Topic: 11.9007 bug - Passengers don't transfer

[Michael Hauber]

Running 11.9007 that I compiled out of the 11.x branch.  No passengers will transfer at a station/stop.  Passengers are generated with a transfer destination, but disappear when the first transfer stop is reached.

[jamespetts]

Thank you for the report. Can you upload a saved game in which this behaviour can be reproduced?

[Michael Hauber]

Here is a save game to see this behaviour easily.

http://simutrans-germany.com/files/upload/No_transfer_demo.sve

[neroden]

The file is gone now.

Can you see if this persists on my ncn-devel branch (look up neroden's github)?  There were a lot of problems on the 11.x branch.

[Michael Hauber]

Had a quick look for 'Neroden's Github' and can't seem to find it?

[Junna]

Had a quick look for 'Neroden's Github' and can't seem to find it?

https://github.com/neroden/simutrans/tree/ncn-devel

I can't compile it myself, however... I get:

1>simgraph16.obj : error LNK2005: "void __cdecl mark_rect_dirty_wc(short,short,short,short)" (?mark_rect_dirty_wc@@YAXFFFF@Z) already defined in simgraph0.obj
1>simgraph16.obj : error LNK2005: "void __cdecl activate_ribi_clip(int)" (?activate_ribi_clip@@YAXH@Z) already defined in simgraph0.obj
1>simgraph16.obj : error LNK2005: "void __cdecl clear_all_poly_clip(void)" (?clear_all_poly_clip@@YAXXZ) already defined in simgraph0.obj
1>simgraph16.obj : error LNK2005: "void __cdecl add_poly_clip(int,int,int,int,int)" (?add_poly_clip@@YAXHHHHH@Z) already defined in simgraph0.obj
1>simgraph16.obj : error LNK2005: "void __cdecl display_set_clip_wh(short,short,short,short)" (?display_set_clip_wh@@YAXFFFF@Z) already defined in simgraph0.obj
1>simgraph16.obj : error LNK2005: "struct clip_dimension __cdecl display_get_clip_wh(void)" (?display_get_clip_wh@@YA?AUclip_dimension@@XZ) already defined in simgraph0.obj
1>simgraph16.obj : error LNK2005: "void __cdecl display_set_height(short)" (?display_set_height@@YAXF@Z) already defined in simgraph0.obj
1>simgraph16.obj : error LNK2005: "short __cdecl display_get_height(void)" (?display_get_height@@YAFXZ) already defined in simgraph0.obj
1>simgraph16.obj : error LNK2005: "void __cdecl display_set_actual_width(short)" (?display_set_actual_width@@YAXF@Z) already defined in simgraph0.obj
1>simgraph16.obj : error LNK2005: "short __cdecl display_get_width(void)" (?display_get_width@@YAFXZ) already defined in simgraph0.obj
1>simgraph16.obj : error LNK2005: "short __cdecl display_set_base_raster_width(short)" (?display_set_base_raster_width@@YAFF@Z) already defined in simgraph0.obj

[neroden]

https://github.com/neroden/simutrans/tree/ncn-devel

I can't compile it myself, however... I get:

You are somehow compiling BOTH the 16-bit graphics version AND the 0-bit graphics version (command-line version) simultaneously, which will never work. 

If you are using the Makefile, please check the settings in config.default and make sure that COLOR_DEPTH is set to 16 for the graphics version or 0 for the command-line version.

If you are compiling in some way other than the Makefile, please make sure that you are compiling only *one* of simgraph16.cc and simgraph0.cc -- not both.

[Michael Hauber]

I had the same problem.  I compile by opening the .sln file in Microsoft visual studio, and updating the library and include directory locations.  This worked for the 11.x branch and the master branch.  I selected an option to exclude simugraph0 from the current project and that seemed to work.  This drops the file totally out of the project explorer view, whereas I can see both files in this view for the 11.x branch project, so I'm not sure how it was done there.

And is someone actually updating these .sln files?  Or do the simutrans developers use other compilation software which is nice enough to generate various files required to define a project in other common IDEs?

Anyway I get a working exe out of this which does not have the failure to transfer bug.

[neroden]

I had the same problem.  I compile by opening the .sln file in Microsoft visual studio, and updating the library and include directory locations.

