Question: Why is Update test not transactional?

[robfig]

Out of curiosity, why is the Update test not required to be transactional? It seems like that makes it much less realistic -- I doubt that anyone is doing read-updates outside of a transaction in real life. And exercising the framework's transaction support seems like a good thing.

[bhauer]

Hi @robfig. That's a great idea. I've added "Transactional update test" as a new test type on #133.

Rather than change test 5, I'd prefer to test transactions as a new kind of test. I'd welcome a draft mini-spec for the test. :)

[robfig]

Heh, not sure I have any ideas for it other than what the Update test is already. Do you have any thoughts?

[bhauer]

@robfig Nothing in particular comes to mind, but I always like asking. Some really good ideas have come from the community. :)

[bhauer]

I'm closing this since I have this captured as a new test type for the future.