- sha265.c is using some C99 specific features, in particular static
array dimensions in a function parameter array (see section 6.7.5-7
of the C99 spec). A #ifndef check was in place to prevent compilation
under MSVC but it actually needed to check for C99 compliance instead.
Even still, the C99 code fails on a few compilers (PGI being one) so for
compatibility reasons, the C99 version of the function declaration is
removed entirely, leaving only the C89 version.
- CHECK_SYMBOL_EXISTS is used to determine the presense of bswap
functions from byteswap.h. Most compilers re-dedefine the bswap_N
functions as a __bswap_N function implemented by the compiler. Since
bswap_N is usually defined as a macro then it's mere presence passes
the check. Some versions of the PGI compiler though have shipped
broken headers for byteswap.h, in particular 11.3 for x64 linux
provides byteswap.h but is missing an associated bits/byteswap.h which
causes some of the bswap_N macros to be defined but broken and unusable.
The bswap_N checks have been converted to CHECK_SOURCE_COMPILES to
ensure that the bswap_N calls are actually usable and not just merely
defined.
Any "#define restrict ..." line may conflict with MSVC header files
that use "__declspec(restrict)". Define our own LZMA_RESTRICT macro
in "Utilities/cmliblzma/config.h.in" and transform liblzma code to
use it:
git grep -lE '\<restrict\>' Utilities/cmliblzma/liblzma |
xargs sed -i 's/\<restrict\>/LZMA_RESTRICT/g'
Explicitly invoke python script through the interpreter since
windows does not act on hashbangs.
Use the found qcollectiongenerator executable rather than what
happens to be in PATH.
Add identifiers for variables, properties, policies and modules.
This will allow QtCreator to show relevant documentation if it learns
more about the context of the contents of cmake files.
The error is caused by a workaround for UNICOS. The workaround
incorrectly uses the _CRAYC macro to determine if building for
UNICOS. This macro will always be defined for the Cray C and
C++ compiler regardless of the target platform, even when
building for Linux. The correct macro should be _CRAY, which as
per Cray documentation is only defined for UNICOS targeted
builds.
Newer HP-UX versions (e.g. 11.31) also offers F_SETTIMES, but with a
different struct for the timing information.
Suggested-by: Eric Berge <ericmberge@gmail.com>
Newer HP-UX versions (e.g. 11.31) also offers F_SETTIMES, but with a different
struct for the timing information.
Suggested-by: Eric Berge <ericmberge@gmail.com>
Teach set_property and get_property an "INSTALL" property type to be
associated with install-tree file paths. Make the properties available
to CPack for use during packaging. Add a "prop_inst" Sphinx domain
object type for documentation of such properties.
Use the CPack DragNDrop generator instead of the deprecated PackageMaker
tool to package CMake itself. This provides an installation experience
that is more consistent with other products on OS X and allows users to
select the destination directory easily. It also avoids installing
"/private/var/db/receipts/com.Kitware.CMake.*" receipts that must be
removed by "pkgutil --forget com.Kitware.CMake" before another version
of CMake can be installed.
The DragNDrop installer does not support a post-flight script, so drop
our configuration of it. The cmake-gui has an option for installing
symbolic links to enable command-line use. In practice users may simply
add "/Applications/CMake.app/Contents/bin" to their PATH instead.
In the domain clear_doc method, avoid removing entries from a dictionary
while iterating over it. Instead accumulate a set of entries to remove
at the end.
67f5f0a9 libarchive: Use _snprintf on Windows, not snprintf
fcfbb0a9 libarchive: Drop LIBARCHIVE_ADDITIONAL_LIBS, CMake does not need it
61a649d9 libarchive: Update README-CMake.txt for new snapshot
4533560c Merge branch 'libarchive-upstream' into update-libarchive
44d6b82f libarchive: Disable all whitespace checks in third-party code
37f225b7 libarchive 3.1.2-246-ga5a5d28b (reduced)
6ab7c326 libarchive: Avoid left-shift overflow of signed integer
In libarchive/archive_write_set_format_zip.c there are two calls to
archive_le32enc whose second argument is of the form
archive_entry_mode(zip->entry) << 16
However, the return type from archive_entry_mode may be a signed integer
so the shift may overflow. Since the second argument of archive_le32enc
expects uint32_t anyway, simply cast to that prior to shifting.
