[Lldb-commits] [PATCH] D55328: [CMake] Revised LLDB.framework builds

Stefan Gränitz via Phabricator via lldb-commits lldb-commits at lists.llvm.org
Mon Dec 10 10:24:12 PST 2018


sgraenitz marked 3 inline comments as done.
sgraenitz added inline comments.


================
Comment at: cmake/modules/LLDBConfig.cmake:64
+  set(LLVM_CODESIGNING_IDENTITY lldb_codesign CACHE STRING "" FORCE)
+endif()
+
----------------
labath wrote:
> sgraenitz wrote:
> > sgraenitz wrote:
> > > @labath Could this be a way to phase out LLDB_CODESIGN_IDENTITY and move to the LLVM one?
> > > 
> > > * An **explicitly set** LLDB value takes precedence
> > > * Warn if a non-empty LLVM value is overwritten
> > > * Default to lldb_codesign if none is set explicitly
> > Not sure if it is too intrusive.
> You say "phase out", but it's not clear to me how would that happen. What would be the next step here? I don't think llvm will ever get `lldb_codesign` as the default value for LLVM_CODESIGN_IDENTITY.
> 
> Given that we seem to have two parts of the project with different default policies on code signing, I think the cleanest approach would be to have code signing identity overridable on a per-call basis.
> 
> You might not even need to modify `add_llvm_executable` for this to work. I think you might be able to just locally do `set(LLVM_CODESIGN_IDENTITY ${LLDB_CODESIGN_IDENTITY})` in the scope of `add_lldb_executable` and the signing code should pick that up instead of the cache variable (the cmake scope rules always confuse me, but I think this how it works).
> You say "phase out", but it's not clear to me how would that happen.

Yes sorry, that was dump phrasing. Explained in more detail in D55013.


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D55328/new/

https://reviews.llvm.org/D55328





More information about the lldb-commits mailing list