================
@@ -41,9 +41,25 @@ COFF Improvements
 * Added support for ``--time-trace`` and associated 
``--time-trace-granularity``.
   This generates a .json profile trace of the linker execution.
 
+* Prefer library paths specified with ``-libpath:`` over the implicitly
+  detected toolchain paths.
+
 MinGW Improvements
 ------------------
 
+* Added support for many LTO and ThinLTO options.
+
+* LLD no longer tries to autodetect and pick up MSVC/WinSDK installations
----------------
mstorsjo wrote:

In mingw mode, one shouldn't use MSVC/WinSDK (headers/)libraries - there's 
really no use case for that.

The root cause is that in mingw mode, one always link by calling the compiler 
driver (just like for regular unix targets), which passes all the necessary 
search paths to the linker. In MSVC style environments, one usually invokes the 
linker directly (`link.exe` or `lld-link`), and it picks up a bunch of implicit 
search paths (from the environment variable `LIB`, and LLD can also look for 
MSVC installs in default paths and/or based on registry or other things, IIRC). 
When operating in mingw mode, those implicit default paths are only a potential 
source of unexpected behaviours and surprises, so there's no opt-in.

What about:
> LLD no longer tries to autodetect and use library paths from MSVC/WinSDK 
> installations when run in MinGW mode; that mode of operation shouldn't ever 
> be needed in MinGW mode, and could be a source of unexpected behaviours.


https://github.com/llvm/llvm-project/pull/80011
_______________________________________________
llvm-branch-commits mailing list
llvm-branch-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/llvm-branch-commits

Reply via email to