[PATCH] D97717: [SYCL] Rework the SYCL driver options

[Aaron Ballman via Phabricator via cfe-commits]

aaron.ballman added a comment.

In D97717#2599728 <https://reviews.llvm.org/D97717#2599728>, @bader wrote:

> @aaron.ballman, it looks like unittests should be updated as well. Please, take a look at failures in pre-merge checks.

Thanks, I'll correct those!



================
Comment at: clang/include/clang/Basic/LangOptions.def:252
 LANGOPT(SYCLIsDevice      , 1, 0, "Generate code for SYCL device")
+LANGOPT(SYCLIsHost        , 1, 0, "SYCL host compilation")
 ENUM_LANGOPT(SYCLVersion  , SYCLMajorVersion, 1, SYCL_None, "Version of the SYCL standard used")
----------------
bader wrote:
> I'm okay with that change, but IIRC, @ABataev suggested to have both -fsycl and -fsycl-is-device options to align with OpenMP mode (full discussion is here - https://reviews.llvm.org/D72857#inline-674377).
Thanks for pointing this out!

As far as the frontend is concerned, being in "SCYL" mode is not a useful construct because you don't know *which* SYCL mode you're in (host or device). Downstream, we're finding that having `SYCL`, `SYCLIsDevice`, and `SYCLIsHost` introduces bugs because developers may check `SYCL` when they mean `SYCLIsDevice` or they do things like check `SYCL && SYCLIsDevice` which is redundant.


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D97717/new/

https://reviews.llvm.org/D97717