Skip to content

[C2y] Add test coverage for WG14 N3370 #115054

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 3 commits into from
Nov 7, 2024

Conversation

AaronBallman
Copy link
Collaborator

This paper added case ranges in switch statements, which is a GNU extension Clang has supported since at least Clang 3.0.

It updates the diagnostics to no longer call this a GNU extension except in C++ mode.

This paper added case ranges in switch statements, which is a GNU
extension Clang has supported since at least Clang 3.0.

It updates the diagnostics to no longer call this a GNU extension
except in C++ mode.
@AaronBallman AaronBallman added clang:frontend Language frontend issues, e.g. anything involving "Sema" c2y labels Nov 5, 2024
@llvmbot llvmbot added the clang Clang issues not falling into any other category label Nov 5, 2024
@llvmbot
Copy link
Member

llvmbot commented Nov 5, 2024

@llvm/pr-subscribers-clang

Author: Aaron Ballman (AaronBallman)

Changes

This paper added case ranges in switch statements, which is a GNU extension Clang has supported since at least Clang 3.0.

It updates the diagnostics to no longer call this a GNU extension except in C++ mode.


Full diff: https://github.com/llvm/llvm-project/pull/115054.diff

4 Files Affected:

  • (modified) clang/include/clang/Basic/DiagnosticParseKinds.td (+7-2)
  • (modified) clang/lib/Parse/ParseStmt.cpp (+9-1)
  • (added) clang/test/C/C2y/n3370.c (+105)
  • (modified) clang/www/c_status.html (+1-1)
diff --git a/clang/include/clang/Basic/DiagnosticParseKinds.td b/clang/include/clang/Basic/DiagnosticParseKinds.td
index 78510e61a639fa..0da509280068ad 100644
--- a/clang/include/clang/Basic/DiagnosticParseKinds.td
+++ b/clang/include/clang/Basic/DiagnosticParseKinds.td
@@ -203,8 +203,13 @@ def err_expected_equal_designator : Error<"expected '=' or another designator">;
 def ext_gnu_old_style_field_designator : ExtWarn<
   "use of GNU old-style field designator extension">,
   InGroup<GNUDesignator>;
-def ext_gnu_case_range : Extension<"use of GNU case range extension">,
-  InGroup<GNUCaseRange>;
+def ext_gnu_case_range : Extension<
+  "case ranges are a GNU extension">, InGroup<GNUCaseRange>;
+def warn_c23_compat_case_range : Warning<
+  "case ranges are incompatible with C standards before C2y">,
+  DefaultIgnore, InGroup<CPre2yCompat>;
+def ext_c2y_case_range : Extension<
+  "case ranges are a C2y extension">, InGroup<C2y>;
 
