[Patch] Improved Timetable Management [V2.31tr SVN15778]

[PhilSophus]

. The auto-fill will then set this waiting time on any stop, unless the real waiting time will be longer.

That's actually a small problem for me. Normally I set the time to 0:02 for all stops but the last ("loop") ones, where the trams wait to catch up (they usually get 30-40mins, depending how many other lines intersect the course of that particular tram). I realise this might be only my style of playing, but I guess - if you don't ask you don't get

That's also a bit like my playing style. Maybe it is possible to use a stop in a depot to let the trams catch up there instead of blocking a station?

That's not really the issue here. The problem is that pshemko wants the desired waiting time to be set even if the real waiting time is longer. And BTW you can't set waiting times on depot orders.

I play similarly although I set the waiting a bit above the real waiting time (but not enough to really catch up delays) and set a longer waiting time at one or both ends of the line or an important transfer station.

@pshemko: As auto-fill now rounds to the unit you selected (i.e. days, minutes or ticks), I wonder if there really is a problem. Okay, it gets rounded up to the next minute, but so what?

[pshemko]

Usually I try to have a few main tram lines intersecting with each others and then another set that service 'local areas' (i.e. they concentrate in one major transfer station). Due to high number of shared stops between various lines I try to keep the stop time there to a reasonable minimum and then let the trams catch up with the timetable at the last stops (which normally wouldn't be shared between the lines). I guess if I could make them wait in the depot that would make timetabling easier, but less realistic, on the other hand one could argue that even with the 'loop' design of the last stops that I use its not very realistic, since we have no one-way tram tracks and all trams effectively clog up the stop very quickly (since they can't move until the timetable time, even if they wait at the very first part of multiple-tile tram stop).

So I would like to ask for either one of the improvements:
1. Ability to set a predefined time per stop, preferably per vehicle type (trains can safely wait 30mins without causing backlogs)
2. Ability to set wait time in a depot (which has infinite capacity )

[PhilSophus]

Usually I try to have a few main tram lines intersecting with each others and then another set that service 'local areas' (i.e. they concentrate in one major transfer station). Due to high number of shared stops between various lines I try to keep the stop time there to a reasonable minimum and then let the trams catch up with the timetable at the last stops (which normally wouldn't be shared between the lines). I guess if I could make them wait in the depot that would make timetabling easier, but less realistic, on the other hand one could argue that even with the 'loop' design of the last stops that I use its not very realistic, since we have no one-way tram tracks and all trams effectively clog up the stop very quickly (since they can't move until the timetable time, even if they wait at the very first part of multiple-tile tram stop).

So I would like to ask for either one of the improvements:
1. Ability to set a predefined time per stop, preferably per vehicle type (trains can safely wait 30mins without causing backlogs)
2. Ability to set wait time in a depot (which has infinite capacity )

The latter definitely is harder to do, as there currently is no code that already does the waiting as in stations. And I don't see how that would help on the issue of minimal waiting time in intermediate stations.

If the first was done, it would certainly be done in a way that timetables take precedence over the predefined time, as vehicles not respecting their own timetable would be quite counter-intuitive IMHO. And if there was no timetable, it wouldn't make sense to have a default time in the first place. The only reasonable place to put such a feature is auto-fill, but I'm not sure that too many different options should be done there. Shouldn't the philosophy behind it be, if you want finer control, do it yourself instead of using auto-fill? BTW, if non-destructive auto-fill goes to trunk it will probably have the option to overwrite waiting times even if they are shorter. At least, that is what Rubidium requested in FS#1124.

Anyway, I'm too tired to think that through, now.

[Tekky]

Sorry for the wait. Next time, I will simply use the new openttd_useful.zip instead of attempting to compile all the external libraries myself.

Here are the new Win32 binaries of ITiM 2.11 with cargodest. I have again supplied two binaries, one with and one without the passenger reduction patch.

I have scanned both archives with VirusTotal and no viruses were found.

Legal Information:
Please note that this program is copyrighted and licensed under version 2 of the General Public License, which means -among other things - that it is distributed without any warranty. Please see the file COPYING.txt for further information, which is included in the .zip archive. I have supplied this notice because it is mandatory according to section 1 of the GPL.

EDIT: This version of ITiM is now out of date. More recent binaries are available later in the thread.

[rbn2903]

Hi PhilSophus,

I've a question related to the Headway GUI. Everything you click keeps beeing activated, but you can't click (read: activate) an order where there is no train waiting/driving. OTOH if a order is already activated and all the trains continue, it will say "No vehicle at this order.", which works just fine. So is this behaviour intended or did you never really notice this? I'd really like it, if I'd be able to activate such orders with no verhicles.

Thanks,
Rainer

[PhilSophus]

So is this behaviour intended or did you never really notice this? I'd really like it, if I'd be able to activate such orders with no verhicles.

