[llvm-branch-commits] [libcxx] [libc++][modules] Adds module testing. (PR #76246)

[Chuanqi Xu via llvm-branch-commits]

================
@@ -131,13 +138,65 @@ def parseScript(test, preamble):
     script += preamble
     script += scriptInTest
 
+    has_std_module = False
+    has_std_compat_module = False
+    for module in modules:
+        if module == "std":
+            has_std_module = True
+        elif module == "std.compat":
+            has_std_compat_module = True
+        else:
+            script.insert(
+                0,
+                f"echo \"The module '{module}' is not valid, use 'std' or 'std.compat'\"",
+            )
+            script.insert(1, "false")
+            return script
+
+    if modules:
+        # This flag is needed for both modules.
+        moduleCompileFlags.append("-fprebuilt-module-path=%T")
----------------
ChuanqiXu9 wrote:

I didn't read the patch completely so I don't know the concatenated command line. But the behavior matches current expectation of clang. Now when clang sees an `import` for a named module, clang will only find that named module by `-fprebuilt-module-path` or `-fmodule-file=<module-name>=<path-to-the-module-file>`.

I guess you're talking about the case about:

```
clang++ std.compat.pcm use.cc -o use
```

where use.cc uses std.compat module. 

This command line will be split into 3 processes like:

```
clang_cc1 std.compat.pcm  -c -o std.compat.o
clang_cc1 use.cc -c -o use.o
clang_cc1 std.compat.o use.o -o use
```

So the second step won't know about `std.compat.pcm`.

While it is technically possible to improve that, I feel it is not so useful since the major expected users of modules should use a build system instead of concatenating the command line manually.

https://github.com/llvm/llvm-project/pull/76246