As commented in bug 421611, it would be nice to have a --log-file option. This would be very useful for Fennec, and I imagine others would find it useful.

I think this would be useful on Windows as well because (from my testing at least) unless you have NO_EM_RESTART set, you won't get any output from reftest, which is a real awful setup. Perhaps I should add that to the MDC article on reftests if it isn't already there...

Clint: it should work fine with "make reftest", as that explicitly runs the app twice, with -silent the first time to avoid the restart pain.

(In reply to comment #0) > it would be nice to have a --log-file option. Do you actually need this option, or are the log files from bug 468913 and bug 484298 enough?

thanks for revisiting this. We don't need a --log-file option for reftest/xpcshell anymore. I have found these logging features to provide enough information.

Let's morph the bug then, as I think it would still be nice to replace '| tee ./$@.log' in { 73 RUN_REFTEST = rm -f ./$@.log && $(PYTHON) _tests/reftest/runreftest.py $(1) | tee ./$@.log } and when calling runreftest.py directly (as in bug 383136).

Why? If somebody wants stdout to go to a file, they can; you don't need to add it as a feature to every program.

(In reply to comment #6) > Why? Easy usage, synchronized behavior between tests harnesses, ...

one example of why this was originally consider is for windows ce where we don't have the luxury of stdout and file redirection.