #18 on: January 15, 2009, 08:13:28 PM »
@ DirrrtyDirk : Sorry for digressing so much, but I really think that these are related observations which are better to be placed together. Besides, while it is more appropriate for my previous post to reside in the Help section, starting a new thread in it without initiating a question (since I am not asking questions with that post) would seem rather odd. So, I will now change this thread's title to "Railway Signals [formerly Long Block Signal]" to better reflect the coverage of this thread as wel as to remind people of the old thread title. Hope that resolves the problem.
« Last Edit: January 15, 2009, 08:36:55 PM by knightly »

Offline kierongreen

  • Dev Team, Coder/patcher
  • Devotee
  • *
  • Posts: 2346
Re: Railway Signals [formerly Long Block Signal]
« Reply #19 on: January 18, 2009, 06:56:41 PM »
The long block signals were a bit of a hack to cope with very specific situations. These were generally where you had stations on single track line sections in for example a tram system or in a low traffic rural railway. They weren't designed for spacing purposes, rather for pathing ones. Changing the behaviour would be quite difficult...

Offline fabio85

  • *
  • Posts: 32
Re: Railway Signals [formerly Long Block Signal]
« Reply #20 on: March 01, 2009, 04:51:01 PM »
Hi all!
I have read all the choose signal's post and I have noticed that choose signal looks at the schedule and not at the single station. But the quest is: how can I avoid train's deadlocks if I have more that one line in a single couple of rails?
In other words: I have a line with a couple of tracks and many stations that take up all the two tracks.
                         C                                      C                                                 C
________________________________________________________________ _________ >
__X_________X______________X_________X_____________________X____ _____X__ <
 C     AAAAAA                     C        BBBBB                                  C      DDDDD
Before and after each station I have built a track crossing. The two tracks are single way and, before entering a station, trains meet a choose signal because they can choose one of the two platforms. I use this line for passengers and trains must in each station (A B D): I haven't any problem...
When I have added a commercial line that must stop only in A and D problems started: my commercial train remains locked at the B station's choose signal because there are passengers train on its way (far...)!!!
I have found a solution to this: to add station B to the commercial train schedule. I add an useless stop to the schedule, but I avoid deadlocks...
There is another way, more "elegant", to avoid deadlocks? I tried to select the track between choose signal and platform to avoid the train to stop, but it cannot choose the platform: if there is another train at the platform i have another deadlock!!!
Thanks!

Offline Lodovico

  • *
  • Posts: 80
Re: Railway Signals [formerly Long Block Signal]
« Reply #21 on: March 02, 2009, 08:11:47 AM »
To my experiences a choose signal works ONLY for convoys that DO stop at the station next to signal.
It would be VERY usefull, if any ST guru write some patch for choose signal before station where convoys do not stop.

knightly

  • Guest
Re: Railway Signals [formerly Long Block Signal]
« Reply #22 on: March 02, 2009, 08:44:19 AM »
To my experiences a choose signal works ONLY for convoys that DO stop at the station next to signal.

Actually a choose signal still works if the train doesn't stop at the next upcoming station.  ;) It's only that it doesn't work with the next upcoming station as one would normally assume, but instead works with the next stop in the train's schedule, as I have explained in one of my posts above.

It would be VERY usefull, if any ST guru write some patch for choose signal before station where convoys do not stop.

If any Simutrans guru is kind enough, I would be most grateful  :D if the whole signalling code be revised so that
(1) all signals' behaviour is dependent only on track layout and respective signal placement.
(2) all stations act as de facto signal points, whether they are on or off train schedules
In that way, new players (and even experienced players) would find the signalling system more intuitive and easy to manage, instead of being confounded by inconsistent, perplexing signal behaviour which increases the likelihood of deadlocks as well as train waiting time at signals.
« Last Edit: March 02, 2009, 08:50:36 AM by Knightly »

Offline fabio85

  • *
  • Posts: 32
Re: Railway Signals [formerly Long Block Signal]
« Reply #23 on: March 02, 2009, 04:49:50 PM »
So there isn't a more "elegant way"... It's better to add all stations in a train schedule even if there isn't a "commercial" necessity (the only necessity is to avoid deadlocks)...

