Unfortunately, we can't force the specific inclusion of a header file.
So if anything add /opt/local/include to the include path--such as
libintl--then other dependencies might be drawn from /opt/local at
compile time, even though we detected them elsewhere at configure time.
This, in turn, causes issues with mixed versions, such as the iconv.h
header being pulled from /opt/local/include, but linked against the
library in /usr/lib--which can be mismatched versions.
So, despite CMake's best effort to treat /sw and /opt/local as just
another system area, we really need to give them preferential treatment.
To do this, we add them to CMAKE_PREFIX_PATH.
This fixes an issue discovered while re-enabling iconv in #1370.
The -O3 optimization level can often lead to dangerous (and sometimes
incorrect) optimizations being performed. So let's use a level that's
more stable.
If you aren't just building everything into build/, then the functional
tests fail because they can't find the nvim executable. Let's pass in
the location of the nvim executable, and set NVIM_PRG environment
variable accordingly.
Remove build warning:
The target name "test" is reserved or not valid for certain CMake
features, such as generator expressions, and may result in undefined
behavior.
Now that the lua client is available, python/lupa are no longer necessary to run
the functional tests. The helper functions previously defined in
run-functional-tests.py were adapted to test/functional/helpers.lua.
- cmake: git_timestamp() returns last commit time formatted as
`YYYYMMddHHmm`.
- Always include commit hash in :version and --version output.
`nvim --version` sample output:
NVIM 0.0.0-alpha+201410070245 (compiled Oct 7 2014 05:30:45)
Commit: f747b2b1ff7bfe7eb00cc2be82d7af87c98f1111
The 'lupa' python package provides a simple way to seamless integrate lua and
python code.
This commit replaces vroom by a python script that exposes the 'neovim' package
to a lua state, and invokes busted to run functional tests. This is a temporary
solution that will enable writing functional tests using lua/bused while a lua
client library is not available.
The reason for dropping vroom is flexibility: Lua/busted has a nice DSL-style
syntax while also providing the customization power of a full programming
language. Another reason is to use a single framework for unit/functional tests.
Two other changes were performed in this commit:
- Instead of "gcc-unittest/gcc-ia32", the travis builds for gcc are now
identified by "gcc/gcc-32". They will run unit/functional tests for both 64
and 32 bits.
- Old integration tests(in src/nvim/testdir) are now ran by the 'oldtest' target
The install() command will create the parent directories, but it does so
with the user's umask. We want to do our best to make sure the correct
permissions are being set, without clobbering existing permissions.
To do this, this commit introduces an install_helper(), which is similar
in signature to the install() command, to help ensure that directories
are created ahead of the actual install() command. This will attempt to
use 0644 permissions for files and 0755 permissions for directories by
default--though they can be overridden.
To make this work correctly, without trying to introduce some mechanism
with setting the umask, it meant that there's a small portion that makes
use of an "internal" version of the file() command. It has been tested
on CMake 2.8.11, 2.8.12, and 3.0.2, and works correctly on all versions.
This fixes#1201 and #1086.
`-Wstrict-prototypes` warn if a function is declared or defined without
specifying the argument types.
This warning disallow function prototypes with empty parameter list.
In C, a function declared with an empty parameter list accepts an
arbitrary number of arguments when being called. This is for historic
reasons; originally, C functions didn't have prototypes, as C evolved
from B, a typeless language. When prototypes were added, the original
typeless declarations were left in the language for backwards
compatibility.
Instead we should provide `void` in argument list to state
that function doesn't have arguments.
Also this warning disallow declaring type of the parameters after the
parentheses because Neovim header generator produce no declarations for
old-stlyle prototypes: it expects to find `{` after prototype.
- When USE_BUNDLED=OFF and the system has multiple lua interpreters
(luajit AND lua) it may occur that only the later has the needed
dependencies (lua-lpeg, lua-cmsgpack). If we pick luajit then
finding the dependencies FAILS.
- This commit groups detection of the lua interpreter with the lua
dependencies it tries to find them for both the interpreters and
choses the first one that has them.
Helped-by: John Szakmeister <john@szakmeister.net>
This is in preparation for the next step, which is to find a suitable
lua interpreter, rather than just erroring when an interpreter is found
but doesn't have the necessary dependencies.
Helped-by: John Szakmeister <john@szakmeister.net>
CMake purposefully disables the use of the `-isystem` flag on Apple
platforms, but it's overly blunt with the detection. Apple's compilers
have supported the flag since at least 10.4. Let's force the switch to
be on when gcc/g++ is detected on an Apple platform to reduce the
warnings out of the msgpack-related bits.
Stop forcing some platform setting that are really intended to be used
for Travis CI. Under other systems, like Arch Linux, it prevents
dependencies from being correctly located.
- The 'stripdecls.py' script replaces declarations in all headers by includes to
generated headers.
`ag '#\s*if(?!ndef NEOVIM_).*((?!#\s*endif).*\n)*#ifdef INCLUDE_GENERATED'`
was used for this.
- Add and integrate gendeclarations.lua into the build system to generate the
required includes.
- Add -Wno-unused-function
- Made a bunch of old-style definitions ANSI
This adds a requirement: all type and structure definitions must be present
before INCLUDE_GENERATED_DECLARATIONS-protected include.
