These will be changed to conditionally include system Lua headers,
e.g. "lua.h", instead of submodule Lua headers, e.g. "module/lua/lua.h".
If a header named "lua.h" includes "lua.h", the build will fail due to
recursion.
This can't be solved using angle brackets to include system headers,
because macos builds won't find them:
In file included from /Users/runner/work/wesnoth/wesnoth/src/ai/registry.cpp:30:
In file included from /Users/runner/work/wesnoth/wesnoth/src/ai/composite/aspect.hpp:24:
In file included from /Users/runner/work/wesnoth/wesnoth/src/ai/lua/lua_object.hpp:25:
/Users/runner/work/wesnoth/wesnoth/src/lua/lua.h:4:14: error: 'lua.h' file not found with <angled> include; use "quotes" instead
#include <lua.h>
^~~~~~~
"lua.h"
Renamed with (requires GNU sed):
$ for f in src/lua/*.h; do
> git mv "${f}" "src/lua/wrapper_${f#src/lua/}";
> done
$ git grep -El -- '#[ \t]*include[ \t]+"lua/[^"]+[.]h"' src | \
> xargs sed -Ei -- '
> s|(#[ \t]*include[ \t]+"lua/)(lua[.]h")( )?|\1wrapper_\2|;
> s|(#[ \t]*include[ \t]+"lua/)(lualib[.]h")( )?|\1wrapper_\2|;
> s|(#[ \t]*include[ \t]+"lua/)(lauxlib[.]h")( )?|\1wrapper_\2|;
> '
This additionally:
* Makes all copyright notices identical aside from the starting year for Wesnoth-specific source files. Files not included: mariadbpp, lua, spirit po, xbrz, and bcrypt (crypt_blowfish).
* Removes all attribution from the files, since the vast majority of them are outdated or seemingly just outright incorrect. For example, I would guess that Dave is no longer the sole author of the majority of Wesnoth's current code.
The change to static_cast for the definition of LUAL_BUFFERSIZE replaces the fix previously used (d0100758f8) for Lua 5.3. 5.4 removes the static alternative for LUAL_BUFFERSIZE. A better solution would probably be to disable the old-style-cast warning for luaconf.h, but I can't figure out how to do that so using static_cast is the easiest solution. Do note that change will have to be applied each Lua update like the aforementioned commit.
Apparently, even on Boost 1.73 without Boost placeholders specialized as is_placeholder, they were somehow
getting used in the global namespace. Would explain all the "Boost placeholders in the global namespace is
deprecated" warnings I was getting after 23d1db043f.
When testing with BOOST_BIND_NO_PLACEHOLDERS, even though I had `using namespace std::placeholders` in utils/functional.hpp,
compilation still failed in places. This confirms even more that Boost global placeholders were being used. Honestly,
it was simplest just to specify std::placeholders for everything. This also means we can remove the hack in utils/functional.hpp
designed to allow Boost placeholders to work with `std::bind`.
This was needed to get the build working with vcpkg's version of SDL, where all the SDL files are
in their own SDL folder. However, our cmake config also has a note saying it was a deliberate choice
to move our SDL files *out* of their SDL2 folder due to certain distros (FreeBSD is mentioned) not
putting the files in said folder in the first place.
This could otherwise be used to escape the lua sandbox, as described in
multiple sources. For example one can use it to reenable the os.execute
function to do shell commands
The affected functions were
load,loadstring,wesnoth.dofile,wesnoth.require and various places in the
wesnoth source where lua chunks were loaded for example by the ai code.
This commit also changes the lua source to change luas load (which is
the same as loadstring), alternatively we could add a wrapper around the
original load function that always passes "t" as third parameter, i went
this way mostly because it was easier to implement, but also because i
was not 100% sure that is is impossible to query the upvalues of a
function via lua (wesnoth disables debug.getupvalue but still).
There is also an occurance in the application_lua_kernel that was not fixed
because i assumed that umc cannot contian application lua scipts.
