Chill's patchpack v14_7

[ChillCore]

Just post your savegame, one can see the list before loading the game (see attachment) ... and with the developers tools set to on on one can load games while missing GRFs ... -> !! For testing/checking only !!

By custom GRFs do you mean GRFs that you have written and/or modified yourself? Those you may PM to me. If you are not the original author do not post them please.
If you have "normal" grfs in your game that that I can not find and are absolutely needed for me to load the game I'll let you know. (I have a big collection already.)

I will test again with higher daylenght factors as my last (the one mentioned in the CaroDist thread) was with daylenght set to 1.

About the other two problems you mention ... I'll have to do some testing.
I'll try and change the DAY_TICKS CargoDist uses to the original ones if you are sure that solves the problem (after I am completely done bumping and changing everything that needs bumping and changing).

[timofp]

I had to upload it to megaupload, it's a 20mb savegame (big map i know, i created it myself)
I'm using R21900M
I'm also not using any custom grf's

a few examples of wrong links that should not be there,

*from Station Condinbury Noord to Station Butford Centraal
*from Station Condinbury Centrum to Station Butford Centraal
*from Station Sardstone to Station Kinfingfield
* and so on....

Savegame
http://www.megaupload.com/?d=0C6MLJB6

[Coxx]

I started that game with r20982, later switched r21667 (all builds from this thread). I noticed that trains now only run on 2/3 of their topspeed, I also got a lot of error messages about lost ships. Now I upgraded to r21900. The trains are ok now, but the ships are still lost. It seems to be a problem with NPF. Switching to the original pathfinder solves that problem, but ships now get lost for other reasons .

Great patchpack, thank you!

[jimmydanny]

Anyone here wanna build me a OTTD with the newest pack and post it here/RapidShare?

My x64 screws everything over for some odd reason, and I am not removing Windows to get Linux just for this game.

Jim

[ChillCore]

timofp,
I can load your game fine, even with a few GRFs missing, and will check better in a bit while adjusting CargoDist to ignore the daylength factor.


Coxx,
Thank you for the kind words.
Can you please PM me or point me at the csplatforms and the 2CC ship set GRFs. I can not find them ...
I can not load your game without it crashing. Also I need to know the exact version of the W2W houses set in your game as I think my replacement is what makes the game crash. (*)
There have been some improvements recently in regards of ships getting lost ... so the next version should work better.


jimmydanny,
Page 55 has a precompiled Win32 binary of my last posted version.



Talking about crashes ...
Even when staying on the title game the game asserts after a while.
"Assertion failed at line 128 of ~/src/core/pool_func.hpp: this->checked != 0"
My current version is r22116M and I have not yet checked if the same still happens in clean trunk.

[Alberth]

Talking about crashes ...
Even when staying on the title game the game asserts after a while.
"Assertion failed at line 128 of ~/src/core/pool_func.hpp: this->checked != 0"
My current version is r22116M and I have not yet checked if the same still happens in clean trunk.

We have no such report currently, afaik.

Recently, checks were added to verify that you call Pool::CanAllocateItem() for every pool item that you make.

[ChillCore]

Thank you for the reply Alberth.

I know the previous reports about these pool crashes have been closed but r22088 suggest to me that there are still cornercases in trunk.

Also I am not yet sure if I correctly adjusted my code to reflect r22111 to r22116 as I had just bumped before posting and was searching for Coxx's missing GRFs when it crashed, still at titlescreen but with missing NewGRF list in front ... so I do not know from where it was called or what happened.
The only thingh I do know for sure is that no GRF was involved as it was the titlescreen.
I can try to reproduce it in a debug build and trace it as it is reproducable with my current code ... I do not know how usefull this would be as my code is heavily modified.

Would it hurt a lot if I did the same as has been done in r22088 or do I better not do that?

[Coxx]

...
Can you please PM me or point me at the csplatforms and the 2CC ship set GRFs. I can not find them ...
I can not load your game without it crashing. Also I need to know the exact version of the W2W houses set in your game as I think my replacement is what makes the game crash. (*)
There have been some improvements recently in regards of ships getting lost ... so the next version should work better.
... .

Sorry for sending a savegame with such a mixed array of grf´s. Wouldn´t help finding out what´s wrong.
2cc ships are here:
http://www.tt-forums.net/viewtopic.php? ... 60#p729660
w2w (both versions) are here:
http://www.tt-forums.net/viewtopic.php?t=40234, Comparing the screenshots, I would say I got the second one. Am I allowed to post the one I have, if neccesary?
The CS Platforms I cannot find anymore, they are supposed to be found here: http://ttd.tycoonez.com/, but they are not. I don´t think they are responsable, since I´ve got the same problem without them.

Two more savegames from different games, same problem. This time without CS platforms.