 // Generic errors.
 def err_expected_expression : Error<"expected expression">;
diff --git a/clang/lib/Parse/ParseStmt.cpp b/clang/lib/Parse/ParseStmt.cpp
index 6470e55e521add..9ba3b112254933 100644
--- a/clang/lib/Parse/ParseStmt.cpp
+++ b/clang/lib/Parse/ParseStmt.cpp
@@ -890,7 +890,15 @@ StmtResult Parser::ParseCaseStatement(ParsedStmtContext StmtCtx,
     SourceLocation DotDotDotLoc;
     ExprResult RHS;
     if (TryConsumeToken(tok::ellipsis, DotDotDotLoc)) {
-      Diag(DotDotDotLoc, diag::ext_gnu_case_range);
+      // In C++, this is a GNU extension. In C, it's a C2y extension.
+      unsigned DiagId;
+      if (getLangOpts().CPlusPlus)
+        DiagId = diag::ext_gnu_case_range;
+      else if (getLangOpts().C2y)
+        DiagId = diag::warn_c23_compat_case_range;
+      else
+        DiagId = diag::ext_c2y_case_range;
+      Diag(DotDotDotLoc, DiagId);
       RHS = ParseCaseExpression(CaseLoc);
       if (RHS.isInvalid()) {
         if (!SkipUntil(tok::colon, tok::r_brace, StopAtSemi | StopBeforeMatch))
diff --git a/clang/test/C/C2y/n3370.c b/clang/test/C/C2y/n3370.c
new file mode 100644
index 00000000000000..0d654fb4ad28b9
--- /dev/null
+++ b/clang/test/C/C2y/n3370.c
@@ -0,0 +1,105 @@
+// RUN: %clang_cc1 -verify=expected,c-expected -std=c2y -Wall -pedantic %s
+// RUN: %clang_cc1 -verify=expected,c-expected,ped -std=c23 -Wall -pedantic %s
+// RUN: %clang_cc1 -verify=expected,cxx-expected,gnu -Wall -pedantic -x c++ %s
+// RUN: %clang_cc1 -verify=expected,c-expected,pre -std=c2y -Wpre-c2y-compat -Wall -pedantic %s
+
+/* WG14 N3370: Yes
+ * Case range expressions v3.1
+ *
+ * This introduces the ability to specify closed ranges in case statements in a
+ * switch statement. This was already a well-supported Clang extension before
+ * it was standardized.
+ */
+
+void correct(int i) {
+  constexpr int j = 100, k = 200;
+  switch (i) {
+  case 12 ... 14: break; /* gnu-warning {{case ranges are a GNU extension}}
+                            ped-warning {{case ranges are a C2y extension}}
+                            pre-warning {{case ranges are incompatible with C standards before C2y}}
+                          */
+  // Implementations are encouraged to diagnose empty ranges.
+  case 15 ... 11: break;  /* expected-warning {{empty case range specified}}
+                             gnu-warning {{case ranges are a GNU extension}}
+                             ped-warning {{case ranges are a C2y extension}}
+                             pre-warning {{case ranges are incompatible with C standards before C2y}}
+                           */
+  // This is not an empty range, it's a range of a single value.
+  case 10 ... 10: break; /* gnu-warning {{case ranges are a GNU extension}}
+                            ped-warning {{case ranges are a C2y extension}}
+                            pre-warning {{case ranges are incompatible with C standards before C2y}}
+                          */
+  case j ... k: break;   /* gnu-warning {{case ranges are a GNU extension}}
+                            ped-warning {{case ranges are a C2y extension}}
+                            pre-warning {{case ranges are incompatible with C standards before C2y}}
+                          */
+  }
+}
+
+void incorrect(int i) { // cxx-expected-note 2 {{declared here}}
+  switch (i) {
+  // The values have to be integer constant expressions. Note that when the
+  // initial value in the range is an error, we don't issue the warnings about
+  // extensions or incompatibility.
+  case i ... 10: break;    /* c-expected-error {{expression is not an integer constant expression}}
+                              cxx-expected-error {{case value is not a constant expression}}
+                              cxx-expected-note {{function parameter 'i' with unknown value cannot be used in a constant expression}}
+                            */
+  case 10 ... i: break;    /* c-expected-error {{expression is not an integer constant expression}}
+                              cxx-expected-error {{case value is not a constant expression}}
+                              cxx-expected-note {{function parameter 'i' with unknown value cannot be used in a constant expression}}
+                              gnu-warning {{case ranges are a GNU extension}}
+                              ped-warning {{case ranges are a C2y extension}}
+                              pre-warning {{case ranges are incompatible with C standards before C2y}}
+                            */
+  case 1.3f ... 10: break; /* c-expected-error {{integer constant expression must have integer type, not 'float'}}
+                              cxx-expected-error {{conversion from 'float' to 'int' is not allowed in a converted constant expression}}
+                            */
+  case 10 ... "a": break;  /* c-expected-error {{integer constant expression must have integer type, not 'char[2]'}}
+                              cxx-expected-error {{value of type 'const char[2]' is not implicitly convertible to 'int'}}
+                              gnu-warning {{case ranges are a GNU extension}}
+                              ped-warning {{case ranges are a C2y extension}}
+                              pre-warning {{case ranges are incompatible with C standards before C2y}}
+                            */
+  }
+
+  switch (i) {
+  // Cannot have multiple cases covering the same value.
+  // FIXME: diagnostic quality here is poor. The "previous case" note is
+  // showing up on a subsequent line (I'd expect the error and note to be
+  // reversed), and "duplicate case value 20" is showing up on a line where
+  // there is no duplicate value 20 to begin with.
+  case 10 ... 20: break; /* expected-error {{duplicate case value '11'}}
+                            expected-note {{previous case defined here}}
+                            gnu-warning {{case ranges are a GNU extension}}
+                            ped-warning {{case ranges are a C2y extension}}
+                            pre-warning {{case ranges are incompatible with C standards before C2y}}
+                          */
+  case 11: break;        /* expected-note {{previous case defined here}}
+                          */
+  case 11 ... 14: break; /* expected-error {{duplicate case value '20'}}
+                            gnu-warning {{case ranges are a GNU extension}}
+                            ped-warning {{case ranges are a C2y extension}}
+                            pre-warning {{case ranges are incompatible with C standards before C2y}}
+                          */
+  }
+
+  // The values specified by the range shall not change as a result of
+  // conversion to the promoted type of the controlling expression.
+  // FIXME: the overflow warnings seem like they probably should also trigger
+  // in C++ as they do in C.
+  switch ((unsigned char)i) {
+  case 254 ... 256: break; /* c-expected-warning {{overflow converting case value to switch condition type (256 to 0)}}
+                              gnu-warning {{case ranges are a GNU extension}}
+                              ped-warning {{case ranges are a C2y extension}}
+                              pre-warning {{case ranges are incompatible with C standards before C2y}}
+                            */
+  case 257 ... 258: break; /* c-expected-warning {{overflow converting case value to switch condition type (257 to 1)}}
+                              c-expected-warning {{overflow converting case value to switch condition type (258 to 2)}}
+                              gnu-warning {{case ranges are a GNU extension}}
+                              ped-warning {{case ranges are a C2y extension}}
+                              pre-warning {{case ranges are incompatible with C standards before C2y}}
+                            */
+  }
+}
+
diff --git a/clang/www/c_status.html b/clang/www/c_status.html
index c896b6124b6ab2..8b677095cee182 100644
--- a/clang/www/c_status.html
+++ b/clang/www/c_status.html
@@ -236,7 +236,7 @@ <h2 id="c2y">C2y implementation status</h2>
     <tr>
       <td>Case range expressions v3.1</td>
       <td><a href="https://www.open-std.org/jtc1/sc22/wg14/www/docs/n3370.htm">N3370</a></td>
-      <td class="unknown" align="center">Unknown</td>
+      <td class="full" align="center">Yes</td>
     </tr>
     <tr>
       <td>New _Lengthof() operator (v4)</td>

