build: removing xcrun from darwin build

[ivanacostarubio]

Hello,

I'm tackling the issue: #18959

Initial discussion: #18743 (comment)

I was not able to remove the xcrun invocation to get the SDK path. All others removed.

We can see below how this behaves when multiple Xcode are present and we switch among them.

$ sudo xcode-select --switch /Applications/Xcode11.7.app

$  make -C depends print-darwin_CXX
darwin_CXX = clang++ -mmacosx-version-min=10.14 -stdlib=libc++ --sysroot /Applications/Xcode11.7.app/Contents/Developer/Platforms/MacOSX.platform/Developer/SDKs/MacOSX.sdk

$ clang++ -mmacosx-version-min=10.14 -stdlib=libc++ --sysroot /Applications/Xcode11.7.app/Contents/Developer/Platforms/MacOSX.platform/Developer/SDKs/MacOSX.sdk --version
Apple clang version 11.0.3 (clang-1103.0.32.62)
Target: x86_64-apple-darwin20.3.0
Thread model: posix
InstalledDir: /Applications/Xcode11.7.app/Contents/Developer/Toolchains/XcodeDefault.xctoolchain/usr/bin

$ sudo xcode-select --switch /Applications/Xcode12.3.app/

$  make -C depends print-darwin_CXX
darwin_CXX = clang++ -mmacosx-version-min=10.14 -stdlib=libc++ --sysroot /Applications/Xcode12.3.app/Contents/Developer/Platforms/MacOSX.platform/Developer/SDKs/MacOSX.sdk

$ clang++ -mmacosx-version-min=10.14 -stdlib=libc++ --sysroot /Applications/Xcode12.3.app/Contents/Developer/Platforms/MacOSX.platform/Developer/SDKs/MacOSX.sdk --version
Apple clang version 12.0.0 (clang-1200.0.32.28)
Target: x86_64-apple-darwin20.3.0
Thread model: posix
InstalledDir: /Applications/Xcode12.3.app/Contents/Developer/Toolchains/XcodeDefault.xctoolchain/usr/bin

[dongcarl]

Not sure we reached a conclusion on that issue, but your contribution here is very worthwhile: the fact that we need xcrun to get the SDK path says to me that we should actually keep all the xcrun calls, since we can expect xcrun to be consistent with itself. What I mean is: the SDK that xcrun returns is more likely to be compatible with the clang that xcrun returns than just a bare clang.

[fanquake]

I think Carl's comments make sense, and we can just leave this as is for now. We can close #18959 as a "wont change". Thanks for the investigation, and for putting together a PR @ivanacostarubio. Feel free to grab another good first issue and ping me if you have any questions.

[DrahtBot]

Guix builds

File	commit 9212e67 (master)	commit e3d010f (master and this pull)
*-aarch64-linux-gnu-debug.tar.gz	7c4e2b7f7b7e08a3...	dda361b1fdb6e1b0...
*-aarch64-linux-gnu.tar.gz	e099872c0b0f3c04...	29c5f899a9599117...
*-arm-linux-gnueabihf-debug.tar.gz	79dbf1c1bb7ba758...	36a7a67770156c17...
*-arm-linux-gnueabihf.tar.gz	0b18756fdb3c85f3...	2aa8e56fc1399d68...
*-osx-unsigned.dmg	2b0245b1fa1ee744...	eea5453668fd17b5...
*-osx-unsigned.tar.gz	efe02c8bd6bf1c56...	ba75a9d32009d9bf...
*-osx64.tar.gz	94512611694d575e...	9fc7825b0a07c238...
*-powerpc64-linux-gnu-debug.tar.gz	bac482b02a75711c...	6b886809fd32a50e...
*-powerpc64-linux-gnu.tar.gz	9ab68ffa85b64620...	a806b3dac972d623...
*-powerpc64le-linux-gnu-debug.tar.gz	2fac9038e6c90ea4...	5ae35e79790621a5...
*-powerpc64le-linux-gnu.tar.gz	67c8bab214dbf998...	234a98529c5f9863...
*-riscv64-linux-gnu-debug.tar.gz	e9334b54f1743f8c...	a3a47cf01273fd70...
*-riscv64-linux-gnu.tar.gz	104be99be035aa07...	de12b896e8c7b715...
*-win-unsigned.tar.gz	01dc26708072002b...	648a21d625d4fd1e...
*-win64-debug.zip	ce517409c7624d28...	852c8cf6ec6e0e59...
*-win64-setup-unsigned.exe	73bcc94ed2e90ee6...	669e132e05bf111f...
*-win64.zip	995d9f34a5779d41...	f89bc2a1cb78f1a5...
*-x86_64-linux-gnu-debug.tar.gz	5ee23de73377ceb5...	b7a899056e2ab8bb...
*-x86_64-linux-gnu.tar.gz	6c0c194cd088df6a...	80bd342025c2ce9c...
*.tar.gz	2e8b970a9ba9bc44...	095688a4a2945353...
guix_build.log	c94b5568ba42535d...	6590e140e1b09b20...
guix_build.log.diff		da0fea5697dbdc10...