News:

Want to praise Simutrans?
Your feedback is important for us ;D.

Discussion of new signalling system

Started by Octavius, October 26, 2015, 09:37:24 PM

Previous topic - Next topic

0 Members and 3 Guests are viewing this topic.

Isaac Eiland-Hall

Quote from: Vladki on June 20, 2016, 08:55:37 PM
Compiled sucessfully. Thank you very much.
Edit: one more install please libpng-dev is needed to compile makeobj. Thanks.

Got it! :)

Ves

Huh, this has taken days to summary up and write!
I apologize on beforehand if I tell the same things more times than one, but as said, this was written over a long range of days with many interruptions.

However, here it is:


Description of the problem:
If you take a map of an existing station, you might not be able to put the signals according to the plans. This, obviously, does not take into account the earlier mentioned dvarfsignals or other shunting signals, but only the signals used for 'real' traffic. Also the T-signals are a subject of its own and also outside the scope of this report.

Description of the relevant real-world signalling practice:
You can make two different ways of placing the signals dependant on the era. I will just here refer to them as "Modern" and "Old". Also there is the "T-semaphore" way, but that is another discussion.

Modern way of placing signals:
Normally a signal is put at the platform end (the platform signal), and then another signal after the last point at the station (the exit signal). The former cannot be CLEAR if the later are locked at DANGER.
This is to ensure that the train does not leave the platform while the exit signal is at DANGER, blocking the passage for other trains. The train may not leave the platform if it has not been given clearance by the dispatcher, which in turn has made sure that the rute is ready to handle the train.
When Dispatcher is ready, (s)he will attempt to set the platform signal to CLEAR, and if the exit signal approves, both signals will go to CLEAR. If the exit signal for some reason cannot be set to CLEAR, the platform signal can also not be set to CLEAR.

Here is a bunch of diagrams (in swedish, unfortunately) that illustrates where the signals are positioned. However, you dont need to read swedish to read the diagrams.
https://web.archive.org/web/20070417163015/http://home.swipnet.se/~w-54462/signalteori/hsifsi1.htm
Translate the number of lights on top of each other on the signals like this:
1 light  = repeater of the next signal. Im not 100% sure, but if light is lit, train may not pass behind the signal
2 lights = 2 state signal
3 lights = 2 state signal (with posibility to show restricted speed)
4 lights = 3 state/choose signal
5 lights = 3 state/choose signal (with posibility to permissive travel)

The signal that is a big circle with multiple lights inside are presignals (two different variations, doesnt matter for now)

Looking at a real chart for a station:
http://www.ekeving.se/r/siEJ/RB/1452-001_1960-1983.jpg
Åkersberga
You will find similar symbols as in the diagrams previously linked. If following the track away from the station to the left, you first encounter a road crossing and after that you will find a 4 lamp signal facing one way and a two lamp signal facing the other. Those two signals are the entry signal and the exit signal. Following the tracks to the right of the station and you will find a similar setup.


The "Old" way of putting signals:
Looking at many earlier drawings from the semaphore time, there are not so many permanent signals. Often you find only the entry signal and no signals what so ever at the platforms and only in some occations an exit signal. Obviously there would be a flagman ready at the platform to send the train away and in that regards, (s)he would substitude as a "signal" in simutrans concept.
The flagmans job would be to keep the train at the platform until it was clear to go, following the same principe with the dispatcher. Whenever the train has gone outside the station area (usually at the entry signal) the tracks are free to use for other trains.


An example of a bigger station Ängelholm in the old days:
http://www.ekeving.se/r/bgd/TKP/T-1943_angelholm.jpg
This is quite messy with a lot of lines, however, it consist of two single tracked lines going in from left and upper left and a three tracks going out to the right. Now I think that two of the three tracks actually is not a part of the main line, but rather dead end tracks.
Note that semaphores are used as entry signals and color light signal as exit signals. Although on the track going out to the lower left, I cant find any signals...
Also note that the two exit signals protecting the line to the right probably is a combination of exit signal and platform signal, as the signals would be visible from the platform. However, Im not sure that there would be no flagman anyway.

Semaphores looks like a very small line with a thick bulk like a semaphore wing to the thin line. If the semaphore has more wings, the printed semaphore would have them in the upright position. You can find one with three wings at the very right of the drawing.



