At this point, it is an interface to the cache. It will be extended
to be a universal interface for access to and manipulation of
configuration-time data (defintions, properties on targets,
directories, source files etc).
This will allow porting all command implementations away
from the cmMakefile and cmTarget classes, and result in something
more-purely related to configuration-time processing of cmake
commands. That should serve at least the following goals:
* Split the CMake implementation more definitively into three
stages: Configuration, computation and generation, and be able to
implement each optimally for memory access patterns etc.
* Make better IDE integration possible by making more configuration
data available.
* Make it possiblte to use a smaller library than CMakeLib.a in
cpack and ctest, resulting in smaller executables.
* Make it possible to run the configure step multiple times in
the same CMake run (#14539).
Manage its lifetime in the cmake class, and add a convenience accessor
to cmMakefile.
This topic was never tested without some follow-up commits. The
GetCacheEntryValue API returns a pointer to memory freed on return.
It will have to be revised along with the rest of the original topic.
With PushScope and PopScope, keeping track of another bit of data for
each scope isn't easy. Instead, store it as another CMake variable so it
gets implicitly tracked along with everything else.
This works in a revert of commit
7d674b5f0b.
This reverts commit v3.1.0-rc1~557^2~2 (ClearMatches: Only clear matches
which were actually set, 2014-03-12). The optimization did not track
the match count in the same scope as the variables, allowing possible
inconsistency.
Resolve conflicts in Source/cmIfCommand.cxx, Source/cmMakefile.cxx,
and Source/cmMakefile.h by moving the changes to the new location
of the code involved.
It gets incremented while entering a loop block (e.g. foreach or while)
and gets decremented when leaving the block. Because scope borders for
example at function borders must be taken into account the counter is
put into a stack. With every new scope an empty counter is pushed on the
stack, when leaving the scope the original value is restored.
This will allow easy querying if the break command is properly nested
within a loop scope.
Signed-off-by: Gregor Jasny <gjasny@googlemail.com>
A common idiom in CMake-based build systems is to have custom commands
that generate files not listed explicitly as outputs so that these
files do not have to be newer than the inputs. The file modification
times of such "byproducts" are updated only when their content changes.
Then other build rules can depend on the byproducts explicitly so that
their dependents rebuild when the content of the original byproducts
really does change.
This "undeclared byproduct" approach is necessary for Makefile, VS, and
Xcode build tools because if a byproduct were listed as an output of a
rule then the rule would always rerun when the input is newer than the
byproduct but the byproduct may never be updated.
Ninja solves this problem by offering a 'restat' feature to check
whether an output was really modified after running a rule and tracking
the fact that it is up to date separately from its timestamp. However,
Ninja also stats all dependencies up front and will only restat files
that are listed as outputs of rules with the 'restat' option enabled.
Therefore an undeclared byproduct that does not exist at the start of
the build will be considered missing and the build will fail even if
other dependencies would cause the byproduct to be available before its
dependents build.
CMake works around this limitation by adding 'phony' build rules for
custom command dependencies in the build tree that do not have any
explicit specification of what produces them. This is not optimal
because it prevents Ninja from reporting an error when an input to a
rule really is missing. A better approach is to allow projects to
explicitly specify the byproducts of their custom commands so that no
phony rules are needed for them. In order to work with the non-Ninja
generators, the byproducts must be known separately from the outputs.
Add a new "BYPRODUCTS" option to the add_custom_command and
add_custom_target commands to specify byproducts explicitly. Teach the
Ninja generator to specify byproducts as outputs of the custom commands.
In the case of POST_BUILD, PRE_LINK, and PRE_BUILD events on targets
that link, the byproducts must be specified as outputs of the link rule
that runs the commands. Activate 'restat' for such rules so that Ninja
knows it needs to check the byproducts, but not for link rules that have
no byproducts.
Teach the add_custom_command and add_custom_target commands a new
USES_TERMINAL option. Use it to tell the generator to give the command
direct access to the terminal if possible.
No call sites pass NULL to the output argument, so take it by
reference to avoid the if(output) conditions. Propagate the
change through the TryCompile APIs that call it.
When CMP0053 is in WARN mode, variables get expanded twice, leaking the
fact that the string was expanded twice and changing behavior. Instead,
suppress variable watches when running the expansion to trigger the
CMP0053 warning.
Allow setting build properties based on the features available
for a target. The availability of features is determined at
generate-time by evaluating the link implementation.
Ensure that the <LANG>_STANDARD determined while evaluating
COMPILE_FEATURES in the link implementation is not lower than that
provided by the INTERFACE of the link implementation. This is
similar to handling of transitive properties such as
POSITION_INDEPENDENT_CODE.
Add properties and variables corresponding to CXX equivalents.
Add features for c_function_prototypes (C90), c_restrict (C99),
c_variadic_macros (C99) and c_static_assert (C11). This feature
set can be extended later.
Add a <PREFIX>_RESTRICT symbol define to WriteCompilerDetectionHeader
to conditionally represent the c_restrict feature.
Introduce a new implementation of ExpandVariablesInString and select
between the old and new implementations based on policy CMP0053.
Instead of cmCommandArgumentParserHelper, use a custom parser with our
own stack. This is much faster and works well for our simple grammar.
The new behavior of CMP0053 should expand @VAR@ syntax only in certain
contexts. All existing EVIS callers use "replaceAt == true" so
hard-code our call to the old implementation. Update the signature to
default to "replaceAt == false" and pass "replaceAt == true" explicitly
in the call sites for configure_file and string(CONFIGURE).
Testing the configure (no generate) step with ParaView shows ~20%
performance improvement.
In terms of complete configure/generate steps, further testing with
ParaView shows a 20% performance improvement over 2.8.12.2 with Unix
Makefiles and minimal with Ninja. Ninja is less because it generate step
is the expensive part (future work will address this) by a long shot and
these changes help the configure step for the most part.
3f517522 StoreMatches: Minor cleanups
ef62fbad ClearMatches: Store match variable names statically
f718b30a ClearMatches: Only clear matches which were actually set
ClearMatches was clearing many variables which were never set in the
first place. Instead, store how many matches were made last time and
only clear those. It is moved to the cmMakefile class since it is a
common utility used by multiple commands.
Stephen Kelly
Brad King
Ben Boeckel
Gregor Jasny
Peter Collingbourne
Nils Gladitz