GetQueryAsync / ProjectTo not working with "UseInMemoryDatabase" db context

[Edgaras91]

Is this a known issue, as I've seen mentions in stack overflow?

In tests (unit) when creating WebApplicationFactory and overriding DbContexts to be in memory, the GetQueryAsync stops working.

Source/destination types

                //When using in-memory database only:
                var hasResults = await _accountsDbContext.AccountAddresses.ToListAsync(cancellationToken: cancellationToken);
                var noResults0 = await _accountsDbContext.AccountAddresses.ProjectTo<AccountAddressDto>(_mapper.ConfigurationProvider).ToListAsync(cancellationToken: cancellationToken);
                var noResults1 = await (await _accountsDbContext.AccountAddresses.GetQueryAsync(_mapper, request.QueryOptions)).ToListAsync(cancellationToken: cancellationToken);
                var noResults2 = await _accountsDbContext.AccountAddresses.GetQueryAsync(_mapper, request.QueryOptions);

//SETUP:

 var accountsDescriptor = serviceCollection.Single(
                            d => d.ServiceType ==
                                 typeof(DbContextOptions<AccountsDbContext>));

                        serviceCollection.Remove(accountsDescriptor);

                        serviceCollection.AddDbContext<AccountsDbContext>(options =>
                            options.UseInMemoryDatabase("AccountsDB"));

Mapping configuration

     public void Mapping(Profile profile)
        {
            profile.CreateMap<AccountAddress, AccountAddressDto>()
                .ForMember(d => d.Title, s => s.MapFrom(x => x.Title.TitleName));
        }

Version: x.y.z

ASP NET6 AutoMapper.AspNetCore.OData.EFCore 4.0.0 Microsoft.EntityFrameworkCore.SqlServer 6.0.13

May be related to this: https://stackoverflow.com/questions/60837497/automapper-projectto-not-working-with-efcore-in-memory-database-unit-tests

[Edgaras91]

Ignore, this was not working on newly added Integration tests, because child tables were not "seeded", so the auto mapper profile could not map to child data, whereas using a real database, already had "good" data with all of it's child tables populated.