At first it was intended or rather a side effect I knew was sub-optimal but accepted as it was just a lot easier to do it this way. As you probably noticed, orders are moving all the time, so it's not really easy to determine the order on which you clicked. Instead the code determines which vehicle is at the same y-position. The order of this vehicle must be the order you clicked on.

But as I wanted <CTRL>-click to work for empty slots, too, empty vehicle slots know the related order for some time now, too. So, the reason for it is actually gone. I'll put it on the todo list.

You just remind me of a related issue: I would also like to avoid that jumping around of orders, when vehicles advance to the next order. But I do not really have a satisfying solution for that, so far. As long as there is a complete timetable and the vehicles are about on-time, the maximum number of vehicles on an order is known, but what to do, when there are more vehicles than estimated because something is wrong in your network? So, at the moment this feature is flagged "too much work for too little benefit"


Edit: And yet another, somewhat unrelated issue. There is some activity around FS#1124 (i.e. non-destructive autofill) which is also part of ITiM. There will probably be an option to turn off the keeping of waiting time. So, Rubidium wondered if there was still a use for destructive autofill then. As you are having non-destructive autofill as part of ITiM for quite some while now, it might be useful to know for which purposes you would still need destructive auto-fill (if so that is).

[Lilman424]

I have ITiM 2.10 and am completely losing all my money, but i'm thinking it's to do with cargodest.

I had a train line looping between 4 cities (Abergrove, Nudingsay Bridge, Great Trestable and Trendean in the saved game) as well as a longer straight line between Teningdon Green, Senhall, and Messide Heath. I was making quite good money, and then I connected the two lines and quickly starting losing all my money. I'm not sure whether this is a matter of poor balancing within cargodest or a matter of bad route building.

It seems to me that my routes are reasonable, but I'm no expert on rail routes >_>

[rbn2903]

ok thanks

But I do not really have a satisfying solution for that, so far.

I have a thought on that: i don't believe you can out a definite height on an order, because what happens if you play with breakdowns enabled (which i do not) oder have a huuuuge deadlock, because you are combining freight with passengers in one network and some station out of nowhere stops accepting a specific freight. BUT: maybe you could reduce the jumping thing a bit. Lets say, you have a shared order with 16 Stops + 16 Ways to there and 60 vehicles sharing this order. then the quotient would be 60/32=1,875 ~ 2 Two lines for each order. But as in my Orders the stopping time is often a lot shorter than the driving time, you would have to take the duration of each order into account. so you'd get: 1 line for 1st order (station), 3 lines for 2nd order (long trip), 1 lines for 3rd order, and so on,...

Of cause I made these numbers up, but I'm sure, you'd be able to put these in little functions depending on "number of vehicles", "number of orders", "duration for each order" and "total duration of timetable". And if anything bad happens the orders would of cause bea ble to show one more line if neccessary.

I hope this idea is helpful.

[...] it might be useful to know for which purposes you would still need destructive auto-fill [...].

I believe that no answer could also be an answer: i always use non-destructive auto-fill. For me it works just fine So was this an announcement to get to know peoples opinions on that?

EDIT: sry, the word "vehicle" hates me, I always write "verhicle"

[PhilSophus]

It seems to me that my routes are reasonable, but I'm no expert on rail routes >_>

You should keep a look on your train list and check the trains that make a big loss. Train 7 seems to be stuck at Messide Heath Mine due to a signaling issue (put a normal path signal facing to the station there).

No, wait, this is very strange: The train says "Loading/Unloading" not "Waiting for free path", despite having the order to go non-stop to another station. Skipping orders frees the train, but nevertheless it just shouldn't happen, no matter what you did. Do you have a savegame before the train got stuck? Do you remember if you did anything with that train before, like skipping or changing orders? I'm not sure, whether this is an ITiM-specific problem or a (potentially already fixed) trunk issue. Could you please update to 2.11 (it's based on 100 trunk revisions ahead, but otherwise unmodified) and watch if it happens again.

@rbn2903: Your suggestion is about what I was thinking about: Giving each order order_time/average_veh_separation+1 slots when it is timetabled. The problem is, that there still may be more vehicles in a slot, so the code has to account for this case, making it a lot more complex.

[rbn2903]

Giving each order order_time/average_veh_separation+1 slots when it is timetabled. The problem is, that there still may be more vehicles in a slot, so the code has to account for this case, making it a lot more complex.

