pytest-dev
Test case verbosity
Allow for the output of test case execution to be controlled independently from the application verbosity level. verbosity_test_case is the new ini setting to adjust this functionality.
Addresses #11639
I also don't love verbosity_test_cases. But I think part of the problem is that even just describing what it controls is more complicated than verbosity_assertions. verbosity_test_execution kind of works, but I could see it being misunderstood that it would affect the "Captured stdout" portion of the output. It also makes the --collect-only part less obvious.
I could see an argument that --collect-only is different enough to be it's own thing. But to me, they both feel like "pytest found each test case and is now displaying info as it iterated over them". I was also trying to avoid an over correction that added too many buckets that could be controlled.
I have addressed the initial comments and corrected the bug I identified while re-working the tests. (Thanks for that push. They are now more thorough and easier to follow.)
@plannigan In https://github.com/pytest-dev/pytest/issues/11387#issuecomment-1722324799 you did some great work listing all of the places where verbosity is used. While the assertions verbosity level was a common feature request, I'm not entirely sure about the others (as opposed to using the global verbosity level).
For me, in order to properly evaluate going forward, it will be helpful if you can list all of the categories you're thinking about and how the uses listed in #11387 will categorize.
I'm not entirely sure about the others (as opposed to using the global verbosity level).
For "test case" specifically, I've seen people wanting to use more verbosity for it, while keeping using less verbosity for assertions (it was internally at work on a specific case), so I think this option has its uses.
it will be helpful if you can list all of the categories you're thinking about and how the uses listed in https://github.com/pytest-dev/pytest/issues/11387 will categorize.
It would be helpful indeed to list them in #11387.
For "test case" specifically, I've seen people wanting to use more verbosity for it, while keeping using less verbosity for assertions (it was internally at work on a specific case), so I think this option has its uses.
Right, but for this they can bump verbosity and decrease verbosity_assertions now. What I'm curious about is whether there's demand for controlling the remaining cases besides assertions separately.
@bluetech I think it is a good idea to take a step back and consider where this work could go. After reviewing my original list of where verbosity is used and things seem to fall into 3 larger categories:
- Things in the header
- Things in the summary (this covers assertions, which has already been merged)
- The things in between (this PR)
There is one other potential category dealing with traceback filtering and truncating locals. Since that is in the summary, there might be an argument that it should also be grouped with assertions, but at first glance, it feels distinct to me. (Maybe if verbosity_assertions was renamed to verbosity_summary I might think differently.)
As for the "would people use this?" question. I agree that people could achieve it as described. But I feel that having this being the answer is:
- harder for people to discover
- a bit clunky (
addopts = "-vv"andverbosity_assertions = 0to do everything in the config file). - less explicit - a reader would have to understand the implications of both being used. Going further, the usages might not be next to each other, making it harder to know both are used.
@bluetech @nicoddemus Checking back in as there hasn't been any activity for 2 weeks since my last reply (hopefully you both took some time off for the holidays).
(hopefully you both took some time off for the holidays).
Indeed this was the case for me.
But the remaining discussion is probably best continued by @bluetech
@bluetech I have updated the branch so that it would be able to be included in the 8.0 release with verbosity_assertions. Do you have any more requests/feedback or can this move forward?
@bluetech It has been two months since I responded to your request for more information. But there has been no response. I updated the branch so that it might be able to be include in 8.0, but that did not happen. There is more work that I'd be interested to contribute, but without any response, I'll likely close the branch and move onto other things.
@plannigan Sorry for not responding earlier and letting it linger -- not very nice.
I read your comment but I am not convinced personally. I think it is better to only add configuration knobs when there is demand for it, and my feeling is that there isn't clear demand for this one. I don't mean to say that the feature isn't useful per se, just that IMO it's not useful enough, it's a balancing act.
So I'm -0 on this feature, meaning I won't merge it myself but if any other maintainer feels otherwise that would be perfectly fine by me.
If @bluetech is -0 on this, I would like to get this merged then. This is something people at work have mentioned in the past so I think it is a nice addition.
I will make a final review tomorrow and merge it then. 👍
Thanks for the responses. I've updated the branch to include the latest of main and reverted that changelog changes I had made to be compatible with the v8 rc. Should be good to go.