Origin Trial - ability to report errors, timeouts

[jonasz]

Hi,

One important goal of the upcoming Origin Trial will be to understand how generateBid works on the large variety of devices it will be invoked on. This includes:

• how often does generateBid time out?
• how often does it crash? (i.e: raise an exception)

With the current FLEDGE spec it seems that if generateBid crashes or times out, the buyer will never learn about that. (So for example it will be hard to tune perBuyerTimeouts, tune the amount of computations, or discover and fix bugs that cause exceptions.)

In the future this kind of functionality will likely be fulfilled by reportLoss and aggregate reporting (relevant issue: #164).

I was wondering, do you have any thoughts on how such observability / debugging could be supported in the Origin Trial? Perhaps we could consider a basic support for event-level reportLoss with limited inputs, analogous to the event-level reportWin?

I'd be curious to hear your thoughts!

Best regards,
Jonasz