Can I do an IQueryable projection from a string[] column in postgres to a protobuf message?

[x2d7751347m]

Feature intro: https://mapperly.riok.app/docs/configuration/queryable-projections/

public class BaseSomething
{
    public BaseSomething(
        string[] someTypes)
    {
        SomeTypes = someTypes;
    }

    public string[] SomeTypes { get; set; }
}

    public static partial IQueryable<BaseSomethingMessage> ProjectToDto(this IQueryable<BaseSomething> q);

message BaseSomethingMessage {
  repeated string SomeTypes = 1;
}

The generated MESSAGE C# class has the following types:

public interface IMessage<T>
  : IMessage, IEquatable<T>, IDeepCloneable<T>
 in namespace Google.Protobuf
 from package Google.Protobuf v3.31.0

MyProject.csproj

    <ItemGroup>
        <PackageReference Include="Google.Protobuf" Version="3.31.0"/>
        <PackageReference Include="Grpc.AspNetCore" Version="2.71.0"/>
        <PackageReference Include="Grpc.Tools" Version="2.72.0">
            <PrivateAssets>all</PrivateAssets>
            <IncludeAssets>runtime; build; native; contentfiles; analyzers; buildtransitive</IncludeAssets>
        </PackageReference>

        <Protobuf Include="Proto\BaseSomethingMessage.proto">
            <Access>Public</Access>
            <AdditionalImportDirs>.\Protos</AdditionalImportDirs>
            <GrpcServices>Both</GrpcServices>
            <ProtoCompile>True</ProtoCompile>
            <CompileOutputs>True</CompileOutputs>
            <OutputDir>obj\Debug\net9.0\</OutputDir>
            <Generator>MSBuild:Compile</Generator>
        </Protobuf>
    </ItemGroup>

I tried this mapping, but it didn't auto-generate correctly, and I tried the following manually, which was not possible.

/// Cannot be created correctly.    
    public static partial IQueryable<BaseSomethingMessage> ProjectToDto(this IQueryable<BaseSomething> q)

/// System.InvalidOperationException: EF Core does not support MemberListBinding: 'new Blog { Posts = { new Post(), new Post() } }'.
 /// at Microsoft.EntityFrameworkCore.Query.Internal.ExpressionTreeFuncletizer.VisitMemberListBinding(MemberListBinding node)

    public static IQueryable<BaseSomethingMessage> ProjectToDto(this IQueryable<BaseSomething> q)
    {
#nullable disable
        return global::System.Linq.Queryable.Select(
            q,
            x => new global::SomeMessagesProto.Generated.BaseSomethingMessage()
            {
                SomeTypes = { x.SomeTypes },
            }
        );
#nullable enable
    }

/// Cannot convert source type 'string[]' to target type 'Google.Protobuf.Collections.RepeatedField<string>'
                BusinessTypes = x.BusinessTypes,

If there is no way, should I raise an issue? Or has it already been reviewed and deemed difficult?

[latonz]

I don‘t think EF core supports projecting to a protobuf repeated field. Does it work if you write the select manually instead of using Mapperly? If so, could you provide the code?

[x2d7751347m]

/// System.InvalidOperationException: EF Core does not support MemberListBinding: 'new Blog { Posts = { new Post(), new Post() } }'.
 /// at Microsoft.EntityFrameworkCore.Query.Internal.ExpressionTreeFuncletizer.VisitMemberListBinding(MemberListBinding node)

    public static IQueryable<BaseSomethingMessage> ProjectToDto(this IQueryable<BaseSomething> q)
    {
#nullable disable
        return global::System.Linq.Queryable.Select(
            q,
            x => new global::SomeMessagesProto.Generated.BaseSomethingMessage()
            {
                SomeTypes = { x.SomeTypes },
            }
        );
#nullable enable
    }

This is a function I built myself. It runs before ToArray, ToList, etc., so the DB side throws an error in the comment when it runs those methods. Of course, the function in the application runs without problems after ToArray, etc.

