CMakeLists.txt contains now the version numbers for DMD and (next) LDC release.
If a .git folder is found then the LDC version is replaced by the first 6 chars
of the Git revision.
Possible improvements:
- If the build is not done at the master branch then it could be useful to check for a tag and use the tag instead of the revision. (for release builds)
- Maybe it is useful to include the branch name.
This fixes issue #366.
There might be still some pieces left here and there, and
there is certainly code that could be rewritten in a nicer
way with the 3.0 requirement out of the picture.
The new command line option conflicted with a builtin LLVM 3.0
one. The functionality will be added back in after the 0.11.0
release branch has been created.
This reverts commit 6a1bc70bd7 and
subsequent fixes.
Even though this argument design conflates two separate concepts
(ABI and hardware/software implementation), I chose to go
with it since users are liekly know it from GCC and the
combination of softloat operations with hardfloat ABI makes
no sense.
I didn't implement it for old LLVM versions, as ARM EABI
exception handling requires LLVM 3.3+ anyway, without which
LDC would be useless anyway.
1. Main include corresponding to .cpp file, if any.
2. DMD and LDC includes.
3. LLVM includes.
4. System includes.
Also updated a few include guards to match the default format.
I hope I have untangled the checks for "native" Windows (Triple::Win32)
vs. Windows/MinGW/Cygwin (Triple::isOSWindows) correctly.
MinGW needs some default libraries as well, has to be fixed later.
I removed the copyright notice. I don't think we are obliged to
display it for DMD; on the other hand, if we do, we might also
have to display the libconfig one (LGPL). Thus, this seemed like
the better choice.
The indentation is to align with the registered target output,
which we cannot control.
This is what DMD does, and arguably the expected behavior – otherwise,
there would be no way other than using -od to prevent races if two
compiler instances could ever compile the same file(s) at the same
time (e.g. for debug/release configurations).
Should we transition to assuming -singleobj by default?
A lot of system specific knowledge is already present in LLVM. This is used to populate several fields in global.params instead of hard coded values in main(). Ensures that the frontend and LLVM have always the same values.
