Terrible timing I'm very aware but should we maybe adjust the print_test_
function signatures before making them public?
Currently print_test_finished
seems to swap the test name and the record for some reason.
function print_test_started(::Type{TestRecord}, wrkr, test, ctx::TestIOContext)
function print_test_finished(test, wrkr, record::TestRecord, ctx::TestIOContext)
function print_test_errorred(::Type{TestRecord}, wrkr, test, ctx::TestIOContext, test_time)
I suggest changing print_test_finished
to:
function print_test_finished(record::TestRecord, wrkr, test, ctx::TestIOContext)
for consistency with the other 2.
And then CUDA.jl can set its ParallelTestRunner
compat to 1.1 or whatever the version we make them public is.
Originally posted by @christiangnrd in #34 (comment)