NEWS

v1.6, not yet released
  * silence an "unused label 'skip'" warning on some sed
    implementations
  * define and use t4s_unset() to work around another NetBSD shell
    bug:  http://gnats.netbsd.org/49595

v1.5, released 2016-02-20
  * remove some unneeded quotes to work around a NetBSD shell bug:
    http://gnats.netbsd.org/50827
  * add t4s_testcase_diff() for comparing two (text) files
  * add a helper function to discard the last N arguments
  * work around a strange issue where a segfault in a testcase can
    cause a t4s_testcase() subshell to exit before the testcase's
    return value is recorded

v1.4, released 2016-01-11
  * add a new '--exit-hook' argument to t4s_setup() to run
    user-specified code on exit (traps on EXIT, HUP, INT, and TERM are
    reserved by t4s_setup())
  * support empty bailout messages
  * if the script exits unexpectedly, log the reason for the exit in
    the bailout message ("unexpected early exit" or signal name)
  * fix t4s_esc() to properly escape arguments with trailing newlines
  * print "command not found" if t4s_subtests() is given an invalid
    command

v1.3, released 2015-11-07
  * run subtests in a subshell to prevent them from interfering with
    the test environment
  * use file descriptor 9 for bailout message communication to reduce
    the chance of conflict with testcases
  * close unnecessary file descriptors before running testcases and
    subtests to prevent tests from interfering with the test
    environment
  * permit positional arguments for the testcase script
  * the tap4sh version is now available in the t4s_version,
    t4s_version_major, and t4s_version_minor variables

v1.2, released 2015-06-14
  * add t4s_give() for marking a prerequisite as satisfied

v1.1, released 2015-06-14
  * t4s_log() now prefixes each line with '# ' when not in a
    testcase (t4s_testcase() already prefixes lines written to
    stdout with '# ')
  * t4s_log() now reads stdin for lines to log if not given any
    arguments
  * t4s_testcase() now supports prerequisites for automatically
    skipping tests if certain previous tests failed (via the new
    --needs and --gives arguments)

v1.0, released 2015-06-12:
  * initial release