Comment 4 for bug 1388586

Ok, so a quick examination shows that this file has lots of unencapsulated stdout. encapsulating it via subunit-filter -s to a new file gets me:

time ../../filters/subunit-ls < ~/Downloads/failed-subunit2.2 > /dev/null

real 0m3.875s
user 0m3.775s
sys 0m0.087s

or as time ../../filters/subunit-ls --no-passthrough < ~/Downloads/failed-subunit2.2

real 0m4.843s
user 0m4.444s
sys 0m0.149s

vs

time ../../filters/subunit-ls < ~/Downloads/failed-subunit2 > /dev/null

real 1m14.881s
user 1m3.278s
sys 0m11.570s

So about 1 second in terminal emulator blocking when not rendering to /dev/null. and the rest in utf8 decode I suspect. Will profile shortly to confirm/deny this. Its still kindof slow either way, of course - there are only 6.5K tests, which makes this nearly a ms per test as processing.