Would be great.
I think this first requires a rewrite of the Base test system to have some sort of TestRunner very similar to the BenchmarkGroup in BenchmarkTools.jl: https://github.com/JuliaCI/BenchmarkTools.jl/blob/master/doc/manual.md#the-benchmarkgroup-type.
We could then pass arguments to that runner and it would choose what tests to execute. What we currently do is just include a file (runtests.jl) and run it from top to bottom. This provides almost no options for customizing the execution behavior.
That makes sense.
Just a random question:
Is it possible to make it like pytest? Specify a test filename, and then do that test?
The way test is documented now is "run test/runtests.jl. We could document test bleh to run test/bleh.jl but it seems like a fairly minor improvement (you could just do include("test/bleh.jl").... I think there needs to be a rewrite of the testing system that is customizable and extensible and until then there is nothing really that can be done from this side.
Until the testing system itself has support for this, nothing to do here.
Most helpful comment
Until the testing system itself has support for this, nothing to do here.