Currently htmldocck allows the text in rustdoc tests to be shorter, but not longer than the expected test. So if the actual text were Option<T and you wrote Option<T>, it would fail the test. But if the actual were Option<T> and the expected were Option<T, it would pass the test.
This behavior is not super intuitive and seems like it could lead to false negatives where it didn't catch regressions. Rustdoc should consider changing it.
Originally posted by @jyn514 in #76934 (comment)
Currently
htmldocckallows the text in rustdoc tests to be shorter, but not longer than the expected test. So if the actual text wereOption<Tand you wroteOption<T>, it would fail the test. But if the actual wereOption<T>and the expected wereOption<T, it would pass the test.This behavior is not super intuitive and seems like it could lead to false negatives where it didn't catch regressions. Rustdoc should consider changing it.
Originally posted by @jyn514 in #76934 (comment)