[clang] Add test for CWG3129

[eisenwave]

Resolves https://wg21.link/CWG3129

Clang has supported floating-point literals with overly tiny or huge exponents since 3.0, with only a warning emitted.

[llvmorg-github-actions]

@llvm/pr-subscribers-clang

Author: Jan Schultke (eisenwave)

Changes

Clang has always supported floating-point literals with overly tiny or huge exponents, with only a warning emitted.

---

Full diff: https://github.com/llvm/llvm-project/pull/206335.diff

1 Files Affected:

• (modified) clang/www/cxx_dr_status.html (+1-1)

diff --git a/clang/www/cxx_dr_status.html b/clang/www/cxx_dr_status.html
index ef30b5eefea13..4eaf0c44ab082 100755
--- a/clang/www/cxx_dr_status.html
+++ b/clang/www/cxx_dr_status.html
@@ -21720,7 +21720,7 @@ <h2 id="cxxdr">C++ defect report implementation status</h2>
     <td>[<a href="https://wg21.link/lex.fcon">lex.fcon</a>]</td>
     <td>DR</td>
     <td>Clarify which <I>floating-point-literal</I>s are valid</td>
-    <td class="unknown" align="center">Unknown</td>
+    <td class="full" align="center">Clang 2.7</td>
   </tr>
   <tr id="3130">
     <td><a href="https://cplusplus.github.io/CWG/issues/3130.html">3130</a></td>

[Endilll]

Sorry, a bit more work is needed here.

1. You need to add tests to clang/test/CXX/drs/cwg31xx.cpp.
2. You need to take your tests to Compiler Explorer to see what was the first Clang release that exhibited the expected behavior (call it R).
3. You need to leave a special comment in the test specifying R.
4. You need to run clang/www/make_cxx_dr_status to update the cxx_dr_status.html. If it produces more changes, include only the changes relevant for CWG3129. If the script terminates with an error about an unrelated Core issue, let me know, and I'll fix that.

[Endilll]

The PR description would also benefit from a link to CWG3129.

[eisenwave]

You need to take your tests to Compiler Explorer to see what was the first Clang release that exhibited the expected behavior (call it R).

I've tried that, and I couldn't find a version old enough where it wouldn't do what the issue wants. My impression is that 2.7 is being used to mean "since forever" in these issue resolutions, but that's so old, it's not even on Compiler Explorer.

Hmm nevermind, 2.7 doesn't accept it, so 3.0 is actually the fix verison https://godbolt.org/z/zW3c8r6eP

[eisenwave]

The one problem with putting it into cwg31xx.cpp is that I had to get rid of the // cxx98-14-no-diagnostics comment; otherwise this test just wouldn't pass.

Maybe that means it should have been in a separate file after all, or is there some clever way to keep the comment?

[Endilll]

The one problem with putting it into cwg31xx.cpp is that I had to get rid of the // cxx98-14-no-diagnostics comment; otherwise this test just wouldn't pass.

This is totally fine. Special needs I mentioned before are scoped to RUN lines.

You should change the title of the PR to something along the lines of "Add test for CWG3129".

Let me know when it's ready to be merged.

[eisenwave]

@Endilll should be good to go.

OH wait, we got some CI failures.

[eisenwave]

Very weird. The CI failures is because it's printing

1.18973149535723176508575932662800702E+4932

as the maximum for long double in CI but the test expectations is

1.18973149535723176502E+4932

which did work for me locally. OH well ...

[eisenwave]

@Endilll should be good to merge now.