How xdist according to the test file allocation process？

[linlin547]

Hello, my English is very poor, I want to use xdist allocation process, not according to the distribution of the test method, Need a process distribution a test file, what should I do?Thank you very much.

[RonnyPfannschmidt]

with your question, i dont see a good way to answer since you didnt even remotely outline your use-case

your English skill is indeed a major communication blocker there

[linlin547]

How are you I want to use xdist plug-in distribution test files, a process to distribute a test file, instead of distribution according to the amount of test method, what should I do?Thank you very much.

[RonnyPfannschmidt]

@linlin547 again, if i cant make sense of your question, i cant hope to help, if your own English is too bad please get help from someone with better English skills,

else i will have to close this issue as not actionable due to language barrier

[nicoddemus]

I'm going on a limb here, but perhaps @linlin547 means that he wants to distribute the tests to workers grouped by test module instead of by individual tests sent to each worker individually? @linlin547 can you confirm? If that's the case, there's #191 going on to fit exactly that.

[linlin547]

all methods execute in a same thread from a same file .

[linlin547]

execute tests group by class in a same thread @RonnyPfannschmidt @nicoddemus

[linlin547]

the same problem

https://stackoverflow.com/questions/36586422/pytest-distributed-testing-xdist-by-class

[nicoddemus]

Thanks for finding an example demonstrating your problem. As I said there's work regarding this on #191. I recommend to watch that issue closely. 👍

Closing this for now.

[linlin547]

example :
1.I have three test files, they are test1,test2,test3
2.now pytest.ini addopts= -n 3
3.I hope thread1 run test1 file,thread2 run test2 file,thread3 run test3 file

@RonnyPfannschmidt @nicoddemus

[nicoddemus]

Yep it's clear now.

The short answer is: that can't be done right now, but there's work being done in #191.