System.InvalidOperationException: EF Core does not support MemberListBinding: 'new Blog { Posts = { new Post(), new Post() } }'.
   at Microsoft.EntityFrameworkCore.Query.Internal.ExpressionTreeFuncletizer.VisitMemberListBinding(MemberListBinding node)
   at System.Linq.Expressions.ExpressionVisitor.VisitMemberBinding(MemberBinding node)
   at Microsoft.EntityFrameworkCore.Query.Internal.ExpressionTreeFuncletizer.Visit[T](ReadOnlyCollection`1 expressions, Func`2 elementVisitor, StateType& aggregateStateType, State[]& expressionStates, Boolean poolExpressionStates)
   at Microsoft.EntityFrameworkCore.Query.Internal.ExpressionTreeFuncletizer.VisitMemberInit(MemberInitExpression memberInit)
   at Microsoft.EntityFrameworkCore.Query.Internal.ExpressionTreeFuncletizer.Visit(Expression expression, State& state)
   at Microsoft.EntityFrameworkCore.Query.Internal.ExpressionTreeFuncletizer.VisitLambda[T](Expression`1 lambda)
   at Microsoft.EntityFrameworkCore.Query.Internal.ExpressionTreeFuncletizer.Visit(Expression expression, State& state)
   at Microsoft.EntityFrameworkCore.Query.Internal.ExpressionTreeFuncletizer.VisitUnary(UnaryExpression unary)
   at Microsoft.EntityFrameworkCore.Query.Internal.ExpressionTreeFuncletizer.Visit(Expression expression)
   at Microsoft.EntityFrameworkCore.Query.Internal.ExpressionTreeFuncletizer.Visit[T](ReadOnlyCollection`1 expressions, Func`2 elementVisitor, StateType& aggregateStateType, State[]& expressionStates, Boolean poolExpressionStates)
   at Microsoft.EntityFrameworkCore.Query.Internal.ExpressionTreeFuncletizer.Visit(ReadOnlyCollection`1 expressions, StateType& aggregateStateType, State[]& expressionStates, Boolean poolExpressionStates)
   at Microsoft.EntityFrameworkCore.Query.Internal.ExpressionTreeFuncletizer.VisitMethodCall(MethodCallExpression methodCall)
   at Microsoft.EntityFrameworkCore.Query.Internal.ExpressionTreeFuncletizer.Visit(Expression expression, State& state)
   at Microsoft.EntityFrameworkCore.Query.Internal.ExpressionTreeFuncletizer.ExtractParameters(Expression expression, IParameterValues parameterValues, Boolean parameterize, Boolean clearParameterizedValues, Boolean precompiledQuery, IReadOnlySet`1& nonNullableReferenceTypeParameters)
   at Microsoft.EntityFrameworkCore.Query.Internal.ExpressionTreeFuncletizer.ExtractParameters(Expression expression, IParameterValues parameterValues, Boolean parameterize, Boolean clearParameterizedValues)
   at Microsoft.EntityFrameworkCore.Query.Internal.QueryCompiler.ExtractParameters(Expression query, IParameterValues parameterValues, IDiagnosticsLogger`1 logger, Boolean compiledQuery, Boolean generateContextAccessors)
   at Microsoft.EntityFrameworkCore.Query.Internal.QueryCompiler.ExecuteCore[TResult](Expression query, Boolean async, CancellationToken cancellationToken)
   at Microsoft.EntityFrameworkCore.Query.Internal.QueryCompiler.ExecuteAsync[TResult](Expression query, CancellationToken cancellationToken)
   at Microsoft.EntityFrameworkCore.Query.Internal.EntityQueryProvider.ExecuteAsync[TResult](Expression expression, CancellationToken cancellationToken)
   at Microsoft.EntityFrameworkCore.Query.Internal.EntityQueryable`1.GetAsyncEnumerator(CancellationToken cancellationToken)
   at System.Runtime.CompilerServices.ConfiguredCancelableAsyncEnumerable`1.GetAsyncEnumerator()
   at Microsoft.EntityFrameworkCore.EntityFrameworkQueryableExtensions.ToListAsync[TSource](IQueryable`1 source, CancellationToken cancellationToken)
   at Microsoft.EntityFrameworkCore.EntityFrameworkQueryableExtensions.ToArrayAsync[TSource](IQueryable`1 source, CancellationToken cancellationToken)

If I map it to a plain string[] rather than an IMessage from Google.Protobuf like this, it works fine. Would it still be difficult to query it into my application and process it with Select before importing?

            var temp = await filteredQuery.Select(
                    x => new BaseSomething()
                    {
                        BusinessTypes = x.BusinessTypes,
                    }
                ) // Query, DB Side
                .ToArrayAsync(ct);

[latonz]

If your custom written code doesn‘t work, unfortunately there is no way for Mapperly to handle this case.

[x2d7751347m]

I didn't think it would work, but I thought there was another way to do it, but there wasn't. Thanks for the answer!

[mansimov]

Model ladder update — explicit checkpoints pinned

Per @keunwoo-ortet's K5 clarification request, here is the explicit model ladder with pinned checkpoints for #9798. All entries are post-trained (instruct/chat) variants unless noted otherwise.

1. Smoke test — small, fast iteration

• Model: Qwen/Qwen3-4B-Instruct-2507
• HF checkpoint SHA: cdbee75f17c01a7cc42f958dc650907174af0554
• Role: First end-to-end validation on Prime RL (alphabet-sort / reverse-text configs). Proves tokenizer/template parity, rollout→learner log-prob parity, weight broadcast, checkpoint save/restore, and basic NCCL sync at 1–2 node scale before any large model loads.
• Why: Already present in Prime RL basic configs (configs/basic/alphabet-sort/rl.toml), minimal download time, fast failure signal.

2. Small MoE — framework stress at moderate scale

• Model: Qwen/Qwen3-30B-A3B-Instruct-2507
• Role: Prime RL's built-in B200 baseline model. Proves multi-node trainer/inference topology, EP/CP layouts, and scaling ladder behavior before committing to 100B+ checkpoints.
• Note: Added as intermediate step because it is the only model with upstream B200 benchmark rows we can sanity-check against.

3. RL target candidate — large MoE

• Model: deepseek-ai/DeepSeek-V4-Flash (post-trained, not DeepSeek-V4-Flash-Base)
• HF checkpoint SHA: 60d8d70770c6776ff598c94bb586a859a38244f1
• Semantics: Post-trained / instruction-tuned. 284B total / 13B active parameters.
• Role: Primary RL scaling candidate once Prime RL proves model loading, rollout/learner log-prob parity, weight broadcast, checkpoint save, and fresh restore at smaller scales.

4. Alternative / comparison candidate

• Model: thinkingmachines/Inkling-Small
• HF checkpoint SHA: 8cc5877b44d343f88b92086aa1fb72897950f06a
• Role: Secondary candidate for comparison or fallback if DeepSeek-V4-Flash hits framework compatibility issues.

Notes

• model_semantics: posttrained applies to all entries; no base checkpoints are required for this benchmark lane per Keunwoo's K6 correction.
• Exact tokenizer/template, precision, context limits, and immutable checksums will be pinned in the formal run card before promotion.
• Will coordinate here before checkpoint prep to avoid duplicating work if @keunwoo-ortet prepares a compatible artifact first.