Illustration of the problem:
I will go through the working methods, one by one, and illustrate what challenges, problems and possibilities there are to achieve this:

To ENTER a station of any size, old or modern:
Usually no problem with any working method, as there should always be placed a choose signal at the entry of the station. To my knowledge, all choose signals possible to create in Experimental using different working methods, all will adapt to some kind of absolute block when reserving a platform. I do not consider "drive by sight", "one train staff" and "token block" here, as there doesnt really exist any choose signal for them and they shouldnt anyway be used to enter a station with.

The absolute block'ish reservation all choose signals will make, means that the trains will wait at the choose signal for an empty platform. When it gets one, it will book the entire rute. When the train leaves a tile, it will unreserve it.

To EXIT a station the modern way:
A big station usually have several lenghts of tracks and may even contain waiting positions for trains waiting to enter the line. If one can design the station to have a passing loop for each single track line branching out from the station, the desired working method could be applied to the specific tracks and you would not have any issues. If your lines are only doubble tracked, you would likely also not run into any troubbles, but more of that below.

However, if you are using single tracked lines and cannot design the station to have a passing loop per line, you risk running into troubbles:

I will use the example station diagram I made earlier and we are only talking about the right side of the station:

                                            / /
              __________________P_______E__/ /
West   ______/__________________P__/_\__C___/             East
       _____/___________________P_/___\_EC_____________________
                [station house]
P = Platform end signal or flag man
E = Exit signal
C = Choose signal (entry signal)


(1) East branching line shall use token block:
That means:
P = Absolute block
E = Token block
Results: Train will get released from its track, approaching the token block signal. If the token block signal is at danger, the train will get stuck there, blocking other trains.
Suggestion: Either a 2 block signal that will hold the train at the P signal until the E signal can show clear, alternatively a direction reservation made from the P signal which could then only be made if there are no head on traffic (as that would be in the wrong direction. However, a directional reservation that would not prohibit the train from waiting at the E signal, blocking the junctions.

(2) East branching line shall use one train staff:
That means:
P = Absolute block
E = One train staff cabinet
Results: This will result in the same behaviour as with token block.
Suggestion: Here, a 2 block signal would definetively be the best, since direction doesnt matter in one train staff working method.


(3) East branching line shall use absolute block:
That means:
P = Absolute block
E = Absolute block
Results: This will result in the same behaviour as with token block.
Suggestion: Either a 2 block signal that will hold the train at the P signal until the E signal can show clear, alternatively a direction reservation made from the P signal which could then only be made if there are no head on traffic (as that would be in the wrong direction. However, a directional reservation that would not prohibit the train from waiting at the E signal, blocking the junctions.

(4) East branching line shall use track circuit block/cab signal
That means:
P = Track circuit block
E = Track circuit block
Results: Trains will get released from the platform, approaching the E signal. At the E signal, it will attempt to reserve until the following signal and at the same time, try to create a directional reservation behind that signal (as that will be a doubble faced signal). However, if the next block is occupied, or it fails to reserve a directional reservation, the train will stuck at the E signal.
Suggestion: In this circumstance, it would feel more intuitive that there would be made a directional reservation directly from signal "P". However, a 2 block signal would also solve the problem, just as in normal absolute block.

(5) East branching line shall use moving block signals
This is different than all the other working methods, and to my knowledge, there are no specific swedish way of placing the beacons. However, it would be nice if the placement where consistent with the other types of signals.

(6) East branching line shall use time interval (with telegraph, as it is a single tracked line)
I am sorry that I still dont completely understand the time interval with telegraph. I guess there still are some bugs in it, I will open a second thread about that.
However, my tests shows that it might work. It depends on what I see in my tests are bugs or intended :)

I mentioned earlier that I am currently focusing on the single tracked line, but for the doubble tracked line, a train will also drive out and wait at the Exit signal. If the signal is put far enough away it would not get the big consequenses, but if one are placing the signal according to a map or limited by space or other restriction, it might be closer to the station than a train length, and you would get a train stuck at the points.


