Refactor get_crop to always lookup the thumb from the database.

[ctbarna]

No description provided.

[earthboundkid]

We now have two deprecated kwargs. Maybe we should write a new tag?

[ctbarna]

That does the same thing as the old tag?

[earthboundkid]

I think that dropping two kwargs is a sign that we need to think about what went wrong with the design of this tag.

[earthboundkid]

For example, maybe a lot of the work the tag is doing should be in the model instead.

[ctbarna]

We can remove the deprecation warnings if it makes you feel better. Passing size and exact size in have no effect besides to warn that they don't do anything.

[ctbarna]

Unless you write some magic to pass the arguments through the relation, model logic won't work in the template.

[whatisjasongoldstein]

Well, it predates a lot of cropduster/generic-pluses functionality (prefetch etc). It makes sense than we'd refactor it to work with the updates.

The hard part about moving it onto the field is we'd have to get rid of crops with -'s in the names, or automatically infer they're the same as _, since we could do something like article.lead_image.lead_large in the template, but not article.lead_image.lead-large, iirc.

[earthboundkid]

I don't like article.lead_image.lead_large, but we could make it so the tag is just a simple call to a method with a cropname (which we wouldn't need in Jinja).

[ctbarna]

That's what this is…

[earthboundkid]

It's still calling related_image, which is a model detail that breaks the Law of Demeter.

[whatisjasongoldstein]

Before this is deployed we should grep for _raw and see check/fix queries on pages that predated pretch_related working with generic plus.

[fdintino]

@ctbarna similar question for this one. This is basically abandoned, right?