Support adapters to set seat for a bid #2257
Conversation
8bf979b to
8cd19b3
Compare
| @@ -667,12 +675,12 @@ func (e *exchange) buildBidResponse(ctx context.Context, liveAdapters []openrtb_ | |||
| // Create the SeatBids. We use a zero sized slice so that we can append non-zero seat bids, and not include seatBid | |||
| // objects for seatBids without any bids. Preallocate the max possible size to avoid reallocating the array as we go. | |||
| seatBids := make([]openrtb2.SeatBid, 0, len(liveAdapters)) | |||
| for _, a := range liveAdapters { | |||
There was a problem hiding this comment.
Cannot loop over liveAdapters as now, getAllBids() could bring extra bidders
exchange/exchange.go
Outdated
| // Do we need to address duplicate names here?. What if different BidderNames have exta bids with same seat value. | ||
| // Ex. 'pubmatic' adapter has extra bids under new 'groupm' seat and 'appnexus' also comes with additional bids under seat 'groupm' | ||
| // Above example might be more relatable with 'allowUnknownBidderCodes' | ||
| adapterBids[openrtb_ext.BidderName(seatBid.seat)] = seatBid |
There was a problem hiding this comment.
Do we need to address duplicate names here?
There was a problem hiding this comment.
Do we clubs the bids in case multiple bidders returns the same alternate bidder code?. Ex. appnexus and Pubmatic returning groupm as a new bidder
@bretg thoughts?
There was a problem hiding this comment.
updated code to append the bids
|
Implemented validation in a separate PR PubMatic-OpenWrap#309 against this branch |
|
I think we don't want to merge and release this without phase 2, as that will give PBS hosts some control over what adapters are allowed to do. Or perhaps include an off switch will prevent adapters from being able to rewrite the seat in seat bids. I can see a potential issue that downstream consumers could be caught off guard on assuming a one to one relation between seat and bidder, and this breaking some logic that depends on that. As an example, Pub A is using Pubmatic and thus has a relation to them. It is not guaranteed that Pub A has a separate relationship with GroupM. Pub A may have some client logic to generate billing reports based on the seat names of the seat bid. This would result in billing reports for "GroupM" that Pub A does not know what to do with, since they don't have a direct relation there. Pub A can get the information from the adaptercode meta, but that will require them retooling their scripts to account for this new definition of seat. We can't require all publishers to retool their script in lock step with a new PBS release. |
…-bid-1-validation-1 Validate extra-bidder prebid#2257 prebid#2174
Merged phase-2 implementation (PubMatic-OpenWrap#309) in this branch. Will address all the comments here :) |
|
Raised a new PR as this one is from wrong account. New PR: #2266 Apologies for the inconvenience. |
Implements #2174
seatbid.bid.ext.prebid.meta.adaptercodewith the adapter code.seatof bid(s).seatfor bids with value frombid.ext.marketplace.seatfor all adapters.seatandadaptercode.seat.adaptercodeunder newseatof an alias bidder.requestBid()changes.pubmaticadapter test to assert extra-bid and a new response ext fieldmarketplace.