Warning: mch_expandpath (path.h.generated.h) was moved manually. So far it is
the only exception.
It inteferes with development activities by breaking your build in the
middle of a refactor. Instead, let's enable -Werror on the Travis CI
builds via a TRAVIS_CI_BUILD option.
- Leave src as include dir (for includes to recognize 'nvim/' prefix).
- Change subdirectory from src to src/nvim.
- Fix msgpack generation.
- Fix some other paths to new locations.
This adds the `SYSTEM` parameter to `include_directories`, which will tell cmake
to use `-isystem` instead of `-I` for specifying include directories. One
advantage is that compilers won't emit warnings for included files that belong
to dependencies.
This allows us to avoid hard-coding paths and using environment
variables to communicate key information to unit tests, which fits
with the overall goal of making sure that folks driving CMake directly
can continue to do out-of-tree builds.
- Build targeting 32-bit with travis
- Code in `before_install`/`after_success` was moved to travis.sh since it
provides greater flexibility for detecting the build matrix environment. This
improves the build speed since we now install only what's necessary.
- Now clint has a dedicated travis worker
Apparently busted 1.11.0 is broken(https://github.com/Olivine-Labs/busted/issues/236)
in a way that is causing the unit tests to fail. This pins the version to 1.10.0
and also fixes a wrong variable set when msgpack was added as a dependency
A lua executable is now required for the build process since a lpeg-based script
is used for generating a dispatch function and metadata for the msgpack API
frontend. This removes the need for setting the LUA_BINARY environment variable.
Fixes#518.
Only provide the unittest target if busted was found. And only build
nvim-test if the unittest target exists by excluding nvim-test from all.
Note: this means nvim-test won't be built by default, but it will be
built when you try to run unittests.
Since libuv.pc is broken at the moment, try to determine libuv's
dependencies ourselves. This ports most of the checks from libuv into
our CMake build, and fixes the build on other unix platforms.
This achieves several goals:
* Less reliance on scripts so we have better portability to Windows
(though we still have a ways to go for proper Windows support).
Luajit, luarocks, moonscript, and busted are all installed via CMake
now.
* Trying to make use of pkg-config to get the correct libraries. The
latest libuv is still broken in this regard, but we'll at least be in
a position to use it.
* Allow the use of Ninja or make. The former runs faster in many
environments, and automatically makes use of parallel builds.
This also allows for system installed dependencies--though not through
the Makefile just yet--and adds support for FreeBSD.
This also make us build libuv and luajit as static libraries only, since
we're only concerned about having static libraries for our bundled
dependencies.
As described in Google's style guide, the basis for Neovim's
> All of a project's header files should be listed as descendants of the
> project's source directory without use of UNIX directory shortcuts .
> (the current directory) or .. (the parent directory).
Add src as an include directory to facilitate this.
- Valgrind configuration removed
- Fix errors reported by the undefined behavior sanitizer
- Travis will now run two build steps:
- A normal build of a shared library for unit testing(in parallel with gcc)
- A clang build with some sanitizers enabled for integration testing.
After these changes travis will run much faster, while providing valgrind-like
error detection.
Because of the '$' in `if(DEFINED $ENV{VALGRIND_CHECK})` EXITFREE wasn't being
defined, so the `free_all_mem` wasn't being included or called in the resulting
binary.
This commit fixes that, and also adds includes needed for `free_all_mem`
compilation.
It seems clang 3.4 thinks the codebase is in fantastic shape and gcc 4.9.0
has only minor niggles, which I fixed:
- fix uninitialized member warning:
In DEBUG mode the expr member doesn't get properly initialized to NULL.
- fix warnings about directive inside of macro's:
On some platforms/compilers, sprintf is a macro. Putting macro directives
inside of a macro is unportable and gcc 4.9 warns about that.
- fix signed vs. unsigned comparison warning:
The in-memory table will luckily not even come close to the limits imposed
by ssize_t. If it ever reaches that, we've got bigger problems.
This prevents an error from CMake when libintl is not found. It's not a
required library, so we must wrap the inclusion in a conditional. It
was already done for the library.
If the compiler is some GNU-alike variant, set the compiler flags to use
the gnu99 dialect of C and enable all warnings.
Non-GNU compilers may have to have their own magic added to set dialect
and enable warnings.
Closes#179.
As noted in #128, if clock_gettime is provided by librt then it does not
end up being linked into the static libuv.a binary. This might be
considered a bug in libuv but we can address it here.
Detect if librt provides the clock_gettime symbol and, if so, append it
to the list of libraries linked into nvim. On non-librt systems the
behaviour should be as before.
We use the standard CMAKE_PREFIX_PATH variable to pass the location of
.deps as a search location on the command line. There is now no need for
explicitly hard-coding it.
CMake ships with a standard FindThreads module which can be used to a)
test for a threading library and b) confirm that it is pthread. It also
allows the hard-coding of the threading library name to be removed from
``src/CMakeLists.txt``.
Make it an error not to have a pthread library installed and indicate to
CMake that we strongly prefer pthread to any other platform threading
library.
- Cleanup source tree, leaving only files necessary for compilation/testing
- Process files through unifdef to remove tons of FEAT_* macros
- Process files through uncrustify to normalize source code formatting.
- Port the build system to cmake