Here is the example station again, but in an earlier variant:

                                            / /
              __________________P__________/ /
West   ______/__________________P__/_\__C___/             East
       _____/___________________P_/___\_C______________________
                [station house]
P = flag man
C = Choose signal (entry signal)

Again, it is only the right side we are interested in.
Now, the big difference is that there are no exit signal outside the station. To ease up on things, the pakset could have some "dummy exit signals" that would in simutrans eyes be a token block signal, a one train staff or a time interval signal. The player should place the "signal" on the line as an exit signal and the "signal" could have the form of a sign or a man or something similar.
By doing this way pakset wise, It is much easier to find a common solution and I dont need to write down again which working methods work and doesnt. Obviously, something like absolute block would work flawless without an exit signal.


Suggested solution:

What I then would like to suggest is:

* 2 block signal

I quote from myself my thougths on how I think a 2 block signal should work:

2 block signals should only be possible to code as "absolute block", "track circuit block" and "cab signalling".

When a train approaches the 2 block signal, the signal will find the next signal of any working method in the trains path.
The 2 block signal will attempt to set the next signal to CLEAR by whatever rules the other signal follows. If the other signal cannot be set to clear, eg due to unable to create a directional reservation or not enough time has passed since its last train, the 2 block signal would remain at DANGER.
If the next signal is another 2 block signal, I suppose the logic has to stack, that would effectively become a 3 block signal.
If there is no signal before the next stop in the trains schedule the 2 block signal would act as a normal signal.
The "end of choose" sign could also terminate the 2 block functionallity, making the signal act as a normal signal.

The argument that it would be physically impossible for a signal to detect whether there is a token slot free or to know whether long enough time has gone, I think is not valid. Remember that I very much would like to simulate the logic of the techniques rather than the techniques them self. Say token block: I will not simulate that there are actual tokens inside the slot machine. In my world, a train entering such a signal is written down on paper inside the signal box, and the guy operating the 2 block signal would get his orders from the same people in the signalbox. Therefore, upon requesting "if there are any tokens left" from a distance, I find very realistic. If you dont want it in the british pakset, maybe I could suggest a proposal, that in the dat-file of the 2 block signal, specify which working methods the 2 block signal DOES work with.

Bringing back the diagram again:


                                            / /
              __________________P_______E__/ /
West   ______/__________________P__/_\__C___/             East
       _____/___________________P_/___\_EC_____________________
                [station house]
P = Platform end signal or flag man
E = Exit signal
C = Choose signal (entry signal)


Some examples of what could be made:
P = 2 block absolute block
E = One train staff cabinet
Results: The train will attempt to leave the platform. The signal checks wether the track behind the one train staff is free (E), if it is not, the train will be held back at the platform.

P = 2 block track circuit block
E = Track circuit block
Results: The train will attempt to leave the platform. The signal checks wether the track behind signal E is free. If it is free, it will be checked wether the following signal is bidirectional and wether a directional reservation can be made. If that is also true, signal E will go to CLEAR and so will the P signal. If it is not possible to crete a directional reservation or if it is occupied, both signals will remain at DANGER.



If you would consider adding more flexibility in regards to directional reservation, that would also be most welcome, as especially the track circuit block and possibly also the time interval with telegraph could benefit alot from that. However, I cannot think of a good way to implement it, so I will leave that to Vladki to suggest if he wants :)

Octavius

What is the exact purpose of the exit signal? The platform signal already guarantees that no train will approach the exit signal when it's not showing PROCEED. To control shunting movements or for redundancy? Neither of those are required in Simutrans. So that leaves stations with different working methods on different exits as its only purpose.

I reality, the platform signal would be some sort of hybrid. Depending on the working method of the used branch, the station master (acting as signal) would either show a green flag of hand over a token, depending on whether the used exit uses absolute block or token signalling. In simutrans that doesn't work, as there is no way to know which working method applies to which exit. What may work is a signal/sign/cabinet on the exit that can change the working method of the train leaving, but does not act as the end of a block. It would only work for some combinations of working method. A change to time interval signalling would probably not work.

Your platform signal could be of the absolute block type. It will create a directional reservation on the single track branch until the first signal on the branch (which would be at the next passing loop) and clear the platform signal. When the train passes the sign after the last points of the station, it changes working method. The reserved track would remain the same (although it may loose its directionality), but, after the rear of the train has passed the sign, the track is no longer unreserved behind the train until it reaches the next passing loop to hand in its token.