James maintains the .sln file.  I use Linux 100% of the time so I only use the Makefile.

[jamespetts]

Try compiling from the dingliste-cleanup branch and see whether that is any better: I have altered the way that MSVC++ handles the simgraph0.cc and simgraph16.cc files recently.

[Junna]

Try compiling from the dingliste-cleanup branch and see whether that is any better: I have altered the way that MSVC++ handles the simgraph0.cc and simgraph16.cc files recently.

That would defy the purpose of trying to compile nerodens devel branch, though, wouldn't it?

[jamespetts]

I'm not sure that I fully understand - what is that purpose?

[Junna]

I'm not sure that I fully understand - what is that purpose?

Are they otherwise the same versions?

(Updated: Compile on nerodens dinglist-cleanup branch fails with same error.)

[jamespetts]

Can you try compiling from my dingliste-cleanup branch? I have fixed a number of Windows compile bugs on my branches.

[Junna]

Can you try compiling from my dingliste-cleanup branch? I have fixed a number of Windows compile bugs on my branches.

Fails with same error, unless I exclude simgraph0 from project, at which it compiles.

I redownloaded nerodens branch and tried again.

Now it fails with the error:

1>c:\simusrc\vehicle\../simworld.h(2060): error C3861: 'snprintf': identifier not found

[jamespetts]

I have recently integrated this branch into my 11.x branch, and fixed the snprintf issue.

[Junna]

Your (James') merge-private-car now compiles successfully without intervention of any kind - but it crashes frequently in play and cannot load save games what-ever.

I still cannot compile nerodens devel branch...

[jamespetts]

I have been working most recently on the 11.x branch - perhaps you could try that?

However, I am hoping to release the next RC soon.

[Junna]

Ah, it did not come into my head that you were referring to a new branch. Indeed, it now compiles.

However, this still fails loading save games - they crash without any errors shown (windows exception error).

Loading some other earlier saves gives an error message:

"FATAL ERROR: sim_new_handler() - OUT OF MEMORY or other error allocating new object. Aborting program execution"

[jamespetts]

Hmm - odd. What settings are you using to compile? Does this occur with all saved games or only some?

[Junna]

Hmm - odd. What settings are you using to compile? Does this occur with all saved games or only some?

I'm only using the settings included with it - I have not changed anything. It does happen with both debug and release compilations.

The game also crashes a lot randomly in play.

Are you able to load all save games with your own compile?

[jamespetts]

I have not had any trouble loading saved games - what compile preset are you using -  debug or release?

[Junna]

I have not had any trouble loading saved games - what compile preset are you using -  debug or release?

Problem occurs with using both.

[jamespetts]

Are you using the .sln or the .vcxproj file?

[Junna]

Are you using the .sln or the .vcxproj file?

.sln. I was under the impression that one loaded the other. Is there a difference?

[jamespetts]

I think that the .sln is more complete - it's odd if you're having problems with that one. Hmm. What is the name of the .sln file that you are opening?

[Junna]

I think that the .sln is more complete - it's odd if you're having problems with that one. Hmm. What is the name of the .sln file that you are opening?

Simutrans-Experimental. That is the right one, I hope? Could you upload a compile of yours, so I can see if it is the compile that is the issue and not something else?

[jamespetts]

A couple of my what - binaries or .sln files?

[Junna]

A compiled binary, yes. I said compile, not couple, but it's easy to misread; I'd like to be able to determine if the issue is something I'm doing wrong with the compilation or if it is something else.

The ones I compiled, in addition to failing to load save games, crash on the last minute of a month (6.23.xx), but the exact moment is variable (sometimes at 16, 13 or 30 or 42 past 6.23. The new download feature from github is odd and I don't quite get it, as it seems to only offer the first page loaded (i.e. I load "master", it says "master.zip", even if I click on the 11.x branch... I downloaded 11.x by opening it in another browser, but maybe something is still off, who knows.)

[jamespetts]

Hmm - you are downloading .zip files? Wouldn't it be better to check out the repository into a local directory?

As for compiled binaries - I am hoping to release a release candidate to-day, so that should deal with that.

[jamespetts]

I have now released the new release candidate: please try with that to see whether this works any better.

[jamespetts]

Not having had any furhter reports on this issue, I am transferring this to solved bug reports.