---

This section details on the original issue you should resolve

<issue_title>Undocumented / incorrect requirement for import directives to be alphabetically sorted to trigger IDE0055</issue_title>
<issue_description>Version Used:

.NET SDK Version: 9.0.201
SDK-Style Project
Microsoft Windows 11 Business 10.0.22631
Microsoft Visual Studio Enterprise 2022 (64-bit) - 17.13.5

Steps to Reproduce:

1. Have an SDK-Style C# project
2. Set dotnet_separate_import_directive_groups = true and dotnet_diagnostic.IDE0055.severity = error in .editorconfig
3. Add several usings in alphabetical order to a .cs file without separating line breaks
4. Observe an IDE0055 formatting rule violation is triggered
5. Move one of the directives such that they are no longer all alphabetically sorted
6. Observe that the IDE0055 formatting rule violation is no longer triggered

# Triggers IDE0055
namespace ktsu.IDE0055Test;

using Azure.Storage.Blobs;
using Azure.Storage.Sas;
using NuGet.Versioning;
using System.Diagnostics;

# Does not trigger IDE0055
namespace ktsu.IDE0055Test;

using Azure.Storage.Blobs;
using Azure.Storage.Sas;
using System.Diagnostics;
using NuGet.Versioning;

There is a comment in the analyzer rule that alludes to this as if it were intended behavior:

roslyn/src/Workspaces/SharedUtilitiesAndExtensions/Compiler/CSharp/Formatting/Rules/TokenBasedFormattingRule.cs

Line 225 in a35ddc3

private AdjustNewLinesOperation AdjustNewLinesAfterSemicolonToken(

        // if the user is separating using-groups, and we're between two usings, and these
        // usings *should* be separated, then do so (if the usings were already properly
        // sorted).

However the documentation for configuring the rule makes no mention of this requirement:

https://learn.microsoft.com/en-us/dotnet/fundamentals/code-analysis/style-rules/dotnet-formatting-options#dotnet_separate_import_directive_groups

Diagnostic Id: IDE0055: Fix formatting

Expected Behavior:

UsingsAndExternAliasesOrganizer.NeedsGrouping() indicates that two namespaces should belong to the same grouping if the first token of each is different.

roslyn/src/Workspaces/SharedUtilitiesAndExtensions/Compiler/CSharp/Utilities/UsingsAndExternAliasesOrganizer.cs

Line 66 in a35ddc3

public static bool NeedsGrouping(

The rule should not enforce alphabetical sorting, or the requirement should be documented, and if the sorting is intentional it should properly respect dotnet_sort_system_directives_first instead of allowing either completely sorted or sorted with System namespaces first.

In summary the rule should:

• Not query for alphabetical sorting at all

or

• Query for alphabetical sorting in a way that respects the dotnet_sort_system_directives_first configuration, and be documented as such

Actual Behavior:

TokenBasedFormattingRule.AdjustNewLinesAfterSemicolonToken() will only enforce separating newlines between grouped usings if all of the usings are sorted either alphabetically, or alphabetically with the System group first, regardless of the configuration of dotnet_sort_system_directives_first.

roslyn/src/Workspaces/SharedUtilitiesAndExtensions/Compiler/CSharp/Formatting/Rules/TokenBasedFormattingRule.cs

Line 225 in a35ddc3

private AdjustNewLinesOperation AdjustNewLinesAfterSemicolonToken(

            if (usings.IsSorted(UsingsAndExternAliasesDirectiveComparer.SystemFirstInstance) ||
                usings.IsSorted(UsingsAndExternAliasesDirectiveComparer.NormalInstance))

This check also does not match the documentation for dotnet_sort_system_directives_first in that it states:

Sort System.* using directives alphabetically, and place them before other using directives.

But actually checks for:

Sort System.* using directives alphabetically, and place them before other alphabetically sorted using directives.

</issue_description>

<agent_instructions>@copilot we should respect dotnet_separate_import_directive_groups if the usings are in groups, even if the groups are not sorted. </agent_instructions>

Comments on the Issue (you are @copilot in this section)

@JoeRobich I think having a separate analyzer/codefix for organized imports would make sense.