Doesn’t matter, as long as it’s a static class. It’s all about convention.
Our convention is that each “layer” (web, services, data) has a single file called AutoMapperXConfiguration.cs
, with a single method called Configure()
, where X
is the layer.
The Configure()
method then calls private
methods for each area.
Here’s an example of our web tier config:
public static class AutoMapperWebConfiguration
{
public static void Configure()
{
ConfigureUserMapping();
ConfigurePostMapping();
}
private static void ConfigureUserMapping()
{
Mapper.CreateMap<User,UserViewModel>();
}
// ... etc
}
We create a method for each “aggregate” (User, Post), so things are separated nicely.
Then your Global.asax
:
AutoMapperWebConfiguration.Configure();
AutoMapperServicesConfiguration.Configure();
AutoMapperDomainConfiguration.Configure();
// etc
It’s kind of like an “interface of words” – can’t enforce it, but you expect it, so you can code (and refactor) if necessary.
EDIT:
Just thought I’d mention that I now use AutoMapper profiles, so the above example becomes:
public static class AutoMapperWebConfiguration
{
public static void Configure()
{
Mapper.Initialize(cfg =>
{
cfg.AddProfile(new UserProfile());
cfg.AddProfile(new PostProfile());
});
}
}
public class UserProfile : Profile
{
protected override void Configure()
{
Mapper.CreateMap<User,UserViewModel>();
}
}
Much cleaner/more robust.