Tracking model decoding time

[zKoz210]

Problem Statement

The biggest problem Sentry doesn't show is model decoding.

For example, I have a query with a condition that returns 250 models where I need to get a couple of fields. I'll see a 2-5ms request and sit happy, but if I start measuring the application's performance, I'll find that I'm wasting about 50ms when the DB facade could have been used at that point

Solution Brainstorm

I can't suggest how this could be implemented properly

[cleptric]

Could clarify what you mean by Sentry "doesn't show" model decoding?