[PATCH] D55150: Emit warnings from the driver for use of -mllvm or -Xclang options.

Thu Dec 6 10:47:58 PST 2018

jyknight added a comment.

In D55150#1321705 <https://reviews.llvm.org/D55150#1321705>, @kristina wrote:

> > There is a cost to having people encode these flags into their build systems -- it can then cause issues if we ever change these internal flags. I do not think any Clang maintainer intends to support these as stable APIs, unlike most of the driver command-line. But, neither -Xclang nor -mllvm obviously scream out "don't use this!", and so people may then add them to their buildsystems without causing reviewers to say "Wait...really? Are you sure that's a good idea?".
>
> This is why I proposed a compromise, allow this warning to be disabled completely for people actively using those flags, which are pretty much exclusively toolchain developers (well basically what I proposed, since it's not clear what counts as a build made by someone who is working and debugging a pass, being fully aware of those flags, using the subset of them specific to that pass/feature, I would say assertion+dump enabled builds are closest, but having an explicit build flag for it would be nicer). It's more unified than everyone either adding workarounds into build systems or disabling it completely (by just removing it).

I mean, I'm not much opposed to adding that -- just that any new build-time options is always a minor shame. But you didn't respond to the other part of my message -- is adding `-Wno-experimental-driver-option` to your compile-flags not already a completely sufficient solution for your use-case?

> Besides, I don't think this really ever surfaced as an issue, until Chandler's idea with regards to an unsupressable warning for performance tests for 7.x.x

The primary impetus for me was actually the discovery that Boost's build system was using "-Xclang -include-pth" a few weeks back.

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D55150/new/

https://reviews.llvm.org/D55150