Lmt to be a pointer

[evgenigourvitch]

to allow to pass 0 values when required.

[mxmCherry]

Mmm, does it make sense to have this as tri-state (null/absent, 0, 1)?
It is recommended in spec.
And "absence" treatment is not described.
And I cannot think that absence would ever be treated by anyone in any special way.
Probably, there's no need for it to be able to become absent?

Thinking about removing the omitempty modifier (so always pass 0 or 1), just not sure if/how it may affect consumers.

Though, it's not "required", just "recommended"...

[mxmCherry]

@evgenigourvitch could you, please, check if you need the same for DNT attr (the one above Lmt)?
If yes - let's do it in one go.
If not - just leave it (don't change what's not needed/requested).

I think that it makes sense to adhere to such rules:

• "required" in spec -> no "omitempty"
• not "required" in spec - "omitempty" and make it pointer if zero value has some special meaning

Just to make such decisions simple in future.

[evgenigourvitch]

Regarding the DNT - currently it's fine for me, but "tomorrow" it could be changed and I'll need to send 0 value, so lets do it in one go, just like you said. Changing it.

IMHO, the omitempty modifier is just great! It allows to save traffic costs very easily and believe in LARGE scale its worth it - every byte counts.

I think pointer is a nice solution for tri-state values null/0/1. Absence (null) will easily allow to save traffic costs in most of the cases. And 0/1 will be easily passed when needed.
Sounds reasonable?

[mxmCherry]

Will merge / update README with "omitempty" usage hints / make a new major ver release today in ~ 8-9 hours.

[evgenigourvitch]

@mxmCherry thanks a lot!

[mxmCherry]

@evgenigourvitch released as https://github.com/mxmCherry/openrtb/releases/tag/v11.0.0 , thanks for contributing!