Topic: [r7697] Crossroad choose signs are confusing/bugged

[Spacethingy]

I'm not sure if this is a bug or not, but it is certainly weird behavior.

Placing a choose sign on a crossroad creates the sign facing north. This works fine.

In order to place a sign facing in another direction, you have to click and drag over the crossroad in the required direction.* THEN, a further click on the crossroad leaves a single sign facing in the correct direction (which functions OK).

At the point marked *, you get a set of 3 signs facing in all directions except the one you specified. (At least so I can tell from the direction binary value). This is bugged as only two signs appear. Also, none of the signs have any effect on convoys in any direction.

Any chance the click-and-drag could simply create the single sign facing in the dragged direction?

Cheers! 

[Ters]

I didn't think that should be possible at all. You can't do it with signals. Nor does it make sense, because signs telling you which way to go must come before the intersection. And since choose signs are signs that can make vehicles stop, well, you don't want them stopping in an intersection.

[Spacethingy]

They don't stop in intersections ever anyway. The behavior I described above couples with the rule that convoys don't enter intersections if there isn't a clear exit. This results in the following behavior:



...which is quite useful as it has gained me an extra bus length!

[Ters]

That's just even more confusing.

[Leartin]

But in your example, vehicles would not choose any other way but the direction of the station, even if the same station had another open slot to the south. Therefore, the two examples are functionally different in the first place. If you position the choose sign one space east to the intersection, it should be functionally identical.
Now how often are two intersections right next to each other so that this wouldn't work?

[Dwachs]

I think it should not be possible to build signs/signals on crossings, and also not possible to construct ways through signal to create crossings on the signal tile. Any more opinions?

[Spacethingy]

I think it should not be possible to build signs/signals on crossings...

As a user, I disagree. As I showed above, being able to place single signs on a crossing is a valuable function for fine tuning truck routes in built up areas, bearing in mind how large a fraction of road tiles in cities are crossings. Sure, there are other ways of doing it, but this is the simplest for certain situations, albeit with a rather broken method for building it.

Yes, you're quite right, having a direction sign AFTER a junction would be weird if the sign functioned to give a vehicle a choice of direction for the junction it just passed, but that's not what the signs do in that position.

If you decide to remove this, what will happen to save-games with signs on crossroads? Will they become corrupted?

Cheers, thanks for looking into this!

[TurfIt]

I also disagree. I make extensive use of signals on track junctions to increase throughput. It's on my perpetually growing todo list to have a better interface for this... Removing this feature would be a great loss - too many of these hidden weird construction order things have already been removed. (connection to side of bridge!)
IMHO you should be able to not only put them on junctions, but finely control all the combinations of one way directions created.

To the original question:
pak128 (and others) have their road signs misaligned. As road signs control the entrance to a tile, signs should face traffic at the entry edge. Unlike rail signals that control the exit from a tile.
For the picture with the choose sign above, the signs placed as shown in the second picture is closer to correct (it should have three signs in the graphic) when placed on the junction {not a pak alignment issue - code issue to properly support this}, but the vehicles should be waiting as shown in the first picture without the extra tile.

[Leartin]

correct me if I'm wrong, but when I tried this, the vehicles would not choose a different route to their destination if that required them to turn at the junction the sign was buildt on. But if the sign comes (graphically) before the junction, that very junction would have to be included in alternate route searching.

The graphic in pak128 may be misaligned in a technical sense, but you should always keep in mind where a choose-sign makes the most sense - right before a junction, as shown in the second screenshot.

Add these together and you come to the conclusion that the mistake is to have choose signs work on tile entrance rather than tile exit.


EDIT: I created Screenshots as a showcase. All Vehicles have the middle station as their destination. The first screenshot shows the sign on the junction, yet the vehicles refuse to go to the south. Since in p192c the choose sign has the same "misalignment" as in p128, this is actually somewhat correct, since the sign comes AFTER the junction (visually) - but would be very misleading if the sign was placed on tile entrance.
The second screenshot shows the sign replaced one tile to the east. Currently, this has exactly the same result, since neither the junction nore the sign-tile will be entered.
If it was changed so sign tiles could be entered, the second screen would not be possible, since one of the cars would cross the junction and end up on the tile. But in that case you could place signs like in Spacethingy second screen without loss of space.

[Ters]

As road signs control the entrance to a tile, signs should face traffic at the entry edge. Unlike rail signals that control the exit from a tile.

Isn't this more a difference between different signs/signals than road versus rail? The choose sign has vehicles waiting on its tile, just like the choose signal. And it seems the no entry sign works the same everywhere.

[TurfIt]

The choose sign actually has vehicles waiting on the tile before the sign, hence the large gap that's previously been complained about. The fact the choose sign doesn't choose the route to the south in Leartin's screeenshot is a consequence of the sign being on the junction is not really supported. If official proper support were added, it would make sense for the route to be searched from there.

I'll have to check if there a difference in the no entry signs. For road, the sign definitely prevents entry onto the tile containing the sign, even if pak128 graphically shows the sign aligned with the tile exit.

[Leartin]

The fact the choose sign doesn't choose the route to the south in Leartin's screeenshot is a consequence of the sign being on the junction is not really supported. If official proper support were added, it would make sense for the route to be searched from there.

Yes, but that comes down to "if it would work, it would work".

Pretty much everything you could do with choose signs on an intersection could be done with normal choose signs, up until now the only issue brought forward was the "extra buslength" that could be saved in the first example, a result that could easily be mimicked by placing the sign after the intersection right now, and could be saved by changing the sign cod to work on tile exit.

In order to have properly working signs on an intersection, you would need to have a way to toggle between the 15 possible sign "states" (=directions in which signs face) in an obvious-beginner friendly way, since it seems fairly reasonable to place them on the intersecting tile so beginners will run into it, or block them from being placed easily (like requiring the control key as a special feature)
Furthermore, right now what you called 'misaligned' sign graphics are common. IIRC you even need to 'misalign'them to show up correctly in a curve. They would not work well on intersections, so the change in behaviour would require new (or at least repositioned) graphics.


Personally, I see no gain in choose signs on intersections, while I do see issues with them.

[Ters]

The choose sign actually has vehicles waiting on the tile before the sign

It must have been the alignment of the vehicle graphics that confused me. The vehicles look almost completely inside the tile with the sign (although not quite at the sign), but the query tools reveals that it is not. However, that means that if road vehicles stop one tile later, they will appear to have passed the sign before stopping.

In order to have properly working signs on an intersection, you would need to have a way to toggle between the 15 possible sign "states" (=directions in which signs face) in an obvious-beginner friendly way, since it seems fairly reasonable to place them on the intersecting tile so beginners will run into it, or block them from being placed easily (like requiring the control key as a special feature)

On top of that, one-way signals ensure that traffic move in only one direction. But when there is a one way signal facing south (can only be passed by northbound trains), should it be legal for a train coming from east or west to turn south, or for a train from the south to turn east or west even if the signal is red? (And how to you show such a signal-intersection with ASCII art?)