Offline The Hood

  • Devotees (Inactive)
  • *
  • Posts: 2889
  • pak128.Britain developer
Re: Railway Signals [formerly Long Block Signal]
« Reply #24 on: March 02, 2009, 05:11:13 PM »
Isn't there an "end of choose signal" which limits how far the choose signal will work for?  IIRC this should do what you want, but then again I've forgotton how to play the game now!

Offline whoami

  • Devotees (Inactive)
  • *
  • Posts: 693
Re: Railway Signals [formerly Long Block Signal]
« Reply #25 on: March 03, 2009, 02:44:34 PM »
Most of the issues with choose signals have been discussed earlier, e.g. in http://archive.forum.simutrans.com/topic/05930.0/index.html - Change request: Limiting the effect of choose signals (previous forum's archive).

Offline fabio85

  • *
  • Posts: 32
Re: Railway Signals [formerly Long Block Signal]
« Reply #26 on: March 07, 2009, 10:18:19 PM »
Quote
Most of the issues with choose signals have been discussed earlier, e.g. in http://archive.forum.simutrans.com/topic/05930.0/index.html - Change request: Limiting the effect of choose signals (previous forum's archive).

Reading this thread I have understood that "End of choose signal" can be the solution to my problem. But I have made an experiment to test it with bad results.
I have placed rail signals in that order:
              C                EOC                           C
<________________________________________________________________ _________________<
>____X__________________X____________X___________________________ _______X_________>
   SS                             C           SS               EOC                                        C          SS

C=choose signal
EOC= end of choose signal
S= station with two platforms.
Rails between stations are single way, indicated with < >. I have tried with many trains that stop in all the three stations and one only in first and third station. This last train remains locked at the choose signals if there are trains on his way to scheduled station. With EOC this would be avoided, but it don't happen... why? EOC it's the right choice to avoid trains enter a certain platform: if they are placed between choose signal and the platform, trains will not enter this platform...

Offline whoami

  • Devotees (Inactive)
  • *
  • Posts: 693
Re: Railway Signals [formerly Long Block Signal]
« Reply #27 on: March 07, 2009, 10:34:00 PM »
I have placed rail signals in that order:
So you placed EOCs after the stations? This is useful to avoid that a train at a CS (choose signal) would try to find a route even on that track (it can happen!), but it doesn't do what you want it to do.

Quote
I have tried with many trains that stop in all the three stations and one only in first and third station. This last train remains locked at the choose signals if there are trains on his way to scheduled station.
To avoid this, give that last train a waypoint after each station that it doesn't stop at, roughly where you already put the EOCs. If the next scheduled stop is a waypoint, the CS does not look for alternative paths.

Quote
With EOC this would be avoided, but it don't happen... why?
The EOC only prevents the choice of an alternative path using its tile (that you can use it to exclude certain station tracks, is the main intention of this sign, not its core function). But if a path across it has to be used, the CS forces a reservation to the next station in the schedule.

Offline fabio85

  • *
  • Posts: 32
Re: Railway Signals [formerly Long Block Signal]
« Reply #28 on: March 22, 2009, 07:42:25 PM »
I tried adding a waypoint after the station in the middle at the train that stops only in first and third station: choose signal becomes green only if there isn't another train in his way to the waypoint. It's better than before, but remains the problem that choose signal looks only to the "straight" way to waypoint!
              C                EOC                           C
<______________________________1_________ _______________________________________ ___<
>____X__________________X______2______X__ ________________________________X______ ___>
   SS                             C           SS               EOC      WP                               C          SS
If the train is going from the first station to the third and there a train stopped in platform 2 of second station, it will wait for this last will exit the station to station one. It's not important that platform 1 is free, it will wait for platform 2!
WP is the waypoint's position.
I think that the principal problem is that choose signal controls railways looking at the schedule instead of the effective presence of convoys.

hymie

  • Guest
Re: Railway Signals [formerly Long Block Signal]
« Reply #29 on: April 07, 2009, 12:20:00 PM »
Greetings.

First post, so please be gentle.

I've read through this thread and the wiki page http://en.wiki.simutrans.com/index.php/Traffic_congestion_control#Signals, and the information is very interesting, but I haven't found an answer to a problem I've run into.  My search-fu is not strong, so perhaps I missed it.  If so, I'd appreciate a pointer.

I'm running Simutrans 1.02 with Pak64 on Linux.

