Shlomi Fish wrote: > Well, my original motivation was to have "make test" still use Test::Harness > so one will not need to use Test::Run to run the tests. I don't want to > impose Test::Run. Now instead, those that have Test::Run installed will be > able to say "make runtest" or "make distruntest" (which is harder to > implement as a command line utility.) > > I already have "runprove" which is a Test-Run-based command line utility > like "prove". However, running "runprove -l t/*.t" or in our case, something > using Test::Manifest, is more error prone than typing "make runtest" > and "make distruntest" would be even harder to get right this way.
So here's the question: What's the gain for the user? Why do all this extra work to allow a user to optionally run the tests with a slightly different harness IF they happen realize the new target is available for that module? The output looks exactly the same. Seems the energy is better spent elsewhere. > Over-riding "make test" and "make disttest" would not be a good idea, because > it will necessisate Test::Run in order to test the module. *shrug* That's what prereqs are for. Ironically, Test::Run::CmdLine doesn't test itself with itself. PS examples/mini-harness.plx is an unaltered copy of Test::Harness'. You should probably remove it. PPS examples/eumm-test-manifest.plx in Test-Run uses Test::Run::CmdLine which doesn't come with Test-Run.