5a58efaa libarchive: Avoid using name 'u_long'
e2b02823 Update libarchive configuration within CMake
80883321 libarchive: Do not require includers to have windows.h
dfb0458e libarchive: Convert literal LL suffix to ARCHIVE_LITERAL_LL
b0a9807f libarchive: Update archive_util.c to use CMake zlib and bzip2 headers
debe4dec libarchive: Drop options not present in reduced version
66b0c4fa libarchive: Do not generate a pkg-config file
8092e759 libarchive: Update README-CMake.txt for new snapshot
2f197863 Merge branch 'libarchive-upstream' into update-libarchive
23e4666c libarchive: Disable more whitespace checks in third-party code
64713ae3 libarchive 3.1.2-218-g00f4bd83 (reduced)
In archive_entry.h refer to "BY_HANDLE_FILE_INFORMATION *" using
"struct _BY_HANDLE_FILE_INFORMATION *" to avoid requiring the
includer to already have <windows.h>.
Drop ENABLE_TAR and ENABLE_CPIO related options because we do not build
these command-line tools. Drop ENABLE_TEST and ENABLE_COVERAGE options
because we do not build the tests.
Resolve conflicts in favor of the upstream side where possible.
Resolve a logical conflict in archive_windows.h where the upstream
port to Watcom was done slightly differently from ours.
Assistant in Qt 4.8 does not handle css import paths relative to
the includer. This is fixed in Qt 4.8 commit b95750a275 (Assistant: Set
the url on created QNetworkReply objects., 2014-03-31). It is unknown
whether there will be a further Qt 4.8 release containing that commit.
Use a CMake script to pre-replace the content prior to generating the
qch file. An alternative workaround of moving the files or adding
"_static" to the import path did not seem to work for existing Qt 4.8
versions.
The bug was fixed in the Qt 5 branch before Qt 5.0. The Qt 5 assistant
renders this workaround'ed version correctly too.
In some cases, it was possible for the include directory of the system-wide
libcurl to be added to the include path before cmcurl's, which would result
in them being picked up and causing the build to fail if the curl versions
differ too much.
One way to trigger this is to have OpenSSL installed into a non-default
location together with libcurl (/usr/local, for example). If cmcurl is built
with CMAKE_USE_OPENSSL on, -I/usr/local/include would end up being added
before -I${PATH_TO_CMCURL}.
Since commit 71b14dcb (Utilities/Release: Do not upload doc staging
tarball, 2014-02-26) the prefix upload_release.cmake computes does not
match any files when used with -DVERSION=master as has been done for the
nightly binary builds. Since the version is not actually 'master'
anyway, change the nightly binary upload logic to explicitly pass the
destination directory. Do not pass any VERSION so the default is taken
and matches the binaries.
Name the pre-built docs tarball on the remote machine according to the
release script name so that multiple tarballs going to a single remote
machine do not clobber one another.
Add SPHINX_TEXT to enable the Sphinx 'text' builder. Mark it as
advanced and do not add install rules. This is intended for use
by the release manager to build the release notes in text format
suitable for email.
Avoid requiring all build machines for the upstream packaging process to
have Python and Sphinx installed. Instead create a way to build the
documentation once on the host machine and copy it to each build machine
as a tarball with content to include in the installation tree for
packaging.
Port upstream commit 6cf33c93 (Issue 320: Rewrite (again) to avoid
the left shift that CLang dislikes so much, 2013-12-07) into CMake.
Inspired-by: Tim Kientzle <kientzle@freebsd.org>
Port upstream commit 533e8fda (Rework the sign-extension to avoid
left-shift of an explicit negative number, 2013-06-29) into CMake.
Inspired-by: Tim Kientzle <kientzle@freebsd.org>
Download http://www.gnu.org/licenses/old-licenses/lgpl-2.1.txt and place
it as Licenses/LGPLv2.1.txt in our source tree. When building cmake-gui,
use option CMake_GUI_DISTRIBUTE_WITH_Qt_LGPL to enable notification
in the "About" dialog of how the distribution of Qt is licensed.
