automatic choice of a default +llvmXY variant via selected compiler?

[René J.V. Bertin]

Hi,

The KF5 version of the KDevelop IDE uses libclang for its C/C++ parser, and requires a clang installation that includes llvm-config. Apple doesn't include that utility (and possibly other required components) in its toolchain, so one needs a 3rd party clang installation.

To be more exhaustive, this is the information requested:
../lib/llvm-3.5/bin/llvm-config --version
../lib/llvm-3.5/bin/llvm-config --includedir
../lib/llvm-3.5/bin/llvm-config --libdir
../lib/llvm-3.5/bin/llvm-config --cppflags
../lib/llvm-3.5/bin/llvm-config --ldflags
../lib/llvm-3.5/bin/llvm-config --libs core bitreader asmparser analysis
../lib/llvm-3.5/bin/llvm-config --libfiles
../lib/llvm-3.5/bin/llvm-config --prefix
../lib/llvm-3.5/bin/llvm-config --src-root

Port:kf5-kdevelop will thus propose different llvmXY variants that determine which clang version is used for the parser. At the moment it also picks a default if no variant is set explicitly, but I'm not really comfortable with that (or the alternative, raising an error if no variant is set).

It also turns out I had to blacklist at least the 10.9 Apple clang (602) while apparently the current 3.5 clang release is supposed to build the code without issues.

Either way, there is thus at least one OS X version where it would stand to reason to use the internal compiler version selection procedure to determine the default clang version to use for the parser (default +llvmXY variant). That way users can install the llvm/clang ports of choice, and don't have to repeat that choice through explicit selection of an llvm variant for KDevelop (and evidently for other ports that also depend on llvm or clang libraries). It should also allow for more transparent upgrade paths.

I would think that an adaptive mechanism to determine a default variant isn't incompatible with reproducible builds (or at least an accepted/able exception) because the selected variant will still build the same everywhere.
If so, is the definitive choice of compiler known sufficiently early to use it for defining a default variant, and/or is there already something in place to do this? I don't have access to my Mac right now so I cannot just check for myself.

R.