Small Ping. @rsmith - did you have anymore thoughts about this patch?

• Remove the call to setImplicitlyInline() within setImplicitlyConstexpr() and call setImplicitlyInline() directly for function concepts.

Ping

• Initialize IsConstexprSpecified
• Remove unnecessary parenthetical in comment
• Fix non-portable test

• Address Aaron's comments by putting comments next to the magic numbers

• Address Hubert's comments about the quoted section of the TS.

• This update removes the parameter in TemplateDecl::setConcept since there isn't a need to mark a concept false.

Marking some comments Done which were fixed in previous updates.

• Fix a couple of comments to reflect the Patch.
• Clang-format the changes in this Patch.

@rsmith - would you also mind clarifying the comment regarding setConcept(bool IC) at to whether it should exist at all or simply not have any params?

• Take qualifiers into account.
• Check VarDecl is valid when checking declaration type to account for constexpr being specified. Is there any opinion on a better way to handle this? I *could* check for both diagnostics in the test...
• Remove erroneous check for auto (containsPlaceholderType) and associated function concept test.
• Removed the use of Diagnostic Range because we'd need to enumerate over the various qualifiers.

Please let me know if the subject or summary of this Patch is ambiguous.

In D15421#326144, @rsmith wrote:

Bikeshedding on the name a bit... how about __type_pack_element?

Ping. Now that the holidays are over-ish, as Aaron said in one of his Patches.

• Store the IsConcept boolean flag in TemplateDecl by making TemplatedDecl an IntPointerPair, and move the associated member functions into TemplateDecl.
• Remove unnecessary quoted comment.
• Remove an extra space where the diagnostic is used.

Updating to r256247

• Move IsConcept bit and associated member functions from FunctionDecl to FunctionTemplateDecl.
• Set the IsConcept flag using getDescribedFunctionTemplate.

Hi Louis,

Updating to r254337

• Remove marking a variable concept invalid when specialized since we'll only look at the primary template downstream. This removal let's us use the same recovery path as before when 'concept' is specified on a non-template.

• replace hyphen and apostrophe with equivalent ASCII characters

• Cover variadic arguments in check for no params.
• Add tests to cover variadic arguments.

Updating to r251898

Addressing Richard's other comment regarding the check that the FunctionDecl is not a concept. Also, not marking the specialization as invalid if it's declared with the concept specifier since the downstream parts of Clang should look at the primary template.

Moving tests to the correct file.
Modifying the diagnostic id and message so it's more applicable to the checks in this Patch. Update the quoted sections of the standard as well.
Use ternary operator for Partial Specialization check.
Add missing period in comment.

Ping

Change the wording of diagnostic message and quote the invalid specifier. Add 'friend' as an invalid specifier and diagnose function concepts when 'friend' is specified. Reorder the list of invalid specifiers and corresponding checks in the code to match the TS.

Ping

Patch addressing comments; fix comment/documentation wording, scoping of diagnostic and setting invalid declaration, and fix test.

Ping

Addressing comments which were discussed on the mailing list - Apply the same text when diagnosing a free standing declaration as suggested by Aaron. Replace diagnostic identifier err_concept_decl_non_template with err_concept_wrong_decl_kind as suggested by Hubert and Richard. Fix tests corresponding to the changes.

Hi Piotr,

• Update Patch based on comments; rephrase exception spec comment, adjust nesting of check for FunctionTypeInfo, remove SourceRange check, fix test

Updated Patch based on comments; fix diagnostic spacing and phrasing, add getExceptionSpecRange, check FunctionTypeInfo exists, use PartialDiagnostic, add static_assert test for function concept being treated as noexcept(true)

I'll make the fixes based on the comments, but had question about the FunctionTypeInfo in the meantime.

Putting diagnostic on one line.

Removing failing test case (in Parser) created by this Patch

Thanks Richard.

If deemed necessary, I can split this patch out to two patches; one for adding IsConcept to the NonParmVarDeclBitfields bitfield and associated VarDecl member functions and another patch for the Diagnostic/tests

modifying diagnostic message to: 'concept' can only appear on the definition of a function template or variable template