[Alberth]

Hai ChillCore,

I know the previous reports about these pool crashes have been closed but r22088 suggest to me that there are still cornercases in trunk.

That might be the case indeed, the check was added very recently.

r22088 is only to make sure that a release does not suffer from any such corner case. Since in general there is still room in a pool, it is worse to crash due to a missing check than it is to blindly create a new pool item without checking.

Would it hurt a lot if I did the same as has been done in r22088 or do I better not do that?

Since it is in the title screen and reproducable, I'd test it in clean trunk, especially if the timing is reliable.

As for removing the check, obviously you can do that (or more easily, remove the assert). However you are then disabling a safety check, designed to catch bugs, so it does not seem like a good idea to me (ie it similar to assert(), also added to ensure the code behaves sanely).

[APDAF]

I need help installing the patchpack, could someone help me please.

[ChillCore]

Sorry for sending a savegame with such a mixed array of grf´s. Wouldn´t help finding out what´s wrong.

That's OK, is it your savegame and it gives me the oppertunity to expand my precious collection.

I found the 2CC ship GRF.
I tried the three GRFs posted at the W2W thread (there is one at the end of the thread too) but none of them is the correct one ...

I have found the csplatforms GRF, it is hiding in the stations zip but it seems to me that the links are broken on the frontpage.
When hovering over the link to the stations grfs it points at ~/?id=24 but is located at http://ttd.tycoonez.com/?id=152 instead.
I found that one via the coopserver which points at http://wiki.openttd.org/Tycoonez.com:mu ... Czech_Sets , from there I could acces it. Yay new toys.
/me goes looking for some other GRFs linked from that wiki page that I could not find from the front page when searching earlier.

Anyways ... I loaded your savegame after re-compiling my version agains r21900 and I think you need more buoys for your ships.
They seem very spaced out it seems to me. Something like 20 tiles apart should be close enough according to what I have read somewhere (but do not quote me on that).

Also one of your savegames (IIRC Jones&co) has a lot of vehicles with void (-> invalid) orders, most likely caused because of removing stations ... If you do not see that it is caused because of me missing GRFs.

I know the previous reports about these pool crashes have been closed but r22088 suggest to me that there are still cornercases in trunk.

That might be the case indeed, the check was added very recently.
r22088 is only to make sure that a release does not suffer from any such corner case. Since in general there is still room in a pool, it is worse to crash due to a missing check than it is to blindly create a new pool item without checking.

Would it hurt a lot if I did the same as has been done in r22088 or do I better not do that?

Since it is in the title screen and reproducable, I'd test it in clean trunk, especially if the timing is reliable.

As for removing the check, obviously you can do that (or more easily, remove the assert). However you are then disabling a safety check, designed to catch bugs, so it does not seem like a good idea to me (ie it similar to assert(), also added to ensure the code behaves sanely).

I have not yet been able to reproduce it in clean trunk, win binary nor self compiled.
So most likely something is not yet (correctly) updated in my code.
Will have to do some more digging I guess.

I need help installing the patchpack, could someone help me please.

Assuming you want to install the pre-compiled binary ...

Unzip the zip to a folder of your choice where you have write access, add openttd.cfg next to the .exe before running (so you do not mess up your shared one).
Put the flat_blacktiles.grf and copy_paste,grf in the data folder.
If you do not have OpenGFX, -SFX and -MSX (or the original files if you have those) in your shared folder, add them to the data folder also. (shared or not)
Double click openttd.exe and you're good to go.

[APDAF]

Thank you.

[ChillCore]

^^^ My pleasure.




I found my crashy problem ... Three guesses ...
1. ... No.
2. ... Nope.
3. As usual ... I forgot to make some needed changes in CargoDist.

Anyhoo while that issue is fixed, I have the title screen running for about half an hour now and another reported_bug.sav I could no longer load without crashing loads again, I found two other issues. Who would have guessed ...

1. I do not think it's OpenTTD's fault but gcc 4.4 and 4.5 can no longer build debug builds using the "./configure --enable-debug=3" command. (see codeblock after issue 2)
Lucky me, I still have my Ubuntu 9.04 installation and could test with gcc 4.3 which compiles it correctly and gave me the needed backtrace to get an indication on what was wrong with my code and fix the error.

2. I did not get Ubuntu 9.04 to compile with lzma ... liblzma-dev is not recognised as a package ... and installing lzma-dev did not solve my issue.
Anyone familiar on how to solve this? I would appreciate any help as that would enable me to debug compressed savegames since I do not think that I will be able to build debug builds on 10.10 by tomorrow ...


