New map features

[aantono]

I'm using zlib version 1.2.8 installed via homebrew (pkg mgr for OSX)

After making the 'template' changes you suggested, the compiling FINALLY succeeded, and I was able to build a Mac bundle.

BTW, here is what my final/working ./configure looks like:

Code: Select all

./configure --enable-static  LDFLAGS="-stdlib=libc++ -std=c++11 -Wno-narrowing" CFLAGS="-stdlib=libc++ -std=c++11 -Wno-narrowing" CC=/usr/bin/clang++ CXX=/usr/bin/clang++

So I'm attaching it to this post for people to try out.

I'm also attaching a patch.diff of all the changes I had to make to get the thing to compile. Hopefully you won't mind applying them to the code, so I can continue to provide OSX builds in the future.

[cirdan]

Glad it worked!

I'm also attaching a patch.diff of all the changes I had to make to get the thing to compile. Hopefully you won't mind applying them to the code, so I can continue to provide OSX builds in the future.

Thank you. However, I cannot blindly apply your patch to the source tree; what fixes your build may very well break somebody else's. But I will look into each change in detail and decide how to deal with it (for instance, I will likely fix some of your casts in a different way).

I'm using zlib version 1.2.8 installed via homebrew (pkg mgr for OSX)

Can you post your /usr/include/zlib.h and /usr/include/zconf.h files (preferably tarred or zipped)?

After making the 'template' changes you suggested, the compiling FINALLY succeeded, and I was able to build a Mac bundle.

According to a bug report in the clang mailing list, my code may be at fault here. I am using a construction that is not completely correct but works under GCC and MSVC. GCC also accepts the correct syntax, but I will try it with MSVC before making the change (you never know when doing things right may get MSVC out of whack).

[aantono]

Here are the headers...

[cirdan]

Thank you. So there is something about your zlib installation: you said that you installed zlib 1.2.8, but those headers are from zlib 1.2.5, which lacks ZLIB_CONST support. This likely means that your system came with 1.2.5 in the usual place (/usr) and you installed 1.2.8 somewhere else (/usr/local, perhaps?). Now the question is which one your compilation setup picks up, and my bet is on the preinstalled library, given that you had to use a const_cast. You should try to find out where your zlib 1.2.8 installation is and point the configure script to it.

[aantono]

That is a good question.

The output of ./configure only shows that

Code: Select all

checking zlib... found

, but does not indicate the actual version or the location.

I'll try to reconfigure with

Code: Select all

CPPFLAGS="-I/usr/local/opt/zlib/include"

to see if that works.

[Eddi]

try the file "config.log", it may contain additional details, e.g.

Code: Select all

detecting zlib
  trying /usr/include/zlib.h... found
checking zlib... found

[aantono]

Yep, that's where it is reading it.

And for some reason adding

Code: Select all

CPPFLAGS="-I/usr/local/opt/zlib/include"

didn't make it use the new location of zlib either
Any suggestions on what is the right switch to tell ./configure to use zlib from a different location?

[Eddi]

something like "--with-zlib=location", but don't take my word for it

also, "CPPFLAGS" is probably not a thing. "CFLAGS" or "CXXFLAGS", plus "LDFLAGS"

[cirdan]

Any suggestions on what is the right switch to tell ./configure to use zlib from a different location?

There is none, actually.

You can say

Code: Select all

--with-zlib=/path/to/your/zlib

but this will just check that a file by the name provided exists, and ignore it otherwise (try "--with-zlib=/bin/cat" and you will get exactly the same generated makefiles). It only makes a difference if you are doing a static build (not recommended), in which case the argument selects the location of the static libz.a. But there is no way whatsoever to point the configure script to an alternate directory for the include files, which is precisely what you need to get rid of your compilation error.

Some day I will autotoolise the package and stop all this madness.

[te_lanus]

Bug report ahead

This morning I tried to construct a "path" signal in front of a depot, and it crashed.

The thing that triggered the error is that there was a train entering the depot at the time. Tested with 1.5.0-RC1 and doesn't happen in main trunk. Tested it few times, and everytime it happened

Crash reason:
Exception: E1212012
Location: 7C812AEB
Message: Assertion failed at line 2730 of C:/MinGW/msys/1.0/home/Adminguy/fttd/src/train_cmd.cpp: v->trackdir != TRACKDIR_DEPOT

[cirdan]

Another update.

