CodeFormatter should accept SyntaxTree

[belav]

There are cases where someone has a SourceText and being forced to convert it to a string is an extra step that shouldn't be required. CodeFormatter could pretty easily accept SourceText as well.

[TwentyFourMinutes]

I actually noticed that a SourceText is already a finalized visual form of the SyntaxTree, so it therefore really should instead have an overload accepting the actual SyntaxTree, what do you think?

[belav]

Yeah that makes sense. Maybe it has an overload for both of them?

[TwentyFourMinutes]

I don't see any obvious way to convert a SourceText to a SyntaxTree without converting it to a string and back. Therefore the consumer should really be aware what's happening IMHO.

[belav]

CSharpSyntaxTree.ParseText has an overload that accepts a SourceText that you may have missed.

[TwentyFourMinutes]

Oh I think that we are both talking about different things. When I am creating a source code through the SyntaxFactory API I am currently first converting the SyntaxTree > SourceText > string. Now from my perspective it would make the most sense if I could directly pass the CodeFormatter the SyntaxTree. Not because it is necessarily less code to write, but rather because it is a lot faster, as you are converting the passed string/SourceText back to a SyntaxTree as far as I am aware of.

[belav]

Yeah, I was thinking CodeFormatter would have a few overloads

public static string Format(string code, CodeFormatterOptions? options = null)

public static string Format(SourceText code, CodeFormatterOptions? options = null)

public static string Format(SyntaxTree code, CodeFormatterOptions? options = null)

I assumed you were looking at how to implement those overloads in csharpier and didn't know about that CSharpSyntaxTree.ParseText had an overload for SourceText

[TwentyFourMinutes]

Oh I see 😄. Well that would make sense however do you think that it should be encourage to pass it a SourceText? CSharpSyntaxTree.ParseText will more or less convert it to a string anyway and the performance hit would just be hidden. Also I can't think of any quick scenario in my head where I'd have the SourceText, but not the SyntaxTree representation of a piece of code.

[belav]

I am good with just having the overload for SyntaxTree, no need to add another overload if no one is going to need it.