Chill's patchpack v14_7

[ChillCore]

one last question:

can you recommend any AI for this patchpack?

Not really as they all behave differently.

If you enable infrastucture sharing and want control over all companies yourself then 'Iddle' or 'Iddle more' is the way to go. -> download content.
My own AI 'Aiai_Caramba' builds only roads and nothing else. -> AI forum.
All the other ones have a different personality, some focus on trucks others on rails and they all have their own building style ...
Just keep in mind that if you enable the more height levels setting and/or play on very big maps they will be a little slower.

There are a few AI's in the AI forum that are not available from the download content. When downloading those make sure you have the needed libraries also available for them.

If you used MinGW and MSYS then you can just do the "make bundle" command.

The way i did it when i used vs2008 was to download another binary that was created as a bundle and replaced the exe.
Because "make" is a MSYS only thing.

That sounds like a bad idea and might give problems in some cases.
The grfs that come with OpenTTD itself are version specific (whenever something changes) and/or the language files provided in that bundle might not match if not replaced also.

That is also the reason that none of the files that come in a downloaded binary should be put in the shared folder ...

tried to compile, failed!
...

I get the same error.

I worked out arround by changing the lines in economy.cpp :
...

I did not get that error ...
@MagicBuzz:
Before I change it in my version I have a little question.
Is there a reason you did ((int)_date_daylength_facor) the first time and (int)_date_daylength_factor the second?
Also the maximum value for the daylength factor is 255, maybe something smaller then an int would work? int16 instead?
I can not test as for me it compiles correctly in that area.

@ColdIce:
Thank you for the build log.

[MagicBuzz]

There is not "real" reason I used ((int)day_length_factor).

It's resulting of several tests I made while trying to fix the error. The first time I didn't noticed there were 2 lines in error, so I tried to fix manyu times the first line, with (), different types, etc.

You can remove the useless ( ).

[MagicBuzz]

In overflowsafe_type.hpp, at line 100.
=> Add the following line, and let the economy.cpp file without explicit cast.

It will compile correctly.

Code: Select all

FORCEINLINE OverflowSafeInt  operator /  (const uint8            divisor) const { OverflowSafeInt result = *this; result /= (int16)divisor; return result; }

It's cleaner I guess, and this time it uses only a int16 variable.

Here is the extracts of the build log where there are still warnings :

Code: Select all

1>------ Début de la régénération globale : Projet : strgen, Configuration : Debug Win32 ------
2>------ Début de la régénération globale : Projet : version, Configuration : Debug Win32 ------
2>C:\Program Files\MSBuild\Microsoft.Cpp\v4.0\Microsoft.MakeFile.Targets(32,5): warning MSB8005: La propriété 'NMakeCleanCommandLine' n'existe pas.  Elle sera Ignorée.
1>  alloc_func.cpp
1>  strgen.cpp
1>  string.cpp
1>  strgen_vs100.vcxproj -> C:\in\ottd\chillcore\projects\..\objs\strgen\strgen.exe
2>  Microsoft (R) Windows Script Host Version 5.7
2>  Copyright (C) Microsoft Corporation 1996-2001. Tous droits réservés.
2>  
3>------ Début de la régénération globale : Projet : langs, Configuration : Debug Win32 ------
3>  Generating afrikaans language file
3>  Generating arabic_egypt language file
4>------ Régénération globale ignorée : Projet : generate, Configuration : Debug Win32 ------
4>Projet non sélectionné dans le cadre d'une génération pour cette configuration de solution 
3>  Generating belarusian language file
[...]
3>  Generating welsh language file
5>------ Début de la régénération globale : Projet : openttd, Configuration : Release Win32 ------
5>  airport.cpp
[...]
5>  copy_paste.cpp
5>..\src\copy_paste.cpp(284): warning C4805: '|' : mélange risqué de type 'bool' et de type 'int' dans l'opération
5>  crashlog.cpp
5>..\src\copy_paste.cpp(716): warning C4258: 'tile' : définition provenant de la boucle for ignorée ; la définition provenant de la portée englobante est utilisée
5>          ..\src\copy_paste.cpp(697) : définition de 'tile' ignorée
5>          ..\src\copy_paste.cpp(514) : définition de 'tile' utilisée
5>  currency.cpp
[...]
5>  openttd.cpp
5>..\src\network\network_server.cpp(1696): warning C4018: '>=' : incompatibilité signed/unsigned
5>  order_backup.cpp
[...]
5>  viewport.cpp
5>..\src\vehicle.cpp(871): warning C4800: 'int' : valeur forcée à la valeur booléenne 'true' ou 'false' (avertissement sur les performances)
5>  waypoint.cpp
5>..\src\viewport.cpp(397): warning C4146: opérateur moins unaire appliqué à un type non signé, le résultat sera non signé
5>..\src\viewport.cpp(398): warning C4146: opérateur moins unaire appliqué à un type non signé, le résultat sera non signé
5>  widget.cpp
[...]
5>  thread_win32.cpp
5>     Création de la bibliothèque C:\in\ottd\chillcore\projects\..\objs\Win32\Release\openttd.lib et de l'objet C:\in\ottd\chillcore\projects\..\objs\Win32\Release\openttd.exp
5>  Génération de code en cours
5>  Fin de la génération du code
5>  openttd_vs100.vcxproj -> C:\in\ottd\chillcore\projects\..\objs\Win32\Release\openttd.exe
========== Régénération globale : 4 a réussi, 0 a échoué, 1 a été ignoré ==========

[ChillCore]

In overflowsafe_type.hpp, at line 100.
=> Add the following line, and let the economy.cpp file without explicit cast.
....

