[TIP] TAP vs subunit

Remko Tronçon remko at el-tramo.be
Mon May 16 03:27:01 PDT 2016

Hi Robert,

> offset within the output of that script, and no way to say 'run just
> this one failing case'.

I'm not sure I understand how the output format affects the test runner?

I'm assuming you're talking about some automatic feedback mechanism I'm 
unfamiliar with that maps tests from the output back into the input of the 

In my case, the test IDs are used as the names of the tests in my TAP test 
output, so I just pass back the names of failing tests into my test runner 
on the  command prompt.

> 'unexpected success' is semantic information, provided by Python and
> emitted as plain text (not semantic information) by the TAP serializer
> because TAP only understands 'ok' or 'not ok' for a test outcome.

You're right, semantics of test states are very limited in TAP. You can
add any structured extra information in a YAML block, but nothing
of this is standardized, and none of the standard producers support
it. I personally haven't had the need to have semantics beyond 'ok' and
'not ok', though. Any other details i get from the logs or the YAML.

> Sure,:). You might enjoy subunit, which I started - gosh over a decade

Thanks for the tip! 


More information about the testing-in-python mailing list