[clang] [clang][driver][darwin] DarwinSDKInfo doesn't match sufficiently modified triples, doesn't warn for unsupported architectures (PR #204061)
Steven Wu via cfe-commits
cfe-commits at lists.llvm.org
Wed Jun 17 12:58:49 PDT 2026
================
@@ -107,10 +107,9 @@ TEST(DarwinSDKInfo, PlatformPrefix) {
ASSERT_TRUE(SDKInfo);
EXPECT_EQ(SDKInfo->getPlatformPrefix(Triple("arm64-apple-macos26.0")),
"/System/macOSSupport");
- // The triple's architecture doesn't matter.
- EXPECT_EQ(SDKInfo->getPlatformPrefix(Triple("ppc-apple-macos26.0")),
- "/System/macOSSupport");
- // OSes that aren't specified in the SDK never get a system prefix.
+ // When there are multiple system prefixes, non-matching triples don't get a
----------------
cachemeifyoucan wrote:
`... which has undergone modifications from the original -target value, which can sometimes even include changes to the triple's environment...` Does any test cover this case? I wonder what is the case that trigger this?
https://github.com/llvm/llvm-project/pull/204061
More information about the cfe-commits
mailing list