Install the license file as ${CMAKE_ROOT}/Licenses/LGPLv2.1.txt so that
the dialog can display a path to it.
Since commit 7d47c693 (Drop compatibility with CMake < 2.4, 2013-10-08)
we no longer need to use the configure_file IMMEDIATE option to support
compatibility modes less than 2.0.
Configure our Sphinx conf.py with a copyright line extracted from
Copyright.txt instead of using the year in which the documentation is
built. This will future-proof the reported copyright year range when
building documentation for old versions.
Add the man page description line as explicit markup at the top of each
Help/manual/*.rst file and scan it from conf.py to automatically
generate the man_pages Sphinx configuration value. This reduces the
number of places that need to be changed when a new manual is added.
Include the CTestUseLaunchers module in the standalone build of
Utilities/Sphinx so that it can be built under CTest with the
CTEST_USE_LAUNCHERS option.
When scanning CMake module files for .rst comments, recognize
bracket comments starting in ".rst:" too. For example:
#[[.rst:
Include the bracket comment content terminated by the closing bracket.
Exclude the line containing the bracket if it starts in "#".
Teach the CMakeLib.testRST test to cover multiple bracket lengths
and ending brackets on lines with and without "#".
Update the cmake-developer.7 manual to document the bracket-comment
syntax for .rst documentation.
Generator expressions are supported in many places and are a distinct
concept worthy of their own manual page. The old builtin documentation
was previously represented by preprocessor macros to generate it into
each place that supports them. Factor out the duplicate content into a
dedicated cmake-generator-expressions manual page and reference it from
each original location.
Add a Utilities/Sphinx directory to hold CMake build code to run the
Sphinx (sphinx-doc.org) documentation generation tool. Create a
CMakeLists.txt file there capable of building either as a subdirectory
of the main CMake build, or as a standalone documentation build.
Add cache options SPHINX_MAN and SPHINX_HTML to select output formats
and SPHINX_EXECUTABLE to specify the sphinx-build executable. Add
bootstrap options --sphix-man and --sphinx-html to select output formats
and --sphinx-build=<sb> to specify the sphinx-build executable.
Create a "conf.py.in" file to configure_file into "conf.py" to tell
sphinx-build how to build our documents. Create a "cmake.py" Sphinx
extension module defining:
* The "cmake-module" directive used in Help/module/*.rst files to
scan .rst markup from the corresponding Modules/*.cmake file.
* A Sphinx domain called "cmake" defining documentation object types
for CMake Help/<type> directories: command, generator, manual,
module, policy, prop_*, and variable. Add a "role" for each type
to perform cross-references. Teach the roles to treat "<XYZ>"
as placeholders instead of explicit targets if not preceded by
a space. Add cmake domain directives to define command and
variable objects explicitly in .rst file content. This will
allow modules to define their own commands and variables and
have them indexed and linkable.
* A Sphinx document transform that converts Help/<type>/*.rst documents
into cmake domain objects of the corresponding <type> and adds index
entries for them. This will automatically index all CMake documentation
objects and provide cross-reference targets for them with no special
markup in the .rst files.
Drop the 'documentation' build target. We will no longer use the
executables to generate their own documentation. New infrastructure
will be introduced later to generate documentation.
The final location and name of a build-target is not determined
until generate-time. However, reading the LOCATION property from
a target is currently allowed at configure time. Apart from creating
possibly-erroneous results, this has an impact on the implementation
of cmake itself, and prevents some major cleanups from being made.
Disallow reading LOCATION from build-targets with a policy. Port some
existing uses of it in CMake itself to use the TARGET_FILE generator
expression.
Chuck Atkins
Brad King
Daniel Pfeifer
Nils Gladitz
Stephen Kelly
Rolf Eike Beer
Uwe L. Korn
Mike Crowe
Raphael Kubo da Costa
Jiri Malak
Ruslan Baratov
Alex Neundorf
Daniele E. Domenichelli