Teach cmSystemTools::MakefileColorEcho to use the color printing API
only when we actually have a color to print. This avoids printing
escape sequences that do not change the text attributes.
Since commit v3.3.0-rc1~480^2~4 (Makefile: Fix output during parallel
builds, 2015-02-05) we always use this code path to print progress and
rule messages even when color is disabled (e.g. in a try_compile). It
is important to avoid printing escapes when there is no color because on
MSYS we have no isatty and so assume that vt100 escapes are supported,
but we do not want to print them when capturing try_compile output.
If we capture the vt100 escapes in try_compile output, they contain
unbalanced square brackets. This causes CMakeParseImplicitLinkInfo
to fail to separate lines in a ;-list and therefore fail to extract
information from the link line. Then mixed-language linking breaks.
The Fortran compiler version detection infrastructure added by commit
v3.3.0-rc1~436^2~9 (Fortran: Add infrastructure to detect compiler
version, 2015-02-17) forgot to update CMakeFortranCompiler.cmake.in to
save the compiler version persistently as we do already in
"CMake{C,CXX}Compiler.cmake.in". Add the missing line now.
Refactoring in commit v3.3.0-rc1~76^2 (cmMakefile: Handle CMP0014 before
configuring the generator, 2015-05-14) accidentally left the file name
"/CMakeLists.txt" in the error message. Remove it and add a test case.
Since commit v3.3.0-rc1~183^2 (Xcode: Refine quoting rules for Strings,
2015-04-09) we no longer quote strings containing a period ('.').
However, file names like "icon29x29~ipad.png" still need quoting because
they contain a tilde ('~'). Add tilde to our explicit list of
characters that need quoting because such file names will no longer
happen to be quoted because they contain a period.
The directory is at args[0], not args[1]. Introduced in commit
v2.6.0~489 (... 5889 ... tests are not found in some cases when using
add_subdirectory ..., 2008-01-18).
Since commit v3.1.0-rc1~564^2 (OS X: Use -iframework for system
framework directories, 2014-05-05) we test the version of Clang to see
if it supports -iframework. Fix the version test used for AppleClang
since it uses a different version scheme than upstream Clang.
Commit v3.3.0-rc1~196^2~7 (cmake: Simplify command clean up
loop., 2015-04-12) introduced a bug that built-in commands which
were renamed no longer had their original name restored when
cleanup is performed between configure runs. Check for that
and restore the commands with their original name.
Extend the complex test for this. That test is run by ctest with
the --build-two-config command line option.
Extend the workaround added by commit v3.2.0-rc1~278^2 (FindMPI:
Workaround Intel MPI 5.0.1 exit code problem, 2014-12-04) with an
additional/alternate keyword to recognize the case with recent GCCs.
This setting was added to support QtIFW 2.0 in commit v3.3.0-rc1~70^2~1
(CPackIFW: Add QtIFW 2.0 support, 2015-04-27). We need to load the
CPackIFW module to initialize it if it is not set.
We use a special dedicated structure to store the LINK_LIBRARIES target
property. Do not try to construct a string from a NULL value. Instead
leave the property structure empty when no value is given.
Reported-by: Ghyslain Leclerc <ghleclerc@gmail.com>
Since commit v3.3.0-rc1~397^2 (Check*CompilerFlag: Refactor method used
to pass flags, 2015-02-24) these check modules pass the flags to the
compiler front-end during linking as well as during compilation. This
breaks checks for flags like '-x c++' that are meant only for the
compilation step. Revert the change and add a test covering a
compiler-only flag.
In commit v3.3.0-rc1~352^2~3 (Genex: Allow COMPILE_LANGUAGE when
processing compile definitions, 2015-03-04) the name of the variable
used to pass preprocessor definitions to the Fortran dependency scanner
was changed to be per-language, but the actual dependency scanning code
was not updated accordingly. Update the code and add a test case.
Reported-by: Radovan Bast <radovan.bast@gmail.com>
Replace caret-headers with double-quote-headers and replace dash-headers
with caret-headers. This makes the headers match their level of nesting
according to our documentation style guide in cmake-developers(7).
Since commit v3.3.0-rc1~62^2~5 (cmTarget: Store only cmListFileContext
for CMP0023 handling, 2015-05-18) a call to target_link_libraries on a
target that was defined in another (non-ancestor) directory crashes
because no execution context is left active. Fix this by getting the
execution context from the actual cmMakefile where the current
target_link_libraries call takes place. Test this by verifying that
such calls correctly produce an error diagnostic instead of crashing.
In commit v3.3.0-rc1~49^2~2 (cmake-gui: Add --install option to add
command-line tools on OS X, 2015-05-19) the option default was set to
/usr/bin because that is where the old command line install dialog
placed the symlinks. A better default is /usr/local/bin because it is
meant for locally installed software rather than Apple-installed tools.
Also, as of OS X El Capitan, special privileges are required even for
root to modify /usr/bin but not /usr/local/bin.
The fix in commit v3.2.3~3^2 (Fix assertion failure on unmatched foreach
in function, 2015-05-18) broke handling of unmatched non-loop blocks
because it assumed all function blockers removed during error unwinding
were for loops, essentially switching the set of mishandled cases.
The purpose of the loop block push/pop operations is to define a scope
matching the lifetime of the loop function blockers. Since our function
blockers already have the proper lifetime, simply move the push/pop
operations to their constructor/destructor.
Extend the RunCMake.Syntax test with a case covering this.
The most likely documentation page a project author will read in
response to a policy warning is the page for the policy itself.
Add to every policy documentation page a note explicitly stating
that the OLD behavior is deprecated. Also mention this in the
cmake_policy() command documentation that explains how to set a
policy to OLD.
Suggested-by: Fraser Hutchison <fraser.hutchison@gmail.com>