• Fixed a crash when a road vehicle reached a dead end through a one-way road (reported by romazoon).
• Fixed several slope- and bridge-related bugs in the road pathfinder.
• Minor code changes to make compilation easier with clang in c++11 mode.

aantono:
From the diff you sent me, you should no longer need to patch gfx.cpp, map/tileset.h, misc_gui.cpp or table/newgrf_debug_data.h; the other changes are still required. Please let me know if there is anything else broken.

Bug report ahead

Thanks for your report; I will look into it.


EDIT: I have (hopefully) fixed the serious bug I found yesterday. The game still suffers from the bug reported by te_lanus, which I have already identified. As a workaround until I fix it, try not to modify rail track or signals near a train that is partially in a depot.

[wallyweb]

Another update.
...
EDIT: I have (hopefully) fixed the serious bug I found yesterday. The game still suffers from the bug reported by te_lanus, which I have already identified. As a workaround until I fix it, try not to modify rail track or signals near a train that is partially in a depot.

and the Windows binary (unguaranteed):

[TrueSatan]

And here is the Ubuntu version as usual a bit l8er, because
i thought there will be another update of the source
Replaced with a newer version, thanks Cirdan

[cirdan]

i thought there will be another update of the source

And there is, now. The crash when modifying signals near a train that is partially in a depot, reported by te_lanus, should be gone, and there are several cherry-picks from openttd, but no other user-visible changes since last update.

[TrueSatan]

As u can see Cirdan i am as fast as an Vacuum Tube
Already compiled the newest version.

Thanks and hope to see more.

btw ever thought about 45° tunnels and bridges?
I don't need those Roller Coaster style, but sometimes 45°
like the tracks would be nice.

[romazoon]

Thanks for the fixes

+1 for diagonal tunnels...but i want diagonal bridge too then

personnaly, more on the path of bridgehead, i was wondering if it s technically possible to have a road crossing rail or vice versa on the bridgehead tile ? and if that's a feature you planned/or wish to add ?
and what about starting a new bridge from the bridgehead tile ? or even station tiles on it ? technically doable with your implementation of CBH or am i just a dreamer ?

[cirdan]

btw ever thought about 45° tunnels and bridges?
I don't need those Roller Coaster style, but sometimes 45°
like the tracks would be nice.

Yes, it would be nice, but also a nightmare code-wise. I guess tunnels would be slightly easier, because they essentially do not exist anywhere in the code (just a couple of endpoints and vehicles simply drive through the mountain otherwise), but bridges do have some state. Overall, too many assumptions in the code would break, and it would take a lot of work to change them.

personnaly, more on the path of bridgehead, i was wondering if it s technically possible to have a road crossing rail or vice versa on the bridgehead tile ? and if that's a feature you planned/or wish to add ?

Hmm, never thought about it. It is technically possible, of course (as in Turing-computable); the question is how much code we would have to write. I will think about it.

[TrueSatan]

To bad ga7336630 crashes on my savegame like mad after a few days ingame.
and even worse this time i forget to copy the older version. hope i have the xz somewhere around lol

added the crashlos, since you don't like my savegames from this computer i dunno if you can read
something useful.
it crashes march 8th.

going back now to an earlier version.
Edit : Found g7255c8fe and savegame works there

cheers

[TrueSatan]

btw ever thought about 45° tunnels and bridges?
I don't need those Roller Coaster style, but sometimes 45°
like the tracks would be nice.

Yes, it would be nice, but also a nightmare code-wise. I guess tunnels would be slightly easier, because they essentially do not exist anywhere in the code (just a couple of endpoints and vehicles simply drive through the mountain otherwise), but bridges do have some state. Overall, too many assumptions in the code would break, and it would take a lot of work to change them.

Oh tunnels would be a start and nice to have, guess it would be easier graphic wise as well.
So if you ever have time you could maybe try something out then

thanks and hope to get the new bug fixed soon.

[romazoon]

thanks for the reply