I have to say, I just had a first look into your code. But as my "Notepad2" doesn't highlight it well and I guess I'm in hypoglycaemia, it's hard to unterstand for me right now So I'll do some pseudocode (please don't blame me for errors, I usually write in VB.Net and never in C++ yet!):

Code: Select all

int number_of_linesA = order_time / average_veh_separation + 1;
int number_of_linesB = number_of_vehicles_at_order;             //I guess, you use "this->vehicle.Length()" for that, but I'm not quite sure.
int number_of_lines  = max(number_of_linesA, number_of_linesB); //I hope, there is such function "max", but you'll get, what I mean ;)

Wouldn't that be a possibility? Anyway you are able to determine how many verhicles are at the same order right now, so that can't be the problem, or is it?

[Lilman424]

No, wait, this is very strange: The train says "Loading/Unloading" not "Waiting for free path", despite having the order to go non-stop to another station. Skipping orders frees the train, but nevertheless it just shouldn't happen, no matter what you did. Do you have a savegame before the train got stuck? Do you remember if you did anything with that train before, like skipping or changing orders?

I have a feeling i can guess what's going on with that train, though i have no clue why. Its destination is a steel mill, and i would bet that it went to the steel mill, unloaded its coal, but the steel mill was near its max stockpile (using PBI), so all the coal wasn't accepted. It then returned to the coal station and is where it is now. i'll see if i have a savegame previous to this one and watch it happen.

edit: yes, i did have a save game previous to the train getting stuck (attached), as well as one soon after. oddly, not only did it never get stuck, but my train network is also profitable well down the line. It seems like the combination of those trains running for no good and me LACKING their profits was what put me so far in the hole. Indeed, if I load the save i posted earlier and merely stop the two coal trains (since they are not blocking the routes of any other trains, i start making money back, although slowly. i'll try to keep an eye on those two trains and see if they get stuck again, and see if i have any clue how it happened

edit2: perhaps it has something to do with the fact that using cargodest for....well, cargoes, no cargo is generated unless the other station accepts it. So no coal is generated at Messide Heath Mines unless the steel mill station is currently accepting it. I'm just throwing this out there, though, i'm not really sure how it would affect it.

[PhilSophus]

I have to say, I just had a first look into your code. But as my "Notepad2" doesn't highlight it well and I guess I'm in hypoglycaemia, it's hard to unterstand for me right now

That doesn't astonish me at all. The stuff I did there is quite tricky, certainly not self-explaining and not commented at all, I'm afraid.

So I'll do some pseudocode (please don't blame me for errors, I usually write in VB.Net and never in C++ yet!):

Code: Select all

int number_of_linesA = order_time / average_veh_separation + 1;
int number_of_linesB = number_of_vehicles_at_order;
             //I guess, you use "this->vehicle.Length()" for that, but I'm not quite sure.
int number_of_lines  = max(number_of_linesA, number_of_linesB);
             //I hope, there is such function "max", but you'll get, what I mean ;)

Wouldn't that be a possibility? Anyway you are able to determine how many verhicles are at the same order right now, so that can't be the problem, or is it?

Don't bother too much. Actually, I have an idea how to do it, but it may break some assumptions of the code (namely that there is at most one empty cell per order), so I have to be careful not to break anything. The "vehicle" list also contains the empty cells. Depending on whether a cell is empty or not, it stores the order number (or rather sub-order number which is a concept I invented to distinguish between the traveling and the waiting) or a pointer to the vehicle. I make sure there is an empty cell if no vehicle is at an (sub)order and the list is sorted by the sub-order number. As, I'm doing the painting of the order list along with the vehicle list, I just have to draw a separator and the order text when the sub-order number changes in the associated vehicle cells. So, no, I don't count the vehicles at an order explicitly while drawing the lists. But, I think I do so, when drawing the summary panel, when an order is selected.

@Lilman424: Thanks for the savegames and the description. I've never seen such a behavior although I have played a long game with ITiM 2.0x/2.1x, some ECS vectors (stockpiling enabled) and cargodest for all cargos. Anyway, it must be some very special circumstances that trigger it, so that doesn't mean it couldn't have happened. I'll have a look at it. Unfortunately, I can not tell whether this is an ITiM or a cargodest issue, order handling is certainly something I touched

According to Murphy's Law it's something in the OrderList code, which I just put in a Flyspray task , but since Murphy's Law is recursive it might be something completely different if I'm looking there first



Edit: Faster than you probably thought: V2.20 is out featuring the ability to select orders without vehicles and preventing jumping orders in headway window in many cases when timetable is complete.

[Tekky]

Here are the new Win32 binaries of ITiM 2.20 with cargodest. I have again supplied two binaries, one with and one without the passenger reduction patch.

Legal Information:
Please note that this program is copyrighted and licensed under version 2 of the General Public License, which means -among other things - that it is distributed without any warranty. Please see the file COPYING.txt for further information, which is included in the .zip archive. I have supplied this notice because it is mandatory according to section 1 of the GPL.

