Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Tormy]

The point is to reach a balance that makes sense in whatever world DF is set on.

So, if a 4x4 farm plot can feed 100 dwarves, and each of these farm sprawls have 100x100 farm plots, then each sprawl can feed 6,250 people... so, we're talking what, one to two farm sprawls per city? Sounds about right to me!

[I jest, I jest.]

2-3 / city and 1 / village should be enough.

[FleshForge]

The radius they're allowed to spread from a given town ought to be limited pretty sharply too, rather than this 15-tile type stuff that is going on now.  Also they shouldn't be allowed to be built ON TOP OF rivers, especially major rivers.

[Raufgar]

The radius they're allowed to spread from a given town ought to be limited pretty sharply too, rather than this 15-tile type stuff that is going on now.  Also they shouldn't be allowed to be built ON TOP OF rivers, especially major rivers.

Well, if you want simple irrigation, river's the way to go I'm afraid, most major civilizations will always expand around fresh water sources, for food production and easy transportation. However, in such cases, savagery should play a part, since intelligent life are not the only ones using fresh water.

[FleshForge]

around fresh water sources

Yes but not on top of them!

[faratheking]

I got the newest version for MAC (intel, with osx10.5), but it doesn't work 

It shows de following msg in the Terminal panel

dyld: Library not loaded: /usr/local/lib/libstdc++.6.dylib
  Referenced from: /Users/.../Desktop/df_osx31_13/./dwarfort.exe
  Reason: image not found
/Users/.../Desktop/df_osx31_13/df: line 5: 15329 Trace/BPT trap          ./dwarfort.exe
logout

Anyone tried this one on MAC?

[ledgekindred]

dyld: Library not loaded: /usr/local/lib/libstdc++.6.dylib

same thing here. Toady apparently used an independently built gcc rather than the one that is part of the Mac dev tools.  It's looking for libstc++ library from its install directory. otool shows it will look for:

/usr/local/lib/libstdc++.6.dylib
/usr/local/lib/libgcc_s.1.dylib

from the gcc build location.  Easily solvable by putting the dylibs in with the archive and adjusting the search path but I wonder why Toady used a custom-built gcc in the first place.

[foop]

same thing here. Toady apparently used an independently built gcc rather than the one that is part of the Mac dev tools.  It's looking for libstc++ library from its install directory.

It's definitely a different version from the one in the dev tools (or, at least, the last set of Leopard dev tools) because linking to those libs gives symbol mismatches.  Which I rather expected, but I tried linking anyway because I'm hopelessly optimistic.

"dyld: Symbol not found: __ZNSt8__detail12__prime_listE"

[Totaku]

I got the same error on my mac as well. And I use a Intel OSX Tiger (10.4).

dyld: Library not loaded: /usr/local/lib/libstdc++.6.dylib
  Referenced from: /Users/.../Desktop/df_osx.0.31.13/./dwarfort.exe
  Reason: image not found
/Users/.../Desktop/df_osx.0.31.13/df: line 5:  2319 Trace/BPT trap          ./dwarfort.exe.

This looks like another one for the bug report list.

[Baughn]

Because the OS X version of GCC for Leopard is from the dark ages, and is not going to be further updated. I didn't consider that this would happen, though. It'll be fixed soon.

If you want to fix it sooner, you can run the following script, which will take care of things (slowly):

Spoiler (click to show/hide)

#!/bin/bash -x

set -e

cd
mkdir -p gcc/build
cd gcc
curl http://gcc.parentinginformed.com/releases/gcc-4.5.1/gcc-4.5.1.tar.bz2 | tar xj
cd gcc-4.5.1
curl http://gcc.parentinginformed.com/infrastructure/gmp-4.3.2.tar.bz2 | tar xj
mv gmp-4.3.2 gmp
curl http://gcc.parentinginformed.com/infrastructure/mpc-0.8.1.tar.gz | tar xz
mv mpc-0.8.1 mpc
curl http://gcc.parentinginformed.com/infrastructure/mpfr-2.4.2.tar.bz2 | tar xj
mv mpfr-2.4.2 mpfr

cd ../build
../gcc-4.5.1/configure
make -j4
sudo make install

[devek]

Would it be so hard to use clang instead of GCC?

[Baughn]

Clang doesn't support C++ well, and doesn't support C++1x at all.

[devek]

Without derailing the thread I should mention that Clang supports 4 NEW 1x features and MSVC10 supports 6 NEW 1x features. Of course, you can't ignore existing features when comparing total features.

Dwarf Foreman won't compile on MSVC because I use variable length arrays... an 11 year old feature? http://msdn.microsoft.com/en-us/library/zb1574zs.aspx lol

I will also bet you $100 that clang will have full 1x support before MSVC.

[Baughn]

That is not what http://clang.llvm.org/cxx_status.html suggests, but I suppose I can give it a try. Using clang would definitely be an interesting experiment.

[ledgekindred]

clang is getting a significant update with Xcode 4, especially in terms of c++ support, but unless you have an apple developer subscription you don't get access to it yet.  and I really can't speak at all to the relationship between what apple is shipping and what llvm has.  I'd still be _really really_ surprised if clang compiled DF, or any other even moderately complex c++ app that wasn't written from the start to build against clang, without significant issues. 

If it did, then I'd expect df to get a really nice performance boost because llvm has proved it has way better code generation/optimization than gcc outside of some really weird corner cases.  But I'm not gonna beg for clang compilation any time soon. 

[Baughn]

It's worth a shot. I've successfully compiled libgraphics with clang on linux before, which should be somewhat representative of Toady's coding style.

Problem is, Toady doesn't have access to Xcode 4. He won't, either, unless someone buys him a copy of snow leopard. Which would seem silly if it turns out that that clang version does not in fact compile it.

Not that it matters. It's perfectly possible to build it yourself.