Copy link
Collaborator

@erichkeane erichkeane left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Release note to say we enable this in C2Y?

*/
case 11: break; /* expected-note {{previous case defined here}}
*/
case 11 ... 14: break; /* expected-error {{duplicate case value '20'}}
Copy link
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Oh my :)

@AaronBallman
Copy link
Collaborator Author

Release note to say we enable this in C2Y?

I'm not opposed, but we've always supported this feature. Similar to i and j suffixes, I didn't add a release for imaginary constants which are in the same boat (extensions we've supported forever but are now C2y features). WDYT?

@erichkeane
Copy link
Collaborator

Release note to say we enable this in C2Y?

I'm not opposed, but we've always supported this feature. Similar to i and j suffixes, I didn't add a release for imaginary constants which are in the same boat (extensions we've supported forever but are now C2y features). WDYT?

There is a minor value (at least for messaging!) for a release note that says "we looked at implementing N3370, saw it was implemented, so added sufficient testing to validate it, so now you can use it without getting GNU-ext warnings in C"

Copy link
Member

@Sirraide Sirraide left a comment

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Changes lgtm.

I don’t really have a strong opinion on the release note: on the one hand, writing one for this does feel a bit like we’d be writing a release note for C2y itself rather than for Clang, but on the other, I guess adding something like ‘Clang now diagnoses case ranges as a Clang extension rather than a GNU extension’ would be an option. Iirc I added a similar release note when I changed some variadic macro warnings.

@AaronBallman
Copy link
Collaborator Author

I've fixed up a failing test caught by precommit CI and added a release note.

@AaronBallman AaronBallman merged commit 3d0b283 into llvm:main Nov 7, 2024
9 checks passed
@AaronBallman AaronBallman deleted the aballman-wg14-n3370 branch November 7, 2024 11:53
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
c2y clang:frontend Language frontend issues, e.g. anything involving "Sema" clang Clang issues not falling into any other category
Projects
None yet
Development

Successfully merging this pull request may close these issues.

4 participants