The "ids" argument to "parametrize" again accepts unicode strings in Python 2

[nicoddemus]

Fixes #1905

[coveralls]

Coverage remained the same at 93.078% when pulling 1e10de5 on nicoddemus:parametrize-unicode-id into 722f9ea on pytest-dev:master.

[coveralls]

Coverage remained the same at 93.078% when pulling 1e10de5 on nicoddemus:parametrize-unicode-id into 722f9ea on pytest-dev:master.

[RonnyPfannschmidt]

unicode support on python 2 got removed for some specific reason/regression - i think we want test id#s tobe native strings ins all cases

[RonnyPfannschmidt]

i think i recalled the reason, environmental reasons may make python fail to print/output the unicode id's

i think we need to absolutely ensure the ids are native strings
please lets verify that and add a test

[The-Compiler]

unicode support on python 2 got removed for some specific reason/regression

No, the check was added (by @nicoddemus) in response to #1857 to improve the error message when someone passes e.g. integers to parametrize.

[nicoddemus]

Yep

[flub]

So we can merge this anyway?

[nicoddemus]

From my POV yes 😁

[The-Compiler]

Well then, I dared to click the button 😉