As further security measure we might want to disable printing the function
adress in luas tostring for c functions, this cannot be exploited by itself
but it can be used to defeat ASLR in some cases.
(cherry-picked from commit 2554c166dd)
* Dropped unused CVideo class member references.
* Replaced the lone usecase of the CVideo member in loadgame with the singleton and removed said member.
* Removed CVideo references from a bunch of addon management functions.
* Cleaned up a *lot* of now-unnecessary forward CVideo declarations.
Covers more cases of:
* push_back(std::pair
* push_back(std::make_pair
* push_back(std::make_tuple
* push_back(T) where T is an empty object of type T
Small thing I noticed: this does mean the numbers in font::subset_descriptor are no longer 'cast' to size_t before being
added to the vector, but that shouldn't matter (hopefully).
Instead of passing the video object to the Lua kernel, the game now simply
fetches the video singleton when displaying dialogs. This means there is
no longer any need to store a reference to the video object.
luaconf.h provides a place for us to make changes, avoiding the need to change the original definitions. Move everything down there. This encompases the following changes:
1) Disable compatibility with old versions of Lua in the C++. Compatability is maintained only for the Lua runtime. Only one correction was needed: in application_lua_kernel.cpp
2) Change how the backpointer is defined, for forward compatability with Lua 5.3. This effected only one line: in lua_kernel_base.cpp. Using the Lua 5.3 macro caused a GCC warning, suppressed it for that line.
3) Certain Windows-only features are no longer available in the Lua runtime. These features are all in the Lua io module, which we don't allow access to, so this is a non-change for the runtime.
4) Lua will behave as if it is a standard C environment. This, again, mainly effects the Windows environment and features we don't allow access to in the runtime.
Do not change the released Lua source code. Lua is written as standard C, but we compile it as C++ so we can use exceptions. Change the call sites to avoid the warnings about using a C cast instead of static_cast<int>().
NB: The changes to Wesnoth C++ code will be required to upgrade to Lua 5.3, anyway.
This reverts:
commit cee3ab2084
Author: Mark de Wever <koraq@xs4all.nl>
Date: Sun Feb 5 19:55:32 2012 +0000
Fix compiler warnings.
This is the first step towards creating artificial load and testing if the
GUI2 lobby still performs well under load.
I also implemented ability to leave the game in the GUI2 MP staging dialog.
Previously the server continued to believe that the player remains in the
game when he/she closed the MP staging dialog.
@gfgtdf pointed out that preferences contain MP passwords. We can't allow
untrusted code to access them. Since we only need preference access from
plugins for now, the simplest way to avoid security issues is to block Lua
code outside of plugins from accessing preferences at all.
This constitutes drop-in replacements for:
* boost::shared_ptr
* boost::scoped_ptr
* boost::weak_ptr
* boost::enable_shared_from_this
* boost::static_pointer_cast
* boost::dynamic_pointer_cast
This excludes boost::intrusive_ptr, except for stray includes. Refactoring that is more complicated.
This commit converts the following function calls:
* boost::bind -> std::bind
* boost::function and boost::functionN -> std::function
* boost::ref and boost::cref -> std::ref and std::cref
* boost::bad_function_call -> std::bad_function_call
In the process, it was discovered that std::bind has trouble with overloaded
functions. There were two such cases in the code:
* gui2::twindow had an ancient unused overload to draw(). The overload was removed.
* gui2::trepeating_button was binding tdispatcher::fire. This case was converted
to a lambda.
headers from lua/.. contain macros that can break other headers in
wesnoth, boost or other libraries.
In this case it was a macro
#define cast(t, exp) ((t)(exp))
defined in lua/llimits.h that broke a boost header.
This key was created in the first attempt at application scripting
but it was reappropriated to be used as the key for the
"thread table" when the present solution based on coroutines was
found. The key has been renamed and moved to its site of usage now.
The problem was that intf_require would assume use the wrong
lua_State* since it finds it using the kernel's this pointer. This
commit gives a version of "protected_call" which allows to use the
callback's lua_State*.