I'll do that and hope that nothing else gets broken ... thank you once again

- For the warning in network/network_server.cpp
Does changing line 1696 to

Code: Select all

if (cs->last_frame_server - cs->last_token_frame >= 5 * (uint)DAY_TICKS) {

solve it?
I do not get a compiler warning with the old or with the new line so that is a bit hard to check.

Fix it by dividing by -4 instead of 4 or multiplying by -((int)TILE_HEIGHT). I had that problem too.

Deviding by -4 breaks the intended behaviour.
Replacing lines 397 and 398 with

Code: Select all

	a = Clamp(a, (((uint)GetMaxTileHeight() + 1) / 4) * -(int)TILE_SIZE, (int)(MapMaxX() * TILE_SIZE) - 1);
	b = Clamp(b, (((uint)GetMaxTileHeight() + 1) / 4) * -(int)TILE_SIZE, (int)(MapMaxY() * TILE_SIZE) - 1);

seems to still work as it should but I do not know if the compiler warnings are gone without someone testing as for that file also I do not see the warnings here.

[NekoMaster]

There seems to be a problem with loading AI's which makes the IS patch a little useless in single player games (no Ai's = no companies to share infrastructure with)
I have all the libraries you can possibly get through the content downloader, but the ai's still crash.

So far i've tried nocab, admiral ai, and pathzilla

Will the AI problem ever be fixed?

[MagicBuzz]

AdmiralAi, PaxLink and Chooper works fine on my configuration.

So it looks like something is broken with your version.

Do you have the file "compat_1.1.nut" in your "ai" folder ?
=> Without this file, any AI can't load.

[ChillCore]

I have no problem with the AI's neither but there is a problem with the download content. One I have noticed before and I made a topic about.
I am updating my download content in my nightly build and will digg my topic up once I know what content exactly is problematic. (the AI's are amongst them.)
Only Rubidium can fix it if the problem is the same as before, so you will have to be patient.

In the meantime here is a screenshot of the testgame with AI's that I have been running for a few minutes since you posted, NekoMaster, to check ...
The only AI that you mentioned that was not running is Pathzilla as I do not seem to have that one in my linux' shared folder.
Also AdmiralAI(v22) seems to be stuck doing something but does not crash.

[NekoMaster]

AdmiralAi, PaxLink and Chooper works fine on my configuration.

So it looks like something is broken with your version.

Do you have the file "compat_1.1.nut" in your "ai" folder ?
=> Without this file, any AI can't load.

I have that now, but I still get crashes with NoCab and Admiral AI (the index AIAbstractList does no exist) and pathzilla (the index HasNext does no exist)

[T-Unit]

Hey guys. Been away from the community for about a year. This patchpack seems to be full of very good looking patches.

I attempted to apply the last diff in here but have been struggling for the past hour. I know you dislike people who do this but I wondered if anyone had a compiled 64bit version they could post.

Thanks
T

[ChillCore]

AdmiralAi, PaxLink and Chooper works fine on my configuration.

So it looks like something is broken with your version.

Do you have the file "compat_1.1.nut" in your "ai" folder ?
=> Without this file, any AI can't load.

I have that now, but I still get crashes with NoCab and Admiral AI (the index AIAbstractList does no exist) and pathzilla (the index HasNext does no exist)

I have bumped my old topic here: http://www.tt-forums.net/viewtopic.php?f=31&t=47507

It seems that the AI's that crash on you can not read the needed libraries, they seem to be downloaded but not corectly installed or OpenTTD can not read them ...
What you can try is copy the libraries from the download content folder to the normal ai/library folder unzip them and try again ...
You best do this while OpenTTD is not running as the AI's and libraries are detected on startup if I am to believe the output of the console when OpennTTD is started from there.

Hey guys. Been away from the community for about a year. This patchpack seems to be full of very good looking patches.

I attempted to apply the last diff in here but have been struggling for the past hour. I know you dislike people who do this but I wondered if anyone had a compiled 64bit version they could post.

Thanks
T

Hello T-unit,
The 32bits binary MagicBuzz posted does not work for you?
I need to post a new diff as you can read from the last few pages ...
I am waiting for some feedback before doing so and I still need to bump my source beyond r21480 ...

Also what is the problem you are struggling with?
If the build fails you can try adding the line MagicBuzz mentioned before compiling.

In overflowsafe_type.hpp, at line 100.
=> Add the following line, and let the economy.cpp file without explicit cast.

It will compile correctly.