Here is the output when trying to compile a clean trunk debug build using gcc4.5:

Code: Select all

[SRC] Compiling pathfinder/yapf/yapf_rail.cpp
In member function ‘bool CYapfRailNodeT<Tkey_>::IterateTiles(const Train*, Tpf&, Tbase&, bool (Tfunc::*)
  (TileIndex, Trackdir)) const [with Tbase = CYapfReserveTrack<CYapfRail_TypesT<CYapfRail1, CFollowTrackT
  <(TransportType)0u, Train, true, false>, CNodeList_HashTableT<CYapfRailNodeT<CYapfNodeKeyTrackDir>,
  12, 16>, CYapfDestinationTileOrStationRailT, CYapfFollowRailT> >, Tfunc = CYapfReserveTrack<CYapfRail_TypesT
  <CYapfRail1, CFollowTrackT<(TransportType)0u, Train, true, false>, CNodeList_HashTableT<CYapfRailNodeT
  <CYapfNodeKeyTrackDir>, 12, 16>, CYapfDestinationTileOrStationRailT, CYapfFollowRailT> >, Tpf = CYapfRail1,
  Tkey_ = CYapfNodeKeyTrackDir]’:
cc1plus: internal compiler error: Segmentatiefout
Please submit a full bug report,
with preprocessed source if appropriate.
See <file:///usr/share/doc/gcc-4.4/README.Bugs> for instructions.
make[1]: *** [pathfinder/yapf/yapf_rail.o] Fout 1

Yes, the docs folder points at 4.4 but I do have 4.5 installed.

Can someone on Ubuntu 10.10 comfirm this please before I go pester the peeps who maintain gcc++.
I think 4.4 is installed by default as I installed 4.5 myself to test.



Now where was I before all me trouble started... Oh yeah, trafficlights, towngrowth and updating linkgraps faster.

Faster linkgraphs is already done.
If you want to test yourself before I have posted the next version, open moving_average.cpp and replace DAY_TICKS with ORIG_DAY_TICKS (2 times in the code and 2 times in the comments.)
Linkgraphs are updated much faster as is expected but I do not yet know if that solves all the issues timofp had noticed ... I did not get to test much yet between updating Ubuntu and finding the crashy thinghy ...

I have a newer version of that towngrowth patch ...
After digging a bit deeper in the code I found that in toyland food is fizzy drinks and goods is sweets so I only had to add some strings and the code to display the correct ones depending on climate. I reverted the toys code I had alreay added but I may re-add it later after getting the basics to work correctly. (-> goods and food).
Will post in a bit after having checked the code for obvious errors. (Sorry for the delay, Dwight, but with the issues I had to solve first ...)

Now that I have a hopefully crashfree game again I can try to add the trafficlights patch. Noone reported errors for a few days now so I guess it is pretty stable. (Unless everybody waits to bombard me after I have included it ... )


Ps:
I have not yet forgotten about your crashreport Muzzly ...

[Eddi]

did you try xz-dev instead of lzma-dev? some distributions have "lzma" package mean lzma1 and "xz" package mean lzma2. openttd needs the latter.

[ChillCore]

Ah, I knew there was another name for lzma2 but had forgotten what it was.
Thank you Eddi.

Unfortunately no xz-dev or lzma2-dev is to be found in the synaptic repository under Ubuntu 9.04 ...
Which leaves me two options I guess. Either I find the source and compile myself for 9.04 or I uninstall gcc 4.5 from my Ubuntu 10.10 installation and install 4.3 instead untill it is fixed ...


/me needs debugging builds and needs to be able to load compressed savegames at the same time, wheeee... ... eeeee... ... ee,
/me whipes tears and goes google ...

[Nickel_Plate]

Running r21900 with your patchpack.

Now your have brought income inline with daylenght, i am transporting people around on buses but am
not earning any TRANSFER credits due to the amounts being so small and a percentage of the transport cost.

[timofp]

Running r21900 with your patchpack.

Now your have brought income inline with daylenght, i am transporting people around on buses but am
not earning any TRANSFER credits due to the amounts being so small and a percentage of the transport cost.

hmm what you are describing might be the problem that causes all my vehicles to have negative profit if i look to the total in the vehicle windows, this only happens with daylength higher then 1, i described the problem in the previous pages.

and that together with the problem of non disappearing and wrong links in the linkgraph (wich i believe chillcore fixed by now) made me switch to play with daylength 1 again, but higher daylength also has some nice advantages for my playing style, so i hope i can use it again in the future

[Sensation Lover]

I have daylength setted to 5 and i have no problems with vehcs incomings. Look at the attached picture.

[Nickel_Plate]

I have daylength set to 32 days. I'm only getting payment for passengers, no payment for passengers
who TRANSFER from one bus to another.

If you have daylength set under 10 you don't seem to have this problem.

[oberhümer]

http://www.tt-forums.net/viewtopic.php? ... &start=110

I'd be very interested in this.