One of the tests in rust-html2text now fails with a stack overflow on beta/nightly in debug mode, but passes on the current stable (1.39). It also passes on beta/nightly with
--release, so I'm not too worried, but I'd quite like to understand it. (Travis report, but verified locally: https://travis-ci.org/jugglerchris/rust-html2text/builds/620857389)
The test (test_deeply_nested_table) simply generates a very deeply nested HTML fragment to test for unbounded recursion. In that sense it's succeeded in finding some. But it has been passing in the past. It's not just a marginal failure - I tried doubling the nesting depth without it failing on stable.
So for some reason since 1.39 there's some change - I guess either one or more stack frames are now significantly larger in non-release mode, or the thread stack size is significantly smaller. I'm curious enough to look into it in a bit more detail when I get another chunk of time, but I'd be interested if anyone knows of a change in rustc which could have changed this behaviour.