Remove omitempty from "w" and "h" of Video object to support passing 0x0 for dynamic ad size

[saifat29]

Many RTB exchanges expect width and height to be 0x0 for them to support dynamic ad sizes for Video ads. If we can remove the omitempty tag from w and h for only the Video object the zero values would begin to pass.

type Video struct {
    W int64 `json:"w,omitempty"`     // <-- Remove this omitempty
    H int64 `json:"h,omitempty"`     // <-- Remove this omitempty
}

Let me know your thoughts.

[saifat29]

I've also opened an issue on https://github.com/bsm/openrtb/issues/105

[VeronikaSolovei9]

Hello @saifat29, thank you for submitting your question. May you please provide a use case where this change is needed? Do you need to distinguish cases where W and H are 0s and nils? Do you need to unmarshal data to the struct other than openrtb?

[bsardo]

Hi @saifat29, just checking in to see if you can shed some light on use cases as requested by @VeronikaSolovei9. Thanks!

[saifat29]

Sorry for the delay.

The issue is that some demand partners make a distinction between the absence of values for w and h and when these values are set to zero. Their system reacts differently to the case where "w": 0, "h": 0 is present compared to when these fields are completely absent. As a result, they have asked us to provide values for these fields even though they are set to zero, but we are unable to do so because json.Marshal disregards these fields due to the omitempty tag.

Also I don't see anything in the IAB spec against it. Let me know your thoughts, if it's a broad enough problem we should solve it.

[SyntaxNode]

Understood. 👍

Discussed within Prebid and several other folks agree with you. We have a PR out now to address this issue. We reviewed the entire OpenRTB 2.x object model and changes a few other fields which have the potential for nil to be interpreted differently from a zero value.

[SyntaxNode]

Fixed in v20.0.0.