Store the main dependency as the first entry in the dependency list plus
a boolean member indicating its existence. Note that this slightly
changes existing behavior: the main dependency was previously the last
entry of the dependency list.
Note 1: `detail::AddCustomCommandToTarget()` resets cc,
since cc is not moved away.
Note 2: In `detail::AddUtilityCommand()`, a few vars are preserved
before using. Their refs will be alive in most cases, but cc might
be destroyed in the future.
To avoid ambiguity on std::string assigment between the following two
cmProp cast operators:
* operator const std::string&() const noexcept
* operator cm::string_view() const noexcept
Remove unnecessary check of policy CMP0049. The policy can never
trigger on our internally-generated name because it has no variable
references.
The rename in commit 0ed5ce4cd8 (cmTarget: Rename AddSource method for
backward compatibility., 2014-03-17, v3.1.0-rc1~688^2~17) made it look
like this code path depended on CMP0049. Then commit 0e1faa28cb
(cmMakefile: Separate custom command setup from actual creation,
2019-09-14, v3.16.0-rc1~85^2) and commit ea1bed34b2 (cmMakefile: Extract
utilities used for creation of custom commands, 2019-09-21,
v3.16.0-rc1~52^2~1) built additional infrastructure to thread that
dependence through the call stack. Remove it all.
Since commit 777ceaea94 (cmMakefile: Delay custom command creation,
2019-10-17, v3.17.0-rc1~352^2) we process custom command declarations at
generate time. Therefore we do not need to look up what source file
holds the custom command producing a given output until generate time.
Since commit 777ceaea94 (cmMakefile: Delay custom command creation,
2019-10-17, v3.17.0-rc1~352^2) we process custom command declarations
at generate time. This includes the append-to-non-existing-command
check, so we do not need it at configure time.
The execution file path stack and the backtrace stack are kept in sync.
At all call sites of `GetExecutionFilePath`, the execution file path
matches the path in the context at the top of the backtrace stack.
The only call sites that pass the explicit file name argument are in
function blocker `ArgumentsMatch` methods for `function` and `macro`.
We already ensure that they are balanced within a file scope, and the
RAII helpers `BuildsystemFileScope` and `ListFileScope` ensure that the
backtrace and execution list file stacks unwind to the matching level.
Therefore we can assume that the file name where we are checking for
matching arguments matches starting file name where those arguments
first appeared, and do not need to pass it explicitly.
#pragma once is a widely supported compiler pragma, even though it is
not part of the C++ standard. Many of the issues keeping #pragma once
from being standardized (distributed filesystems, build farms, hard
links, etc.) do not apply to CMake - it is easy to build CMake on a
single machine. CMake also does not install any header files which can
be consumed by other projects (though cmCPluginAPI.h has been
deliberately omitted from this conversion in case anyone is still using
it.) Finally, #pragma once has been required to build CMake since at
least August 2017 (7f29bbe6 enabled server mode unconditionally, which
had been using #pragma once since September 2016 (b13d3e0d)). The fact
that we now require C++11 filters out old compilers, and it is unlikely
that there is a compiler which supports C++11 but does not support
#pragma once.
This option has been broken since commit b9f9915516 (cmMakefile: Remove
VarUsageStack., 2015-05-17, v3.3.0-rc1~52^2). That commit removed the
check that an initialized variable has actually been used and caused the
option to warn on every variable ever set. This was not caught by the
test suite because the test for the feature only checked that warnings
appear when needed and not that they do not appear when not needed.
The option was never very practical to use. Remove it to avoid the
runtime cost of usage tracking and checks for every variable (which we
were doing even when the option was not used).
Adds a flag to indicate that pipe output from a custom command should be
interpreted as UTF-8 encoded. This change does not introduce a public
way to set the flag, but generators that create internally-generated
commands know if they are calling cmake, which uses UTF-8 pipes.
MSBuild added support for interpreting output of PreBuildEvent,
PreLinkEvent, PostBuildEvent, and CustomBuildStep as UTF-8. This change
will appear in Visual Studio 16.6 Preview 3. It is opt-in, and you need
to add the StdOutEncoding tag. MSBuild treats these as property bags so
if we emit the tag for earlier versions of Visual Studio it would be
safely ignored. This change emits the StdOutEncoding tag and sets it to
UTF-8 whenever the custom command UTF-8 pipe flag is set. This fixes
globalization issues when the output from cmake contained characters
that required MSBuild to interpret as UTF-8 before displaying them.
