Signals ignored?

[Vladki]

Hi, I managed to download and compile the latest stuff from git (devel-new). And found some interesting things:

1. Trains ignore one-way signals (basic signal clicked multiple times) and pass through as if there is no signal.
2. Trains reserve only one or two tiles ahead (max speed cca 50-60 km/h), trains getting stuck on the track in opposite directions.
3. If the train goes in the right direction of signals, then it sometimes reserver many blocks ahead, or stops in front of empty block.

4. I get frequent announcements that trains cannot choose platform because it is too short. But:
- there is no choose signal at all
- the platform is indeed long enough
- the train avoids the last tile of platform, thus not un/loading completely

This makes me wonder if there is some serious redesing of signaling underway?

[jamespetts]

There is indeed a serious redesign of signals underway, which is not entirely complete. Some of the things that you have reported are intentional. Specifically:

(1) one way signals now operate as in real life: they work as signals in only one direction, but allow trains to pass in both directions - one way signs are needed to restrict the direction of trains;
(2) this is drive by sight mode, which is important for street trams - trains default to this when first starting until they come upon a signal, which sets the working mode for the train;
(3) this is too non-specific to understand fully - this may be intended, one or more bugs or some combination of the two;
(4)
(a) I do not think that this, even in Standard, is limited to choose signals;
(b) I suspect that this is linked to (c); and
(c) this is potentially a bug - can you give steps reliably to reproduce, or, preferably, a saved game in which it can be easily and reliably reproduced?

Thank you very much for your feedback.

[prissi]

The message about not choosing because of too short platform you get only for a choose signal. Otherwise this algorithm is not called upon.

[Vladki]

There is indeed a serious redesign of signals underway, which is not entirely complete. Some of the things that you have reported are intentional. Specifically:

(1) one way signals now operate as in real life: they work as signals in only one direction, but allow trains to pass in both directions - one way signs are needed to restrict the direction of trains;

Great news !!! This was one of the highlights on my wishlist. I'm going to test it. Prepare for bug reports

(4)
(a) I do not think that this, even in Standard, is limited to choose signals;
(b) I suspect that this is linked to (c); and
(c) this is potentially a bug - can you give steps reliably to reproduce, or, preferably, a saved game in which it can be easily and reliably reproduced?

Save is here: https://uran.webstep.net/~vladki/simutrans/test.sve
You'll find a new map - one track connecting coal mine and powerplant. When the train is loaded you get the message that stop is too short. Watch it when it arrives to power station. It will leave the last tile of terminal stop unused (might be related to #2 - drive by sight). When you build one more tile of track after the stop, the message will be gone. However, it was not easy to prepare the test. First I made only 2-tile long train with GWR pannier tank, and 2-tile long stops, and everything was OK. Only when I made longer train and stops, I got the message. Another way that seems to fix the problem is to put the waypoints exactly at the last tile of stop.


And one unrelated thing - after a bit of messing with stations and timetables, the coal mine stopped sending coal to the power station... Is there something new I'm missing apart form the usual - industries are connected, stations are connected, right type of platform, and vehicles, train can pass the whole track...

[jamespetts]

Thank you for that: I will look into the saved game when I get a chance.

As to industries, there are the changes to the in-transit percentage calculation. Were there goods already in transit at this time?

[Vladki]

At the beginning, the train did a few trips full, but later the supplies stopped, with all stuff delivered and used by power plant. So 0 in-transit, 0 input store at customer, 0 waitng at stop, full output at coal mine.
Would you consider using JIT2 in experimental? I think it does exactly what is needed withou complicated round-trip  calculations.

[jamespetts]

I had better have a saved game to have a look at this and determine what is happening, as this does not sound correct. As to JIT2, that would require a huge effort to implement which effort could better be spent elsewhere on the long queue of things that are important for balance, and was implemented for Standard because, I believe, it was not possible to have the in transit percentage set by reference to travelling time as travelling time is not calculated in Standard.

[Vladki]

It happened with the same save game as referenced above.

[Vladki]

One more question about the new signalling. I see that presignals work like real distant signals. But i can't find the right distance from stop signal. Any hints at what distance to put them?

[jamespetts]

That will depend on the braking distances of your trains: they should be at least as far in front of the stop signals as the longest braking distance of the trains used regularly on the line.