Or is that exactly what you mean by a "dummy exit signal"?

Ves

Quote from: Octavius on June 23, 2016, 12:13:09 PM
What is the exact purpose of the exit signal? The platform signal already guarantees that no train will approach the exit signal when it's not showing PROCEED. To control shunting movements or for redundancy? Neither of those are required in Simutrans. So that leaves stations with different working methods on different exits as its only purpose.
They are protecting the line. I cannot say why they exist this way and why it has been designed like this. Maybe it was considered easier or safer to have a separate signal there?

Quote
I reality, the platform signal would be some sort of hybrid. Depending on the working method of the used branch, the station master (acting as signal) would either show a green flag of hand over a token, depending on whether the used exit uses absolute block or token signalling. In simutrans that doesn't work, as there is no way to know which working method applies to which exit. What may work is a signal/sign/cabinet on the exit that can change the working method of the train leaving, but does not act as the end of a block. It would only work for some combinations of working method. A change to time interval signalling would probably not work.

Your platform signal could be of the absolute block type. It will create a directional reservation on the single track branch until the first signal on the branch (which would be at the next passing loop) and clear the platform signal. When the train passes the sign after the last points of the station, it changes working method. The reserved track would remain the same (although it may loose its directionality), but, after the rear of the train has passed the sign, the track is no longer unreserved behind the train until it reaches the next passing loop to hand in its token.

Or is that exactly what you mean by a "dummy exit signal"?
Making something like this work would really enhance the experience of signalling old railways. However I guess this is very difficult to code and calibrate, and that is why I was talking about a "dummy exit signal" that would be placed outside the station. The two ones I can think of needed are token block and one train staff, but might need more. But this method would require a 2 block signal.

jamespetts

Thank you all very much for your contributions so far: this is much appreciated. I have not time to consider this in detail now, but will look at it in due course.

In the meantime, I have implemented (but not yet tested) the suggestion made to have signals defined as "longblock" signals behave in the same way as bidirectional signals at present. I have not tested this because I am not sure how to implement this in the pakset (i.e., how to communicate to players with meaningful and realistic graphics and text what these new "longblock" signals do).

I should be very grateful if those working on the Swedish pakset could have a go at testing it there and see whether it works as had been hoped.
Download Simutrans-Extended.

Want to help with development? See here for things to do for coding, and here for information on how to make graphics/objects.

Follow Simutrans-Extended on Facebook.

Vladki

I'll try with czech signals. What should be in the dat file for those signals?

Ves

#111
From the commit on github, it looks like only is_longblocksignal=1  is needed in the datfile.

jamespetts

Ves is correct - just define the signals as long block signals. Make sure that any token block signals, meanwhile, have the is_longblocksignal definition removed. Please let me know when you have tested these.
Download Simutrans-Extended.

Want to help with development? See here for things to do for coding, and here for information on how to make graphics/objects.

Follow Simutrans-Extended on Facebook.

Vladki

Testing server is up and running at server.exp.simutrans.com. At the moment it is running latest simutrans-experimental an pak128.britain-ex. No addons.

Look at the game and the track layout around the city of Trnava. I tried to replicate the real czechoslowak signal layout, with the british track circuit signals.
The differences are:

- double track in reality does not have strictly defined direction for each track. Both tracks can be used in both directions, although there is one "preferred" direction. This is useful in case of failures, delays, repairs, etc. It is not practical to do it in simutrans, so I made the double track strictly defined. No problem with that.
- end of choose signs. Their design replicate (at least in some paksets) end-of-shunt signs, but in real world they would be closer to the station, then the entry signal, and apply only to traffic leaving the station. Such placement would cancel the effect of choose signal on entry, so I put them further outside. No problem with that.
- real departure signals act as 2 or 3 aspect according to the signaling used further on the line towards which the train is heading. For now I just put 2-aspect signals on the beginning of branch lines. It is not optimal, but works.
- crossing - some crossings are interlocked with nearby signals (like in the game at cemetery), but if the blocks are too long (or the line is not split in blocks at all), then the crossings are closed by independent track circuits and have special signals to inform the approaching train that the crossing is closed (or not).
- departure signals for single tracked line should be the new long signals. I did not test them yet. When I do I'll upload an addon pak with czechoslovak signals and modify the game to use them.
- trains should not turn to drive by sight after stopping at an unsignaled stop (trnava predmestie), but continue according to the last signal seen. If a train has to reverse at such stop, token-block or one-train staff should be used.

