[NFC] Prevent shadowing a variable declared in `if`

[ken-matsui]

Summary

Prevents confusion over which S is referenced in the final else branch if such use is added.

[hubert.reinterpretcast]

@ken-matsui, can you provide some rationale for the change (got compiler warning/error)?

[ken-matsui]

@hubert.reinterpretcast,

Sorry to have missed providing a summary.

In most cases, shadowing a declaration of a local variable should be avoided to prevent making others confused and mistakes because we need to figure out the boundaries of the declaration. In this case, the variable declared by const char *S in if can be accessed by every branch. That means that declarations in else if had shadowed the declaration in if, so I changed them to use the same declaration used in if.

[hubert.reinterpretcast]

In D124556#3479833, @ken-matsui wrote:

@hubert.reinterpretcast,

Sorry to have missed providing a summary.

You can still provide one by using the "Edit Revision" link.

I suggest something like:
Prevents confusion over which S is referenced in the final else branch if such a use is added.

[ken-matsui]

@hubert.reinterpretcast,

Thank you for your suggestion! I’ve updated the summary.

[hubert.reinterpretcast]

LGTM; thanks for the patch!

[ken-matsui]

Thank you for your review!

[ken-matsui]

@hubert.reinterpretcast

Sorry, I'm a newbie here, but is there anything I should do after getting approved?

(I could not execute arc land due to 403 error to the llvm-project)

remote: Permission to llvm/llvm-project.git denied to ken-matsui.
fatal: unable to access 'https://github.com/llvm/llvm-project.git/': The requested URL returned error: 403

[hubert.reinterpretcast]

In D124556#3480228, @ken-matsui wrote:

@hubert.reinterpretcast

Sorry, I'm a newbie here, but is there anything I should do after getting approved?

I'm not sure if the instructions are a bit out-of-date: https://llvm.org/docs/DeveloperPolicy.html#obtaining-commit-access

If you intend to contribute going forward, the idea would be to get a few patches approved and committed on your behalf and then to request commit access. Until you have commit access, you will need to ask people on the review to help you commit.

I can commit this patch for you.

[ken-matsui]

@hubert.reinterpretcast

I see. Thank you for the detailed information!

Could you please commit this patch?

[ken-matsui]

Thank you!

[hubert.reinterpretcast]

In D124556#3482224, @ken-matsui wrote:

Thank you!

Thanks in turn.

clang/lib/Basic/Diagnostic.cpp

while (DiagStr != DiagEnd) {
     case DiagnosticsEngine::ak_tokenkind: {
       tok::TokenKind Kind = static_cast<tok::TokenKind>(getRawArg(ArgNo));
       assert(ModifierLen == 0 && "No modifiers for token kinds yet");
 
       llvm::raw_svector_ostream Out(OutStr);
       if (const char *S = tok::getPunctuatorSpelling(Kind))
         // Quoted token spelling for punctuators.
         Out << '\'' << S << '\'';
-      else if (const char *S = tok::getKeywordSpelling(Kind))
+      else if ((S = tok::getKeywordSpelling(Kind)))
         // Unquoted token spelling for keywords.
         Out << S;
-      else if (const char *S = getTokenDescForDiagnostic(Kind))
+      else if ((S = getTokenDescForDiagnostic(Kind)))
         // Unquoted translatable token name.
         Out << S;
-      else if (const char *S = tok::getTokenName(Kind))
+      else if ((S = tok::getTokenName(Kind)))
         // Debug name, shouldn't appear in user-facing diagnostics.
         Out << '<' << S << '>';
       else
         Out << "(null)";
       break;
     }
     // ---- NAMES and TYPES ----
     case DiagnosticsEngine::ak_identifierinfo: {