Cargodest crash + some graphical glitches

[Nagyzee]

Hello!

I'm using r2293 currently and I have run into a cargodest bug. Savegame and necessary info are attached below.

The problem is the following:
I have a bus station (called Osnaburg Valley in the save) that doesn't accept passengers but acts as a changing place between routes at a road junction. As soon as a bus leaves passengers in the station I can't click on it any more. The station is working correctly though, buses are changing passengers there as intended. But when I try to click on the station the game crashes every single time. Also if I extend the catchment area of the station causing it to accept passengers on its own right the problem goes away instantly.

I don't see why cargo couldn't change vehicles at an interim station that doesn't accept it so I guess this is a non-intended bug. Please look into it when you have some time. Thanks a lot in advance.

[Nagyzee]

I have also experienced some graphical glitches in the same game. On the first and second picture you can see that in case of bridges spanning over sea two levels above there is seashore under one of the bridgeheads where there should be none.

On the third picture you can see a problem with text that is haunting me for some time now. Sometimes the names of toolbars (?) show up only partially and the x goes missing. If I hover my mouse over it the extent of the missing text changes back and forth. I have the same problem with displayed station names sometimes, too.

[JGR]

As for the cargodest problem, I will look into that shortly.

The bridge coastline issue seems to me to be a side-effect of the smooth shore graphics. You could always change back to the default ones as they should not display glitches.
It's not really feasible to redraw the bridges, as the dimensions and pillar placement coordinates are standardised (and have been for some time).
I don't know who drew that coastline or what set it's from, (perhaps check your GRF list?), but you may wish to try bringing it to their attention if you can find them/the set thread.

As for the toolbar issue, I've not observed such an issue. Is it just certain toolbars, or all of them?

It would help matters if you could post your ttdpatch.cfg, the savegame showing the cargodest problem, and indicate which version of TTD you have (DOS or windows, and which language).

Thanks for reporting these issues...

[Nagyzee]

The ttdpatch.cfg and the savegame I've already posted. It's at the end of the first post. I use the Windows version, English language.

Also in the meantime I added a 'load only' order for one of the stops and it seems to have screwed up routing. Now passengers wanting to go beyond that stop which has the 'load only' order have become unable to route. Is cargodest supposed to work with the special go to orders? Load only and unload only would be extremely helpful orders in a cargodest game.

As for the toolbar issue it's regularly happening with the railway station toolbar and with station names on the map. Most probably only at these two places, or at least I can't recall seeing it elsewhere.

And thanks for the fast answer on the shoreline problem. It's the New water v0.7 GRF by Leppka, coded by Lakie. If I turn it off the problem still remains though in that case it's definitely harder to notice and less disturbing. (There is no water at the wrong place then but only a little yellowish shoreline.) You can take a look at this problem in the above savegame, too. There is a bridge showing this glitch only 2-3 tiles away from the bugged bus station.


Sorry for keeping you busy with so many problems and thanks again for the help.

[JGR]

Sorry for not noticing you had already attached those

As for the load-only orders, I designed cargodest to still work properly with those, however I've not tested it as I couldn't think of a meaningful scenario in which I'd actually need to use it...
I'll have a look at the code and see what's going on...

As a temporary measure I can suggest using the other station construction GUI (ie. not using the locomotion GUI). I may well have a look at the code for that though and see if I can spot anything obviously wrong.
I use the traditional station name type and have not observed any problems with those, so try that and see if it fixes the problem...

Looking through your cfg file, I first noticed that forcerebuildovl is set to off.
The recommended setting is on, unless you're *sure* to delete the ovl file whenever changes TTDP version or modifying your config file. (Inevitably people forget to do this, so it is best just to set it to on).
Edit: Actually, I'm not sure about this.

I will look at the savegame and start debugging later on, I ought to do some work now...

[Nagyzee]

Well, I have two bus routes in this test game which share the first two stops. The second being the junction stop that is bugged, let's call it B. So I have an A-B-C-D-E and back route and an A-B-F-G and back route. If I set the stops at B in the outward direction as load only then passengers travelling on the first route won't use the buses of the second route for the first hop and vice versa. Otherwise they do which is suboptimal.

Another scenario is when you want to keep your regional lines separated from city lines. With load only you can make it so that your regional buses/trains can stop at more stations in a town collecting passengers yet passengers that want to travel only inside the town won't use them and take up valuable space.

---

In the meantime I managed to make a quite good screenshot on the toolbar and signs glitches. See below. It's not really important though as it's not hindering me in the game in any form. And the station signs usually correct themselves if I move away from that area on the map and then back or at worst when I reload the game. Still, it doesn't look nice. Also if opened when the railway station toolbar is already open then other building toolbars seem to be missing the x sign in the upper left corner and some text as well, but only then.

[JGR]

As for load-only orders, after some looking at the code and thinking, I suspect that some potentially not insignificant rehashes may be needed to get them to work well with cargodest.
The problem is that on arrival at a station, the single hop route creation/update routine builds a route from the last station to the current station. In the case of load-only orders, there are suddenly multiple "last stations", of which only the last is saved.

In your scenario, how does cargo get to point B, if neither of the routes unload cargo there?

