[pytest-dev] Fwd: Re: Proposal: python namespacign for markings/marker objects

[holger krekel]

Hey Ronny,

On Thu, Oct 20, 2016 at 09:07 +0200, Ronny Pfannschmidt wrote:
> forwarding as i send it with the wrong sender address

I find it better to send a wrongly-sent mail afresh (copy+paste) than forwarding because this forwarding looses threading-info.

> > Finally, and I only just remembered, Holger has some lingering code
> > somewhere which does something like:
> >
> > @pytest.marker
> > def blocker(n):
> >     return {'issue no': int(n)}  # or whatever object you want your marker to be
> >
> > @pytest.mark.blocker(123)
> > def test_fun():
> >     pass
> fixtures and markers are not symmetric and i don't see a reason to make
> them be

the similarity IMO would mostly be that both fixtures and markers use functions. Currently markers require non-trivial parsing code for arguments. Whereas for the above example if you have:

    @pytest.mark.blocker(n=138)

it would work because "pytest.mark.blocker(...)" invokes "blocker(...)" and parsing of arguments including kw-arguments works naturally. By contrast, today you have to parse "markerinfo.kwargs" and "markerinfo.args" yourself and few plugins do it correctly.  

Moreover, by forcing all marker instantiation to go through a registered function we have a single normalizing source of documentation.  We can take the marker function's docstring and inspect the arguments.  The current documentation for markers ("pytest --markers") is often not in sync with implementation.  

holger