https://github.com/drodriguez closed
https://github.com/llvm/llvm-project/pull/137884
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
https://github.com/drodriguez edited
https://github.com/llvm/llvm-project/pull/137884
___
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
@@ -0,0 +1,41 @@
+// RUN: %clang_cc1 -std=c17 -DPRE_C23 -fsyntax-only -verify -isystem %S/Inputs
%s
+// RUN: %clang_cc1 -std=c23 -DPOST_C23 -fsyntax-only -verify -isystem
%S/Inputs %s
+
+#include
+#include
+
+int printf(const char *restrict, ...);
+
+#if PRE_C23
--
drodriguez wrote:
The other place that a default sysroot might come is the CMake option
`DEFAULT_SYSROOT`. In my builds it is empty (the default), but it might be
pointing to something in those xpack builds:
https://github.com/xpack/xpack-build-box/blob/64488ebdfefd96e5eec45ab31bc170aa028fed4e
drodriguez wrote:
@ilg-ul: It seems that the driver does not receive any `-isysroot`, but it
seems that `cc1` receives `-isysroot
/Library/Developer/CommandLineTools/SDKs/MacOSX.sdk`, so I am suspicious of
`SDKROOT` maybe being in the environment. Can you check if your environment is
providin
drodriguez wrote:
> With that being said, can you clarify what you mean by "ignoring the
> command-line option", and can you expand on why the current state of upstream
> Clang is broken in your opinion? But for productivity's sake, let's take for
> granted that the canonical state is upstream
drodriguez wrote:
> But after taking a quick look, my impression is that this change reverses the
> bug fixed in #70817.
>
> To recap, that PR preferred the C++ headers available in the toolchain
> distribution over the SDK headers, when the compiler was launched via a
> symbolic link to the
drodriguez wrote:
I agree with Dmitry in that it would be preferred to have the system compiler
behaviour until the system compiler actually changes.
I have tried again with Xcode 14.3.1 (released almost 1 year ago), which
includes `include/c++/v1` side by side with the compiler, and for it, p
drodriguez wrote:
The `Xcode.app` I was using in the previous comment was 15.1. I tried again the
same steps with the last version I have (Xcode 15.3b1, 15E5178i) and I obtain
the same behaviour.
https://github.com/llvm/llvm-project/pull/80524
___
cf
drodriguez wrote:
> @drodriguez I'm not certain why you think AppleClang prefers `isysroot` over
> toolchain headers -- it might be because we simply don't ship libc++
> toolchain headers anymore and we only ship libc++ SDK headers. But if you try
> installing libc++ headers in the toolchain a
drodriguez wrote:
> I need to take a closer look, since at first reading I can't evaluate the
> consequences, especially if this does change the behaviour when -isysroot is
> **not** provided.
Small detail that I think you are aware: the `-isysroot` can be provided, but
it needs to have the `
drodriguez wrote:
@ilg-ul: you might be interested in reviewing this changes, since they change
the behaviour you introduced in #70817 when `-isysroot` is provided.
If some folks at Apple reads this: it would be preferable to have the actual
code instead of trying to guess which is the interna
https://github.com/drodriguez created
https://github.com/llvm/llvm-project/pull/80524
The current Apple Clang behaviour is to prefer `-isysroot` vs libc++ headers
side-by-side the compiler. This has been like that for several Xcode versions,
at least since Xcode 14.
The code was originally wr
Author: Richard Howell
Date: 2021-09-16T14:02:14-07:00
New Revision: e6020b2a42245d888e941eba66fbce9d0c4ab0a8
URL:
https://github.com/llvm/llvm-project/commit/e6020b2a42245d888e941eba66fbce9d0c4ab0a8
DIFF:
https://github.com/llvm/llvm-project/commit/e6020b2a42245d888e941eba66fbce9d0c4ab0a8.diff
14 matches
Mail list logo