Please have a look on the game and comment here. There are also some swedish layouts done by Ves.
I'll update the game when I test the long-block signals locally.

Vladki

A comparison of czech and swedish signalling.

System that Ves describes as modern signalling (separate platform and exit signals), was in use in Czechoslovakia on some places, but was never wide-spread. It was imported by Germans during WWII (mechanic signals) and improved (added color light signals) by Czechoslovak railways in 50's, but abandoned in 70's. Now only a few stations use it. The main difference is that the platform signal looks different from exit signal (while in Swedish practice they seem to look the same). I think the main reason for this kind of setup is to keep separated the function of line signaling and station safety system. The exit signal is simple 2-aspect signal (just as any block signal on the line), and is telling only whether the line outside of station is clear. The platform signal is interlocked with switches and tells, whether the path from the specific platform is set - either for shunting or departure.

Before the introduction of these platform signals, the stationmaster (signalman) had to give the departure signal by hand. Arriving train had to stop at the platform before any switches. Switches were secured by key/lock dependencies, but the responsibility was on the signalman. Usually there was one exit signal for each line going out of the station, or a signal with multiple arms (lights) was used - number of arms was telling to which line the path was set. Shunting was signaled by separate signal placed next to exit signal. Only some stations had exit signals for each station track.

However, during 50's and 60's new signals (and safety systems) were developed, which combined the line signaling and station safety system all in one. The new signals had aspects for shunting as well as for departure (including speed signaling). As a result, the special platform signals were abandoned, and each station track got its own exit signal (in place of the former platform signal). These signals allowed departure without the need for hand signal. This setup is much simpler and also more suitable for simutrans. I would use this setup also for earlier eras in pak.cz, even if it would not be historically exact.

Just one idea for T signals - it would make sense for me to have two T signals at station - one on each end. Then one arm would be the entry signal, and the other exit signal. Then also the "stop before points" would make much more sense - it would be valid only for trains inside the station. Approaching trains would have a good chance to see the signal from safe distance. Take into account the fact, that there were no distant signals in the early years, so signaling station entry by signal in the middle of station seems impossible (or dangerous) to me. IMHO T-signal in the middle of station could have been used only as exit signal for both directions in combination with hand signal by signalman. Entry signal could have been the plate signal. See http://www.ekeving.se/si/mek/index.html - the first picture on left is "Skivsignalen stod vid Nybodatunneln", and a few pictures further, you can find almost the same photo, but with distant signal (forsignal). This corresponds with the practice in Austria-Hungary. The early entry signals were called "distanz signal in German" - as they were built on long distance from station, and later repainted and changed to true distant signals (vorsignal in German).
Probably the same development happened in Sweden.


Ves

I finally found a source telling that an exit signal was put up if the line speed exceeded 90kmh! If line speed was lower, I understand that there would be no exit signals.

Vladki, would that actually not be a two faced signal with only one set of graphics and with different function for each direction (choose in one direction and normal in the other)?
That would actually work also for "my" entry and exit signals, especially if one could choose which working methods it should use (I would in that case paint two signals). That could in fact be interesting! :-)

But yes, the original T signals in Sweden where placed at the middle of the station in front if the station house. The stations with the T signal where not the biggest ones (they used plate signals at the entry). Trains approaching a station with a T signal, just had to drive slow enough to actually spot it and possibly stop. And yes it was a safety risk and eventually got replaced with real entry signals.

jamespetts

#116
I have been looking in some more detail at Ves's long post just now. One point of clarification, if I may: in the case of trains not stopping at the station, how do the signals marked "P" behave in both the old and new layouts?

Edit: Looking at this in somewhat more detail - in actually playing Simutrans with the Swedish pakset, what incentive would players actually have to place separate platform and exit signals, rather than just placing the exit signals on the platforms as in later practice (and as was the standard in UK practice)? The only reason that I can see is if one of the exit signals is of the token block type and there is, unusually, no double track lead into the single track section on which a token block signal can be mounted. I am reluctant to implement signalling features that a rational player seeking to maximise the efficiency of the operation within the parameters of the simulation would have no incentive to use.

