In a perfect world, I would come up with hooks to handle weird exceptions, and I would also have clean running tests every time. However, this is not a perfect world, and right now, getting the tests to reliably work is my priority. Also, I want to be able to rerun tests that fail and see if they are spurious failures or if there's a real problem.
How did i do this? I made a shell wrapper that would tee the output of the console to a rerun file, and then I would capture the failures, recreate a test suite of just the failed scenarios, and run them individually. Repeat until we get a 100% clean run or we confirm that something is genuinely broken.
So what's my problem? The lovely red/green/yellow color scheme that shows me what's happening at a glance disappears, replaced with a monochrome representation of all the steps. Do the tests run? Sure? Does the rerun whittle down my errors until I get to the essential problems? Yep, it does. I just want to have my cake and eat it too, if possible.
If I run without the rerun script (i.e. sans tee):
If I run with the rerun script (i.e. with the tee):
I get that this is because the system is trying to be smart and not deal with colors because the output is going to a pipe and not to the traditional console. My question is, how can I make my system stop being so smart? the -c (force color) option doesn't cut it.
No comments:
Post a Comment