Looks good. Can you update the README.md (adding your name to contributors) and CHANGLOG.md (adding the changes). I can test and merge pretty quickly afterwards.

Done. tested on my side and it works pretty well.

PS : with verbose < 1 the output is pretty sparse. maybe we could add total, passed and failed statistics to the results. Irrespective of verbosity ?

PPS: the output has the F's printed for failure and seems pretty unnecessary..

Thanks for running test_HTMLTestRunner.py. Most people aren't and I think that's key for ensuring that HTMLTestRunner.py stays "semi" backward compatible.

• Are the changes backward compatible within Python 3? Just looking at them I think so, but didn't have a double check against newest Python features. I can merge after you can confirm/double check this as I think the branch is ready.
• Agreed on the verbosity. It would be better to ensure verbose gives a level of detail that is very helpful.
• Disagreed on the "F's printed". I think having that quick indicator to the screen is useful and easy to disregard if not necessary. I know many a time, I've sat watching the test run looking for no F's

the changes are 3 compatible (I think so based on the api documentation) and 2.7 backward compatible too(2.7 compatibility I checked). Just need to check if this code will run on 2.7 if print_function is imported from future and based on version import StringIO from io or from StringIO.

Will check that now and revert back on this thread.

About the F just an F is a quick indicator of whether there is a failure but maybe a counter would be of better use.

PS: ment to say a statistics of sorts.

total tests run, passed tests and failed tests

the errors are already there on the html document.

Another point that is stuck with me.

Is the stderr and stdout redirection still needed ?

It seems unessary. Testcases that I have written(basic tests with assertEqual) work fine and I get the correct report after removing the output and error stream redirection support.