If, as I understand to be the case, token block working was not used in Sweden in any event, would it not be better simply to use track circuit block signals with normal danger to simulate this sort of signalling, using the new type of longblock signals as the platform signals here? Thus, in Swedish practice, where there were no double track leads, there was also no token block working, and in UK practice, where there was token block working, there were also double track leads, so no situation is likely to arise in which there is a practical need for a double blocking signal.

As to the T-signals (station signals in UK parlance), changing the code to allow for these would be a very large amount of work, as this would require some very significant changes to some quite fundamental assumptions that the signalling code makes at present (e.g. that trains will always stop on the same tile as that on which the signal is located, and will seek to reserve a path from that tile), and this work would not result in any significant economic or operational difference (save for a reduction in the number of signal heads that a player would have to place). Given the large number of tasks that need to be done, it does not seem worthwhile to invest such a large amount of effort in producing double-arm T/station signals for such limited economic/operational reward.
Download Simutrans-Extended.

Want to help with development? See here for things to do for coding, and here for information on how to make graphics/objects.

Follow Simutrans-Extended on Facebook.

Ves

#117
In the old days i think that trains did not pass a station without stopping in some way, but I'm not sure. It could just as well be. Today, if train have a route through a station without stopping, to my knowledge, the signals would just be clear.

Edit:
I have trouble with my internet an did not see you edited your post! A more detailed reply:

I can understand your reluctance to implement this. The only scenario that in simutrans practically would need this, is when using token block or one train staff. I am thinking of a way to make an incentive (within the tools you have already created) for a player to place an exit signal for the working methods existing.

As earlier talked about, having a real staff or token where quite unusual, enough to consider them not being implemented in pak.Sweden.

However, you can find the effects of token block signaling in Sweden:
Many lines has small stops along the way for a small passenger rail carriage to stop on the way between two stations. Clearance has been given to the train to travel the entire stretch between the two stations, and the train will stop and go as it will. This heavily represent the token block working method, just without physical tokens, and to my understanding, the longblock signal doesn't resemble that(?). The token is spoken, as earlier mentioned.
Therefore, I want to have token block working method in paksweden.

Now I understand you if you anyway don't want to implement it. I guess it involves quite a bit of work and bug reporting taking away time and in that case, creating token block stretches will just have to be designed like you write they are designed in England :-)

Vladki

Quote from: Ves on July 11, 2016, 09:43:58 AM
In the old days i think that trains did not pass a station without stopping in some way, but I'm not sure. It could just as well be. Today, if train have a route through a station without stopping, to my knowledge, the signals would just be clear.

I have seen that in some cases, the hand departure signal was given to the approaching train. It sure had slowed down, but accelerated again without stopping.

Quote
However, you can find the effects of token block signaling in Sweden:
Many lines has small stops along the way for a small passenger rail carriage to stop on the way between two stations. Clearance has been given to the train to travel the entire stretch between the two stations, and the train will stop and go as it will. This heavily represent the token block working method, just without physical tokens, and to my understanding, the longblock signal doesn't resemble that(?). The token is spoken, as earlier mentioned.
Therefore, I want to have token block working method in paksweden.

I think this is the same as I have suggested earlier. I think that for all signaling methods (absolute, circuit or even time interval), the clearance was given for the entire stretch of line up to the next signal. Simply - if the train stops in the middle of nowhere, it does not turn to drive by sight. Instead it continues according to the last signal seen. From that point of view, you can think that even absolute or circuit block is a kind of token block, with somewhat virtualised tokens.

What I really do not like about the current implementation of token block is that the track stays reserved behind the train, thus blocking the exit points of the station, until the train arrives to the next station. In reality, the reservation should extend backwards only to the next signal (in whatever direction). If implemented like this, there would be almost no difference between token and absolute block. I still feel that there is no difference in the logic behind both methods, it is just the real world implementation, that uses tokens as an extra safety measure. It is just about the proof of track clearance: absolute block - end of train sign was seen, the signalman pushed the appropriate button, and turned his signal to danger. token block - token was handed from train driver to signalman, circuit block - the track circuit is open, and (optionally) the next circuit is closed. But from the simutrans point of view, all these proofs are equal.

jamespetts

Thank you all for your feedback. I have been working most recently on the one train staff mode, which I think that I have fixed now. Would you mind awfully testing this to see whether it does indeed work properly?
Download Simutrans-Extended.

