- Confirmed that libexpat is built statically and linked with xlsxreader
- May need da7cda3a85 for Windows
- Although libexpat's CMake options are all prefixed with LIBEXPAT_, it also adds some cache entries like SIZE_T (from expat/ConfigureChecks.cmake). Unsure if these affect other libs.
- xlsxio may need additional reconfiguration after moving to add_subdirectory() to find libexpat/libzip on non-Linux platforms.
The dependencies introduced in #1620 cause an error when building natively on Windows, according to @ragundo. CMake is looking for build tools `v100`, which are part of VS2010 and won't work with DF/DFHack, even if they were installed.
Some of the Windows build scripts pass in `-T v140_xp` (from VS2015), so passing this option through to subprojects may help: https://cmake.org/cmake/help/latest/variable/CMAKE_GENERATOR_TOOLSET.html
although they declare that they only need cmake 3.0.2, they started
using add_compile_definitions(), which was added in cmake 3.12
git checkout 29b881d fixes this
so it doesn't break us with incompatibile changes. we only need to
update this lib if/when we run into a bug with it. no need to introduce
instability by tracking HEAD
Fixes tsan trace report between lua viewscreen and other threads running
lua without CoreSuspender lock. But I would assume similar races exists
when using lua from console thread, remote thread and vmethods same time.
This makes jsoncpp a submodule that can be build directly from git
sources. This changes depends/jsoncpp to depends/jsoncpp-sub to avoid
filename conflict if someone tries to use git bisect.
jsoncpp library name changes to jsoncpp_lib_static.
jsoncpp version is the latest tagged release.
Passing a 64bit integer to Json::Value fails to find matching. Even
tough long long int is also 64bits gcc considres long long int different
to long int. But fix can't be as simple as removing an long from type
because that would reduce 32bit builds to have only 32bit maximum int
size. But standard offers fixed width integer types that can be used to
get correct underlying type depending on platform.
All platforms seems to either have compiler supporting c++14. Windows
already requires msvc2015 that supports c++14 which should make it
minimal issue to require c++14 support from all compilers because gcc
is pretty easy to upgrade.
Upstream has moved onto tinyxml2, so this is likely to be the last
release of the original tinyxml (2.6.2 came out in 2011). dfhack
should therefore probably ship 2.6.2, since it's unlikely there will
ever be future fixes to this code. The upstream changelog (taken
from "changes.txt" in upstream tarball) contains many bugfixes and
is included in the commit message below.
2.5.4
- A TiXMLDocument can't be a sub-node. Block this from happening in
the 'replace'. Thanks Noam.
- [ 1714831 ] TiXmlBase::location is not copied by copy-ctors, fix
reported and suggested by Nicola Civran.
- Fixed possible memory overrun in the comment reading code - thanks
gcarlton77
2.5.5
- Alex van der Wal spotted incorrect types (lf) being used in print
and scan. robertnestor pointed out some problems with the simple
solution. Types updated.
- Johannes Hillert pointed out some bug typos.
- Christian Mueller identified inconsistent error handling with
Attributes.
- olivier barthelemy also reported a problem with double truncation,
also related to the %lf issue.
- zaelsius came up with a great (and simple) suggestion to fix
QueryValueAttribute truncating strings.
- added some null pointer checks suggested by hansenk
- Sami V�is�nen found a (rare) buffer overrun that could occur in
parsing.
- vi tri filed a bug that led to a refactoring of the attribute
setting mess (as well as adding a missing SetDoubleAttribute() )
- removed TIXML_ERROR_OUT_OF_MEMORY. TinyXML does not systematically
address OOO, and the notion it does is misleading.
- vanneto, keithmarshall, others all reported the warning from
IsWhiteSpace() usage. Cleaned this up - many thanks to everyone
who reported this one.
- tibur found a bug in end tag parsing
2.6.2
- Switched over to VC 2010
- Fixed up all the build issues arising from that. (Lots of latent
build problems.)
- Removed the old, now unmaintained and likely not working, build
files.
- Fixed some static analysis issues reported by orbitcowboy from
cppcheck.
- Bayard 95 sent in analysis from a different analyzer - fixes
applied from that as well.
- Tim Kosse sent a patch fixing an infinite loop.
- Ma Anguo identified a doc issue.
- Eddie Cohen identified a missing qualifier resulting in a compilation
error on some systems.
- Fixed a line ending bug. (What year is this? Can we all agree on a
format for text files? Please? ...oh well.)
Deprecated in Lua 5.3, but still contains some useful things, like extract()
Also helps maintain backward-compatibility with scripts that target older DF
versions (e.g. Mifki's dfremote project)
As best as I can tell, the copy of tinyxml dfhack uses is unmodified
from whenever it was first bundled. This commit adds an option to CMake,
EXTERNAL_TINYXML, that if set to ON, will attempt to link against a system
tinyxml instead of using the dfhack-bundled one. It defaults to OFF,
so there is no change in default behavior.
The DFHACK_TINYXML variable is then set to either "tinyxml" or "dfhack-tinyxml"
so the library (and any plugins that need updating) can link against one
or the other.
The FindTinyXML.cmake script was taken from https://github.com/ros/cmake_modules
(licensed under the 3-clause BSD license).
Add license text to new CMake file.
1. update to tinythread version 1.1, which provides a detach method
2. move the thread disassociation to the join function, which allows
attached threads to be joined even after they finish (this was the main
leak)
Also update RemoteServer.cpp to detach threads and delete the thread
objects instead of leaking them (although they are much smaller than the
actual threads).