I have a tram line set up thusly:

----<------A------B------C------<----
---->/                         \>----
with the arrows being one-way-signals, and A, B, and C being "street" stops.  I have trams running in both directions, so picture a tram approaching from the left and a tram approaching from the right at the same time.

I read the thread, went home, and tried regular one-way signals, long-block-one-way-signals and one-way-pre-signals.  I used the "b" key, which I don't know what "b" stands for, but it shows the rail reservations.  In all three cases, the tram from the left only reserved track up to the "A" station, and the tram from the right only reserved track up to the "C" station, resulting in a deadlock.

This does not match what Knightly said in Reply #9
Quote
blocks other trains from proceeding if another train is running on the portion of track from long block signal up to and including the last tile of the station platform of the last intervening station which is included on train schedule, in case where there are intervening station(s) and at least one of them is included in train schedule.
(it only reserved up to the last tile of the first intervening station)

and does not match the wiki
Quote
Use a long-signal, and a train will only proceed past it if the entire track section up to the next signal is clear.
A train will wait at a pre-signal unless the block immediately adjacent to the presignal is empty and the signal at the far end of that block is green.
(italics copied from the wiki) unless we assume that a station is also a signal.

Is there a solution to the problem I'm facing?

Thanks.

--hymie!

Offline whoami

  • Devotees (Inactive)
  • *
  • Posts: 693
Re: Railway Signals [formerly Long Block Signal]
« Reply #30 on: April 07, 2009, 12:43:16 PM »
----<------A------B------C------<----
---->/                         \>----
with the arrows being one-way-signals, and A, B, and C being "street" stops.  I have trams running in both directions, so picture a tram approaching from the left and a tram approaching from the right at the same time.
Luckily, this is a trivial setup, perfectly suitable for long-block-signals (=:LBS).

Quote
I used the "b" key, which I don't know what "b" stands for, but it shows the rail reservations.
'b' originally stood for "block reservation repair tool". Nowadays, it selects the PBS (path based signalling) reservation repair tool.

Quote
In all three cases, the tram from the left only reserved track up to the "A" station, and the tram from the right only reserved track up to the "C" station, resulting in a deadlock.
Placing an LBS at each entry should be sufficient from the signalling side, but only with matching train schedules and a suitable track layout in between. Regarding schedules: the trains should not turn around in the section between the LBSs (exceptions only for certain cases). Track layout: same, it should not make trains reverse.

Quote
and does not match the wiki(italics copied from the wiki) unless we assume that a station is also a signal.
Reservations are still done as usual, so you are right, stations appear as if they had built-in signals. However, the LBS checks only for existing reservations on the path, but it does not reserve the whole path. This is the reason why trains should not reverse.

Offline jamespetts

  • Simutrans-Extended project coordinator
  • Administrator
  • *
  • Posts: 20684
  • Cake baker
    • Bridgewater-Brunel
  • Languages: EN
Re: Railway Signals [formerly Long Block Signal]
« Reply #31 on: April 07, 2009, 02:03:30 PM »
A station is treated as a signal, but, confusingly, only for trains scheduled to stop there.

knightly

  • Guest
Re: Railway Signals [formerly Long Block Signal]
« Reply #32 on: April 07, 2009, 02:45:58 PM »
Hi Hymie  :)

Maybe you can post your save data here and indicate which part of the map your are referring to, so that we can take a look?

hymie

  • Guest
Re: Railway Signals [formerly Long Block Signal]
« Reply #33 on: April 08, 2009, 11:53:14 AM »
Reservations are still done as usual, so you are right, stations appear as if they had built-in signals. However, the LBS checks only for existing reservations on the path, but it does not reserve the whole path. This is the reason why trains should not reverse.

I tried it out, and I understand now.  Long-signals do in fact work correctly, they just don't look like they work correctly when you use the "b" key.

Thanks for the help.

knightly

  • Guest
Re: Railway Signals [formerly Long Block Signal]
« Reply #34 on: April 08, 2009, 04:21:36 PM »
IIRC when I wrote the observations about the signals months ago in my previous posts above, I was using Simutrans v.100. Back then Long Block Signals really behaved in the way I described.

Now I have just done a quick test on Long Block Signals with Simutrans v.102 and obviously they no longer exhibit any inconsistent behaviour based on the passing train's schedule. This is good news indeed :). So, what I said about Long Block Signal is now obsolete and can be ignored. Please take note.