Permalink
Commits on Nov 17, 2016
  1. v8.0.1

    isaacs committed Nov 17, 2016
Commits on Oct 25, 2016
  1. v8.0.0

    isaacs committed Oct 25, 2016
  2. tmatch@3.0.0

    Also, remove extraneous 'files' entry, and sort files list
    isaacs committed Oct 25, 2016
Commits on Sep 12, 2016
  1. skip sigterm child test on windows

    isaacs committed Sep 12, 2016
Commits on Sep 10, 2016
  1. Handle unfinished promise-awaiting tests

    isaacs committed Sep 10, 2016
Commits on Sep 9, 2016
  1. v7.1.2

    isaacs committed Sep 9, 2016
  2. Use _TAP_COVERAGE_ environ instead of --__coverage__

    Fix #302
    Fix #246
    
    h/t to @martinheidegger for finding the root problem.
    isaacs committed Sep 9, 2016
Commits on Sep 8, 2016
  1. v7.1.1

    isaacs committed Sep 8, 2016
Commits on Sep 7, 2016
  1. Handle bailouts better after test end happens

    isaacs committed Sep 7, 2016
  2. remove 0.12 from travis

    It's as flaky as 0.10, but way slower, and all the language/platform
    features are in 4.0 and above anyway.
    isaacs committed Sep 7, 2016
  3. skip some tests that always fail on 0.10

    isaacs committed Sep 7, 2016
  4. v7.1.0

    isaacs committed Sep 7, 2016
  5. epipe test: Use shared fd's, and don't run on Travis

    isaacs committed Sep 7, 2016
  6. make runner-epipe test more resilient

    isaacs committed Sep 7, 2016
  7. fix smoke test command

    isaacs committed Sep 7, 2016
Commits on Sep 6, 2016
  1. test: don't require('tap'), require('../')

    For CI
    isaacs committed Sep 6, 2016
  2. give tests more time

    Getting a lot of spurious timeout errors, especially on Travis.
    isaacs committed Sep 6, 2016
  3. mkdirp/rimraf used in tests

    isaacs committed Sep 6, 2016
  4. v7.0.1

    isaacs committed Sep 6, 2016
  5. rename output tests so that they're together

    isaacs committed Sep 6, 2016
  6. Report bailouts in child tests more consistently

    Fix #304
    
    Prior to this patch, there was a race condition between a child test's
    output being added (indented) to the parent test's output, and the
    parser seeing that a failed test point had occurred and bailing out.
    
    Making matters worse, output from child test streams was being
    needlessly double-parsed and analyzed line by line in a confusing manner.
    
    Since a Parser was already being instantiated for these child test
    streams, it seemed most reasonable to use the line parsing already built
    into the parser and DRY up some extraneous bailout handling code.
    
    This also eliminates a lot of mid-hierarchy `Bail out!` lines that were
    being included in inconsistent ways, and fixes the sporadic failures in
    the output tests.
    isaacs committed Sep 6, 2016
Commits on Sep 5, 2016
  1. break up big output test into two sets

    otherwise this thing times out a lot
    isaacs committed Sep 5, 2016
Commits on Aug 31, 2016
  1. node 6 changed debugger message

    isaacs committed Aug 31, 2016
Commits on Aug 27, 2016
  1. fix test regression from f038a11

    isaacs committed Aug 27, 2016
  2. v7.0.0

    isaacs committed Aug 27, 2016
  3. Don't filter out indented bailouts or version lines from child test p…

    …rocs
    
    Since indented, these can be a part of yaml strings, they were causing some
    invalid yaml to be output in runner tests.
    isaacs committed Aug 27, 2016
  4. allow SIGKILL in runner timeout tests

    Especially when coverage is engaged, this test can run slow, resulting in
    a SIGTERM not killing the process in time and a SIGKILL being issued.
    isaacs committed Aug 27, 2016
Commits on Aug 26, 2016
  1. add the --100 flag for 100% coverage check

    I keep having to add 4 flags to module tests for full coverage,
    why not just have a single option for it?
    isaacs committed Aug 26, 2016