this PR is an experiment. I'm almost sure that @Dickby's commit caused the surround function test to begin to fail, although travis did not catch it because make test is returning 0 on test failures.
Here I use @lionel-'s PR #135 and revert 5a20c9757 on top of it.
this PR is an experiment. I'm almost sure that @Dickby's commit caused the surround function test to begin to fail, although travis did not catch it because
make test
is returning 0 on test failures.Here I use @lionel-'s PR #135 and revert 5a20c9757 on top of it.