Want to help with development? See here for things to do for coding, and here for information on how to make graphics/objects.

Follow Simutrans-Extended on Facebook.

Vladki

#120
I have added one-train staff to the network game - see the goods sidings at Nemcanka station. It looks good at first glance. Please check if the layout is how you intended it would be used.

Well, at second glance it is not that good - the reservations for second train entering the on-staff area are not correct... And the more I watch the game the worse it gets. Sometimes the train, upon leaving the on-staff area turns to drive-by sight (ok), sometimes it says track circuit, but ignores any signals and causes deadlock

jamespetts

I think that I have at least partly fixed these one train staff issues; would you be able to re-test? I should be most grateful.
Download Simutrans-Extended.

Want to help with development? See here for things to do for coding, and here for information on how to make graphics/objects.

Follow Simutrans-Extended on Facebook.

Vladki

Train staff seems to be OK. Even if I put the staff cabinets a bit further from each other. Please see the server game if the layout corresponds with your original intention how it should be used.

jamespetts

I have just added a feature in devel-new that I believe has been requested a few times on this thread: trains in the absolute block, track circuit block and cab signalling working methods will not revert to drive by sight signalling at a non-reversing station, but will continue to reserve to the next signal in their current working method.

I should be very grateful if people could test this and report whether it works properly. Thank you for all of your feedback so far.
Download Simutrans-Extended.

Want to help with development? See here for things to do for coding, and here for information on how to make graphics/objects.

Follow Simutrans-Extended on Facebook.

Vladki

Wow, thats good news. I'll test ASAP.

jamespetts

Please note that I have just uploaded several fixes, so please test with the latest version.
Download Simutrans-Extended.

Want to help with development? See here for things to do for coding, and here for information on how to make graphics/objects.

Follow Simutrans-Extended on Facebook.

Ves

This sounds very intresting! I will try it out when I get a chamce!

wlindley

Numerous unrecoverable compile errors from vehicle/simvehicle.cc beginning with

    vehicle/simvehicle.cc:5074:30: error: expected unqualified-id before 'or' token
         onward_reservation_type or;

when compiling with gcc version 4.8.4.  Apparently 'or' is reserved?

jamespetts

Thank you for that: I have changed the variable name in question. Are you able to re-test?
Download Simutrans-Extended.

Want to help with development? See here for things to do for coding, and here for information on how to make graphics/objects.

Follow Simutrans-Extended on Facebook.

wlindley


jamespetts

Download Simutrans-Extended.

Want to help with development? See here for things to do for coding, and here for information on how to make graphics/objects.

Follow Simutrans-Extended on Facebook.

Ves

So, it feels like the signals are in general implemented in the game and it's mostly bug fixing left, what subject do you intend to look at after signaling?
Or do you have some more plans for the signals?

jamespetts

There is one more thing that I want to do with time interval signalling, to do with improving the signalling of stations, which I am hoping to do this week-end. After that, I will have to give some consideration as to how best to proceed next. I will need to spend some time working on more signalling tutorial videos, however, as these are of some importance given the complexity of the new signalling.
Download Simutrans-Extended.

Want to help with development? See here for things to do for coding, and here for information on how to make graphics/objects.

Follow Simutrans-Extended on Facebook.

Ves

There is one issue with drive through stations and choose signals that might need addressing in some way:

Look at the brittish servergame at coordinate (146,111), there you will find a small station on a single track (my old example station in fact), featuring two choose signals and a T-signal.
At the choose signals at either end, there is a train holding, waiting to find a free way through the station to the other end.
The trains try to reserve a route through the station to the end of choose sign, located on the outside of the choose signals.
But since the other train is already waiting there, a deadlock is occured.


What should ideally happen is that one train should be taken onto the platform so the other one can pass.
Would it be possible for the trains in this situation to recognize that there is a passing loop? Ie, automatically take the southermost/westernmost track?
The normal way of doing passing loops is to have only two tracks and put one way signs on them, but in this situation with three or more tracks, the station should ideally not have any one way signs.

