0.13 2016-05-03 - Really pass teamcity messages through directly. When running tests in parallel these messages were still being buffered. 0.12 2016-04-21 * If any code prints a teamcity message directly, it passed through immediately by the formatter, rather than being buffered. This lets you issue progressMessage output from inside your test code. 0.11 2016-04-20 * This module now uses TeamCity::Message to generate messages, instead of doing this with private internal code. 0.10 2016-03-15 * Previously a lot of test output (especially diagnostics from failures) could end up being doubled in the output stream. As of this release all output is now encoded as a TeamCity build message, and nothing should be printed directly to STDOUT. This will change how some output appears in the build log, but should overall be an improvement in clarity. Reported by Mark Fowler. GitHub #5. 0.09 2016-01-11 * All TeamCity messages with name/value pairs now include a flowId value as well. This is always the same as the name of the test file being run, which works fine as long as you do not fork additional processes from inside a .t file. 0.08 2016-01-04 * All TeamCity messages with name/value pairs now include a timestamp value as well. This lets TeamCity calculate test length. 0.05 2015-12-08 * This release is nearly a rewrite from prior releases. Changes include: * Added tests for many different scenarios. * Made sure that this formatter works with skipped tests. * Made the formatter handle various test death scenarios cleanly. * Made sure that this formatter works with subtests and Test::Class::Moose. * Made the formatter hide deep recursion warnings triggered by Devel::Cover. * Made the formatter work when running tests in parallel. 0.04 2009-09-09 * Test diagnostics are now passed through to STDOUT and STDERR, and these will be presented by TeamCity in the "details" for each test failure. However, TeamCity strips out the newlines, so the presentation isn't very pretty. The code for doing this is not very pretty either. TeamCity is sensitive about the order of the messages is receives. So all the diagnostic info has to come between the testStarted and testFinished messages. But with TAP, it isn't easy to know where the test begins and ends. * It seems to work for now, but I'll probably re-write it in the near future. 0.03 2009-07-30 * Added descriptions to the NAME sections in POD. 0.02 2009-07-30 * Forgot to use 'eval' when loading optional modules. 0.01 2009-07-30 * Initial release on an unsuspecting world.