Partial properties: attributes

[RikkiGibson]

Test plan: #73090
Related spec change: dotnet/csharplang#8127

[RikkiGibson]

Compare with SourceMethodSymbolWithAttributes.GetAttributesBag

[RikkiGibson]

@jcouv @333fred for review

[jcouv]

        SyntaxList<AttributeListSyntax> indexerNameAttributeLists,

Not related to this PR: it seems the only use we have for indexerNameAttributeLists is to check whether it is empty or not. We should consider passing a boolean flag instead of forcing some callers to allocate empty lists...

---

Refers to: src/Compilers/CSharp/Portable/Symbols/Source/SourcePropertySymbolBase.cs:92 in 3ce6777. [](commit_id = 3ce6777, deletion_comment = False)

[jcouv]

Done with review pass (iteration 3)

[RikkiGibson]

    /// Returns the definition part of a partial method implementation, 

Not related to this PR: is this comment correct? (and the one on SourcePartialDefinition above)

Refers to: src/Compilers/CSharp/Portable/Symbols/Source/SourceOrdinaryMethodSymbol.cs:351 in 3ce6777. [](commit_id = 3ce6777, deletion_comment = False)

For those outside of CodeFlow, context is:

roslyn/src/Compilers/CSharp/Portable/Symbols/Source/SourceOrdinaryMethodSymbol.cs

Lines 338 to 360 in e5d344a

	/// <summary>
	/// Returns the implementation part of a partial method definition,
	/// or null if this is not a partial method or it is the definition part.
	/// </summary>
	internal SourceOrdinaryMethodSymbol SourcePartialDefinition
	{
	get
	{
	return this.IsPartialImplementation ? this.OtherPartOfPartial : null;
	}
	}
	/// <summary>
	/// Returns the definition part of a partial method implementation,
	/// or null if this is not a partial method or it is the implementation part.
	/// </summary>
	internal SourceOrdinaryMethodSymbol SourcePartialImplementation
	{
	get
	{
	return this.IsPartialDefinition ? this.OtherPartOfPartial : null;
	}
	}

These comments seem wrong, it feels most reasonable to me to just delete them.

[jcouv]

LGTM Thanks (iteration 13)

[RikkiGibson]

@333fred for second review

[RikkiGibson]

179203f is all green and the next commit b190153 is comment only changes, so I will go ahead and force merge for expediency.

[RikkiGibson]

@jcouv feel free to follow up in this PR if you had more comments on the more recent changes, I can be sure to address in subsequent PR.

[Sergio0694]

Hey @RikkiGibson, couldn't find this specific bit in the spec, is this planned to work?

// User code
[field: Foo]
public partial string Name { get; set; }

// Generated.g.cs
private string _name;

public string Name { get => _name; set => _name = value; }

Or would this not work because Roslyn wouldn't be able to know what backing field the partial property implementation would be using? IIRC we discussed allowing [field:] on partial properties before, but it's not clear to me whether that was only expected to work if the partial property implementation would be using the field keyword (which won't be in C# 13 anyway). Just making sure I'm following so that I can also plan what I'll need to support and/or work around in my generators. Thank you! 😄

[RikkiGibson]

Or would this not work because Roslyn wouldn't be able to know what backing field the partial property implementation would be using?

That's right.

It's possible that field keyword in properties will also land this release, which might make your generator's job easier, but otherwise, you'll be looking at suppressing the compiler warning about a bad attribute target, and "forwarding" the attribute to the field, as I think you've done before from a field to a property.