Would it be possible to make the choose signals interact with the signals (in this case, the T-signal) on the station, so it could release one train one block and then see if the deadlock resolves?

jamespetts

Resolving this specific sort of deadlock automatically would require some really rather complex additional logic (I am not immediately sure even how in theory it might be done), the time involved in which does not seem entirely worth it given that it can be resolved simply by setting a waypoint so that the non-stopping trains will always take the leftmost path.
Download Simutrans-Extended.

Want to help with development? See here for things to do for coding, and here for information on how to make graphics/objects.

Follow Simutrans-Extended on Facebook.

Ves

I came to realize the other day a difference in the terminology that caused confusion (at least for me) in the beginning. Now it pops up again as I am thinking of the future translation of the game to Swedish. I'm aware that the implemented terminology in Simutrans might be simplified compared to real life, but it is the simutrans way of doing things that I've to relate to:

The term CLEAR, I understand, means
This signal shows the least restricted aspect, ie is it a 2 aspect stop signal, it means that one block is free, if it is a 5 aspect signal, it means that there are 4 blocks free. There is no way of telling how many blocks are free by just looking at the message (CLEAR), you must know which kind of signal it is showing it.
Now the terms for signals at CAUTION (with all their variations) holds the information about how many blocks is free ahead (caution means one block is free and the second is not, preliminary caution means two blocks are free and the third is not etc.).


In Sweden, there is no good way of translating CLEAR. Most articles I've read use the term "kör", which just mean drive. Most aspects, which is not STOP, includes the word kör as an instruction, followed by possible restricting messages. The term CAUTION is not used, because this signal is in fact clear, and that must be presented.

For ease of reading, I translate in the text below kör to CLEAR
The term CLEAR means
This signal allows you to proced to the next signal. Even if this would be a multiple aspect signal, the message CLEAR holds no information about the fortcomming signal. If this signal was to show that also the next signal is clear, the term would be "CLEAR, expect CLEAR". A four aspect signal would therefore show "CLEAR, expect CLEAR, expect CLEAR" etc. If the last signal would show stop, the message would have been "CLEAR, expect STOP".
A standalone presignal would have these aspects: "Excpect CLEAR" and "Expect STOP"


Now, why am I writing this? Simutrans handles it just fine with the english terminology. I just remember it from when we where discussing it a while back (and I could not understand it) and I found that it might be interresting to know how other countries presents their signal messages.

Also, I'm writing this because I want to make a change in the signal infowindow: Make the CLEAR aspect be different, depending on which signal is showing it (Eg, "CLEAR", "CLEAR2" or something similar). For the english translation, all variations of CLEAR would be just "Clear", however, the swedish pakset could then write "Kör, vänta kör" ("Clear, expect clear") accordingly.

jamespetts

Interesting; do feel free to make the change so long as you provide the translations.
Download Simutrans-Extended.

Want to help with development? See here for things to do for coding, and here for information on how to make graphics/objects.

Follow Simutrans-Extended on Facebook.

Vladki

I think the difference comes from the fact that sweden uses blinking distant signals. Thus a 3-aspect signal shows two lights at the same time - one steady and one blinking.  Czech terminology is similar to british, although it does not translate exactly. Best fit would be with austrian or german terms. Stůj - stop - danger, výstraha - caution, volno - clear. However, we have different approach to 4-aspect signalling. Caution (one yellow light) is shown in braking distance (for speeds up to 120 km). In case the block is shorter, a "repeated caution" is shown (yellow + white) on the signal between caution and danger. For higher speeds cab signalling is mandatory, so the driver can start breaking well in advance (one block earlier).

4-aspect line signalling was in place from cca 1960's to 2000. (now abandoned) However repeated caution is used within complex stations or where the main signal is behind a corner.

This would make the czech translation a bit tricky

Ves

Quote from: Vladki on November 08, 2016, 08:05:54 AMThis would make the czech translation a bit tricky
Do you have any wishes in respect to translatable texts?

Vladki

Just to distinguish somehow which caution is on 4. and more aspect signals. English order is: clear, (preliminary caution,) caution, danger. While czech order is clear, caution, (repeated caution,) danger. The aspect in () is only on 4 aspect signals.

So to make proper translation one would need to distinguish caution on 3-aspect signal (or simple distant signal) from caution on 4 and 5 aspect signals.