Just pick a model and run the tests. It’s easy. Find something where divergences go away with the old code as you increase adapt_delta and repeat the exercise with the new code.
You’re telling me it shouldn’t be a problem. I believe you (and I believed you before I asked for these tests). I am only asking you to double check this. I don’t even care what the model is as long as it meets the specification above.
There are a fair number of reasons I think it’s fine that I ask for these tests. If anything the testing process I’ve put this through has been lenient. The e-mail you sent me on October 7th seems to suggest you disagree. But aside from that, as someone reviewing your pull requests, I want to see more self-reflection in the pulls you are making and less combativeness of the reviews. I also want you to respect the opinions of people other than yourself on these issues.
I don’t like that you are resisting review. Your uturn pull request from a couple months ago actually would have broken Stan: https://github.com/stan-dev/stan/pull/2800#issuecomment-525889921. And it’s okay that there was a bug. We all screw stuff up. What is important is we were able to work together and find it (those were your tests I was running, too, so I’m not taking credit here). Getting immediate pushback on some incredibly simple tests that we all expect should pass and are easy to do is really frustrating given this close history.
It is not fun when you’re doing a careful review to find problems and be told that they aren’t problems: https://github.com/stan-dev/stan/pull/2800#discussion_r314950803. I’m trying to help this stuff go smoothly and make sure there aren’t mistakes that come back to bite us later.
If you think there’s something wrong with this process, I’m happy to have this go before the SGB. If they tell me I’m being unfair in asking you to do tests, that’s fine, I’ll respect that. Again, the tests are easy, and I fully expect them to pass.
Also, is this code in a repo anywhere? This is the pull that I should be watching, right: https://github.com/stan-dev/stan/pull/2790 ? The last that was updated was August 8th and I think the Uturn stuff got finished on the 30th so I think it’s out of date.