EDIT: This version of ITiM is now out of date. More recent binaries are available later in the thread.

[rbn2903]

Faster than you probably thought: V2.20 is out featuring the ability to select orders without vehicles and preventing jumping orders in headway window in many cases when timetable is complete.

Wow, I did not see this coming! Thanks!

[PhilSophus]

I just uploaded V2.21 mainly featuring some bug fixes and a little feature, I had already planned for some time:

• Featurette: Show group name of vehicle in vehicle lists when vehicle has no name.
• Bug fix: Mouse-wheel scrolling was not working properly in station timetables.
• Bug fix: Truncation of very late timetables did not work with headway.
• Bug fix: Change date cheat totally confused headway.

Edit: I just noticed that the cargodest variant included the cd-reduce-passengers patch by accident. I just replaced it by the correct one. Sorry to the one person that already downloaded it

[Tekky]

Here are the new Win32 binaries of ITiM 2.21 with cargodest. I have again supplied two binaries, one with and one without the passenger reduction patch.

Legal Information:
Please note that this program is copyrighted and licensed under version 2 of the General Public License, which means -among other things - that it is distributed without any warranty. Please see the file COPYING.txt for further information, which is included in the .zip archive. I have supplied this notice because it is mandatory according to section 1 of the GPL.

EDIT: This version of ITiM is now out of date. More recent binaries are available later in the thread.

[rbn2903]

I know, someone asked this before, but PhilSophus couldn't you make it easier to reach the headway window? Maybe by adding a button in the train window or by combining the functionality of headway and timetable window (or maybe even the order window), as you can already set the scheduled time for an order in the headway window, which is kind of redundant? (yes, I'm aware, that "CTRL+Click" pops up the timetable window directly, which saves one click.)

Keep up the excellent work!

[PhilSophus]

I know, someone asked this before, but PhilSophus couldn't you make it easier to reach the headway window? Maybe by adding a button in the train window or by combining the functionality of headway and timetable window (or maybe even the order window), as you can already set the scheduled time for an order in the headway window, which is kind of redundant? (yes, I'm aware, that "CTRL+Click" pops up the timetable window directly, which saves one click.)

I just had an answer ready when I got an idea: Why not make the order window, timetable window and headway window tabs of a single unified window (like the four tabs of the details window)?

Anyway, I'll think about it, but I don't want to promise anything. Maybe I just fallback to the easier solution, which is increasing the minimum height of the vehicle view and adding a button.

[rbn2903]

I think tabs would be ok, but I never really used the details window and when I had to, I thought it was complicated. (But I guess it's not up to me to decide that )

But I just got another idea. Once I made enough money to build a good network extensively, I usually forget about the fact that I have some trains which are currently "autofilling" (is that a word? ) and opened vehicle windows annoy me while building new routes. Maybe one could get a message (such as a newspaper) when a train/vehicle finished its autofilling round so one could start the headway immediatly? What do you think of this?
Or: The vehicle becomes the "head" right about when it has finished the autofill precedure. "Set intervall" wouldn't be available until then.

Greets, Rainer

[audigex]

Based on the idea of realism (and also based on "this would be a great feature" rather than "this is definately do-able easily and you should") I (and I would be quite confident saying quite a few others) would really like to be able to set one or two stations to be used as the "catch up" station if the train is running early/late.

Would it be possible to have a station that's set to be the one the train waits at if early (to take up the slack), so that we can have larger terminus stations? And also an "if late, leave immediately" or "if late, skip" level?

The first part is probably akin to the way most trains are timetabled to arrive at their terminus 15mins-half an hour before their departure time (more at big stations) to allow for both cleaning, restocking and to avoid setting off late.

The second parts I would guess would be related to conditional orders? But this would be good also for making trains do "quick stops" at stations if they are late. Eg "stay five days or, if late, leave as soon as loading complete".

And a final one (just popped into my head) would be, if the user selected auto-fill, but trains are all taking significantly more/less time to complete a loop, to automatically auto-fill again, with similar slack levels? This sounds a bit wierd to me as I try to describe it, but could the user just set a "travel slack" or "station slack" which is added onto the auto-fill, and then the train can auto-fill again (with the same slack) if the journey time improves/reduces significantly. Akin to the company adjusting a timetable to account for congestion, or after a track upgrade.

Just some ideas/wishes from my point of view - I love what's been implemented so far (well, in theory I like the idea, not had a chance to try it yet)

(((To those few who regularly do this when someone makes a suggestion: please don't just post "that's too difficult/not possible" or "try it yourself" - firstly I'm not demanding it, I'm suggesting/requesting features I'd like: not saying what should be done, and secondly if I had the ability to code [or the time to learn] at a level high enough I'd be helping way more than loitering in the forum. Reasons people do/don't like the ideas are welcome, however xD)))