It is never read externally.  The CollapseFullPath removed in this commit
is a repeat of a similar call inside ReadListFile.

There is no need to store this as a member variable.

The backtrace will soon not be implemented in terms of a stack of
cmListFileContext objects.  Keep the cmListFileContext in the API
for convenience for now.

Move wrapping existing code from cmMakefile, and simplify the
implementation there.

This is cheaper than getting the whole backtrace, and the cmListFileBacktrace
will not always be composed of cmListFileContext objects.

The lexical scope counting added by commit v3.2.0-rc1~332^2~1 (Track
nested loop levels in CMake language with a stack of counters,
2014-11-18) forgot to account for scopes popped by error messages about
unclosed scopes.  Teach the error handler to pop the lexical scope it
reports as unclosed.  Re-order the lexical scope RAII object to be
inside the variable scope RAII object scope so that the lexical scope
is fully closed before we check assertions about variable scopes.

Extend the RunCMake.Syntax test with a case covering this.

For consistency with all other closure-related methods.

The variable is initialized by the constructor already.

Do not depend on the local generator for that.

Access the state from the local generator in the cmMakefile.

Add policy CMP0061 to maintain compatibility for existing projects.

Hide the detail that it is a std::map.

Since commit 80909041 (cmMakefile: Disable copy constructor, 2015-04-18)
the only call to cmMakefile::Initialize is in one constructor.  Inline
it and drop the separate method.

In commit 86f3cd0f (cmMakefile: Require the localGenerator in the
constructor, 2015-04-27) we added use of GetDefinition to the
Initialize method.  Set SuppressWatches before calling this method
since GetDefinition checks it.

Both for convenience, and because cmMakefile should not be the
way to get access to the local generator, so that should go away
in the future.

No callers handle the possibility that it can return null, and that
is no longer possible anyway.

All structural knowledge of the stack of scopes is now external.

The cmDefinitions::Get will change behavior in follow up commits.

Nils Gladitz authored Apr 29, 2015 and Brad King committed Apr 30, 2015

This reverts commit 242c3966 (add_custom_command: Diagnose
MAIN_DEPENDENCY limitation, 2015-03-09) and the follow up commit
b372a99a (UseSWIG: Do not use MAIN_DEPENDENCY on custom commands,
2015-03-26).

I misdiagnosed the underlying issue that prompted creation of policy CMP0057.
The actual issue surfaces when a single custom command's MAIN_DEPENDENCY
is listed in more than one target; this issue will have to be addressed
independently.

Iterate directly over the parent content provided by cmMakefile.

Accept a range of cmDefinitions*.

The purpose of the stack is to allow access only to the top of it.  Access
to items which are not at the top is needed, so cmDefinitions objects
get a Parent pointer.

The existence of the Parent pointer is a workaround for the inappropriate
use of stack in the first place.  Remove it now.

Construct the final list directly in a named return value.  Use
a single set to track bindings already found.

Co-Author: Brad King <brad.king@kitware.com>

This is more efficient and we lose nothing.

Put knowledge of the implementation details in one place.