Code: Select all

FORCEINLINE OverflowSafeInt  operator /  (const uint8            divisor) const { OverflowSafeInt result = *this; result /= (int16)divisor; return result; }

[T-Unit]

Hello T-unit,
The 32bits binary MagicBuzz posted does not work for you?
I need to post a new diff as you can read from the last few pages ...
I am waiting for some feedback before doing so and I still need to bump my source beyond r21480 ...

Also what is the problem you are struggling with?
If the build fails you can try adding the line MagicBuzz mentioned before compiling.

In overflowsafe_type.hpp, at line 100.
=> Add the following line, and let the economy.cpp file without explicit cast.

It will compile correctly.

Code: Select all

FORCEINLINE OverflowSafeInt  operator /  (const uint8            divisor) const { OverflowSafeInt result = *this; result /= (int16)divisor; return result; }

I hadn't tried that as I wasn't sure whether it would work. But I have now and it does appear to be. I seem to have managed to apply the patch without errors to the source code but that was after making many mistakes. When I posted I was downloading and installing Microsoft Visual C++ 2010 Express with a slow network and couldn't quite see where I was meant to go from there. An entirely noobish act on my behalf. Thanks for the heads up on the 32bit one though.
Thanks
T

[NekoMaster]

Still a no go on the AI's after extracting them from their TAR's. I don't know what's causing this to happen because I haven't really done anything special with my install

[ChillCore]

I hadn't tried that as I wasn't sure whether it would work. But I have now and it does appear to be.

Glad it is working now.

Still a no go on the AI's after extracting them from their TAR's. I don't know what's causing this to happen because I haven't really done anything special with my install

Have you untarred the libraries also?
You may have to make the library folder yourself in the ai folder. You should not not change anything in the content download folder to not disturb your mental sanity (and also to not cause problems later on).

[NekoMaster]

Still a no go on the AI's after extracting them from their TAR's. I don't know what's causing this to happen because I haven't really done anything special with my install

Have you untarred the libraries also?
You may have to make the library folder yourself in the ai folder. You should not not change anything in the content download folder to not disturb your mental sanity (and also to not cause problems later on).

I have extracted the libraries to the library folder in the ai folder, but still nothing works.

[ChillCore]

I am sorry NekoMaster but I think that the problem is on your side somewhere, maybe in your shared folder?
I have tested again, this time with the windows binary for v11_4 posted by MagicBuzz.
Admiral (v25), NoCab (v498) and PathZilla (v6) are doing just fine.

I simply have no clue why they refuse duty for you ...
If you try them in clean trunk they work or not?

[NekoMaster]

I am sorry NekoMaster but I think that the problem is on your side somewhere, maybe in your shared folder?
I have tested again, this time with the windows binary for v11_4 posted by MagicBuzz.
Admiral (v25), NoCab (v498) and PathZilla (v6) are doing just fine.

I simply have no clue why they refuse duty for you ...
If you try them in clean trunk they work or not?

I dont have a shared folder (the one that appers in your documents) because I run everything from my install folder (to keep things together and portable). AI's work in other patched builds like cargodist and in clean stables and trunk.

EDIT : I think I've solved my problem, I moved all my essential stuff to a new folder and used the more recent build that MagicBuzz made (after i was asking about a Win 9x build)

[ChillCore]

EDIT : I think I've solved my problem, I moved all my essential stuff to a new folder and used the more recent build that MagicBuzz made (after i was asking about a Win 9x build)

You think .... ?
Can you please comfirm? Please, please, please, as I was just going to lose my mental sanity and step in a little rowing boat, cross the ocean and kick some sense into your computer.

I think you were using one of the incomplete , now removed, versions MagicBuzz had posted before.
Which shows how important it is to distribute complete binaries ...

To meet MagicBuzz however I do not have to cross oceans, just a little 2.5 hour trip across the E17 wich happens to run not even five minutes from where I live.
(Just kidding MagicBuzz I am a non-violent person in general. )

[NekoMaster]

Im pretty sure its working, I ran for about a year and I didn't get any crashes. It's kinda cool that the build in pause mode cheat is now an advanced options :3

[Cadde]

Not that i give a **** but timetables suck **********!
I tried setting 2 trains in a timetable.

Running daylength 20, 222 ticks per minute.
Gave the trains 12 minutes of travel time (10 minutes where given by the "automate" feature)
And 3 minutes to stay at each station, thus a total of 15 minutes for each station (30 minutes for the entire schedule.)
Set the start date (time) to xx:57 and xx:12 respectively.
Give it no more than 10 (irl) minutes and the trains have decided time tables made by GOD (That's me) are for losers and decide to rewrite them on every stop...

Which makes me wanna...

1. Kill a thousand kittens with FIRE.
2. Crash the two trains just to prove a point that hadn't it been for my ability to build signals they would have DIED a long time ago.

So ya, no need to fix it or anything but it would be cool if they actually worked...
/Rant

[ChillCore]

I see, I need to disable the start time button also when automate is on ...

Can you try again please.
This time do not enable the automate button but use the autofill button, when filled adjust and set start times.
Better now?

Automate keeps adjusting the time over and over again without user input. When GOD clicks the button it means that GOD gives contol over timetables out of his hands to do more important things ...

/me likes kittens ...

Edit:
/ me also likes children ... mind your language please. (I do understand your frustration )