Using mapping framework with drastically different Domain and view-models ?

Question

If I have two models that are very different, is it defeating the purpose to even use a mapping framework such as Auto Mapper or Value Injecter?

For example, say I have two models that are structurally identical, but every property name is different, thus rending all by-convention logic useless. Is it just as well to stick to manually mapping these two models?

I guess I'm mainly trying to understand:

1) Is there any benefit to using a mapping framework aside from the code savings I gain from its by-convention mapping?

2) Is it still easier to set up mapping for the above scenario in a mapping framework vs. manually mapping in this case? (It doesn't seem like it.)

AutoMapper is intended to get rid of code you'd already write, by applying conventions. If there is no pattern of mapping, what's the point? More to the point though - how far off are these models? — Jimmy Bogard, Jun 28 '12 at 02:17

danludwig · Accepted Answer · 2012-07-05T03:54:43.183

1) Is there any benefit to using a mapping framework aside from the code savings I gain from its by-convention mapping?

I think so. It's one of those one-off kinds of things. Sure, you can write the code to manually DTO from one object to another, maybe about as fast as you can create all of the custom mapping rules. But the very second time you need to DTO, using the mapping framework pays for itself.

2) Is it still easier to set up mapping for the above scenario in a mapping framework vs. manually mapping in this case? (It doesn't seem like it.)

I think it's about the same:

Mapper.CreateMap<EntityModel, ViewModel()
    .ForMember(d => d.Prop1, o => o.MapFrom(s => s.Property1))
    .ForMember(d => d.Prop2, o => o.MapFrom(s => s.Property2))
    .ForMember(d => d.Prop3, o => o.MapFrom(s => s.Property3))
        ...
    .ForMember(d => d.PropN, o => o.MapFrom(s => s.PropertyN))
;

With this mapping, all of your DTO code will look like this:

var model = Mapper.Map<ViewModel>(entityInstance);

...versus... If you dit it manually, it looks about the same to me:

var model = new ViewModel
{
    Prop1 = entityInstance.Property1,
    Prop2 = entityInstance.Property2,
    Prop3 = entityInstance.Property3,
        ...
    PropN = entityInstance.PropertyN,
};

Another cool thing about AutoMapper over ValueInjecter is that you only need to define 1 mapping, which will then work for mapping both single instances and collections.

var models = Mapper.Map<ViewModel[]>(entityInstances);

... versus

var models = entityInstances.Select(new ViewModel
{
    Prop1 = entityInstance.Property1,
    Prop2 = entityInstance.Property2,
    Prop3 = entityInstance.Property3,
        ...
    PropN = entityInstance.PropertyN,
});

It will be easier to define manual mapping in implicit or explicit operators (conversion), won't be? — k0stya, Jun 27 '12 at 18:32

score 0 · Answer 2 · answered Jun 27 '12 at 18:19

0

For example, say I have two models that are structurally identical, but every property name is different,

It's you that are designing the view models => you have total control over them (contrary to your domain models which might be for example auto generated from a database or something by a designer) => you can name their properties accordingly :-)

answered Jun 27 '12 at 18:19

Darin Dimitrov

1,023,142
271
3,287
2,928

I agree, but sometimes with legacy code (domain models) you want to take the opportunity to clean things up a bit. Or, would you still make your new models match the poorly designed domain models in the interest of easier mapping? – Jerad Rose Jun 27 '12 at 18:25

Using mapping framework with drastically different Domain and view-models ?

2 Answers2