I designed cargodest so that regional and express type scenarios divided cargo automatically. As routing is based on estimated time to target through the next hop (ie. fuzzy metric based packet switched networking). It's possible to make cargo be routed more strictly by adjusting some of the config parameters. (In particular setting cdstrtdiffmax to 0 or cdstrtcmpfactor to 65536 will only permit cargo to use the single quickest route to the destination (which usually means not unnecessarily using local lines, unless your express lines are overloaded or relatively slow)). However, the defaults should also do the same in most cases (and work much better than the above when networks have mesh aspects).
Using lots of load-only orders brings the question of how cargo gets to those stations (unless you're using them as pick up points only).


I've not been able to reproduce the dialog box text title issue at all. So I'm not sure what the issue is.
Try renaming ttdpttxt.dat, re-running mkpttxt.exe and/or fiddling with some of the GUI/game options and see if the problem goes away.


As for the station clicking crash, I've reproduced that and will aim to fix it ASAP.
However, today I really ought to be to be doing my uni worksheets on PDEs and fracture mechanics instead , and on the weekend, I'll be busy and not using a computer as I've got some... business to take care of .

[Nagyzee]

In your scenario, how does cargo get to point B, if neither of the routes unload cargo there?

Both routes would unload cargo there on the return leg. They'd be on load only when coming from A and on unload only (or simply left on normal) when coming from the other direction. This means that B offers the chance for passengers to switch from one route to the other without going into A. For example passengers going from F to C can spare the B-A and then A-B distance.

And don't worry about the time. I'm building a map now and just test parts of it. It might take weeks till I'm finished with the pace I'm progressing. But I wouldn't hurry you otherwise either.

The bad news is that I'm experiencing some problems with through signals right now so I might have to give you even more work to do. Apologies.

[PaulC]

The bridge coastline issue seems to me to be a side-effect of the smooth shore graphics. You could always change back to the default ones as they should not display glitches.

As you can see from the following screenshot, the glitch does not appear to be related to the water grf:

bridge coastline.png (5.48 KiB) Viewed 5283 times

Of course, you could always obscure it by planting some trees...

[JGR]

The crash when viewing the station window after unloading routed cargo at a station which had a previously empty cargo slot due to not setting the enroutesource cargoslot field in the routed cargo unload code should now be fixed as of r2295.

[wallyweb]

... should now be fixed as of r2295.

So I went to http://nightly.ttdpatch.net/latest/ perchance to find a TTDPatch-nightly-r2295-win32.zip for installation 'pon my drive known as hard,
But alas, the file was not there to be found ... a changelog.txt does not an application make,
So I proceeded to the Parent Directory/ but once again, the trunk was similarly bare.
Perhaps a compilation was not compiled, or perhaps some greedy downloader took all the file.
Would that someone would a correction provide in order that we may all enjoy JGR's fine fix this day.

[DaleStan]

No build logs whatsoever. Methinks the compile farm broke.

[WWTBAM]

/me wonders who or what broke it and how long it will take to fix. Probably as long as it takes.

[michael blunck]

Well, cargodest seems to have matured, works fine until now (I´m in 1930). Good work.

However, that bug with "window shading" reported for r2263 seems to be still there:

Code: Select all

TTD V0019EE00 Crash Log by TTDPatch 2.6 alpha 0 r2293 (Windows)

Exception C000001D at 0187:0063B8B6

EAX       EBX       ECX       EDX
00000002  00000007  00000000  000000D5

ESI       EDI       ESP       EBP
004D0FD4  004F8F44  00D4F834  0065BD0B

DS        ES        FS        GS        SS        Flags
018F      018F      1397      0000      018F      00010202
0000FFFF  0000FFFF  00000037  ########  0000FFFF (Segment limits)
0040F700  0040F700  0000F300  ########  0040F700 (Access rights)

Bytes at DS:EIP
0F 0B C3 80 FA 00 74 08 80 FA 09 74 03 80 FA 0D

Stack Dump:
00D4F848  00606A29  00000001  00000016  00000000  0063B279  06090001  0065BD0B
0056FD24  004D0FD4  009D027E  00D30002  018F003E  00C2103E  0056FB3A  004D0FD4
00400170  003E00D3  00554D5F  0056F16C  004F8651  000000D0  0056BAEA  0055B1EE
0055B091  00D4FDA4  00405A18  00000000  81A37BF9  00D20000  00000001  6F6E6D6C

Handler Stack Dump (at 018F:00D4F7D4):
00004003  00000000  400BE420  0000000A  00000000  00001397  0000018F  0000018F
004F8F44  004D0FD4  00000007  000000D5  00000000  00000002  0065BD0B  0063B8B6
00000187  00010202  00D4F834  0000018F  00606AF1  004D0FD4  004D0FD4  00000170
00D4F848  00606A29  00000001  00000016  00000000  0063B279  06090001  0065BD0B

Patch flags:
DFFEFFBF  1FC81BFF  E1F96DFD  185664E5  00000000  00000000  23A83A9D  A0800002
000000A8  00040008  00000000  00001000  00000001  0000000A  01222204  00000000
94000400  00016666  003FFFFF  000001F4  00000002  00000003  000005DC  FDDB005A
009607D0  01900096  016D005A  76260280  13B14EC4  01E00280  13880258  07B75555
003C0064  370A0514  19280A28  5014140A  14501E14  1E14141E  003F3C1E  EF081400
F0F0F0FF  FF0001F0

On another note: would it be possible to make available an updated mkpttxt.exe to get the new strings translated?

regards
Michael

[JGR]

OK, that bug is fixed now. I didn't notice that it had been reported...

As for mkpttxt.exe, I don't know who if anyone is maintaining the compile farm (or even if it works ATM)... I suppose it wouldn't be too difficult for them to change the script to upload that too...