Skip to content
Commits on Feb 9, 2016
  1. doc: start new HISTORY section for post-3.0.2 release

    I actually forgot to update this before cutting 3.0.2 (which I just
    pushed). If I had, I would have noticed that it should have been 3.1, as
    it adds a new feature. So we've deviated from semver a little.
Commits on Jan 31, 2016
  1. doc: Show how to get NeoBundle to autobuild

    As suggested here:
    I don't use NeoBundle myself, but I installed it temporarily to confirm
    that this works. Given that I didn't verify this on other platforms
    (like "cygwin" and "windows") I left those out. The "unix" build command
    runs on OS X ("mac") and presumably "linux" too, obviating the need to
    specify those explicitly (the commands would be the same anyway).
  2. doc: update AUTHORS section

    Instead of manually updating this and then using a fancy Vim macro to
    reformat it, just dump in the result of:
    :read !git shortlog -s|cut -f 2-3|column -c 72
    :retab " after selecting region, then indent with gv>>
    As part of this, get `.mailmap` into shape, adding some missing entries
    and simplifying others.
Commits on Jan 30, 2016
  1. Merge branch 'pull/202' into next

    Originally submitted at: #202
    * pull/202:
      Update Watchman binary protocol documentation URL
  2. @lencioni

    Update Watchman binary protocol documentation URL

    lencioni committed with
    I noticed that this URL was 404ing, so I tracked down where I think it
    moved to and updated the URL here. I considered referencing a commit
    hash instead of master to prevent this from happening again but decided
    against it because it will potentially be sending folks to outdated
Commits on Jan 26, 2016
  1. Add CommandT{WillShow,DidHide}MatchListing autocommands

    Adds `CommandTWillShowMatchListing` and `CommandTDidHideMatchListing`
    autocommands that can be used to do useful things like temporarily
    disabling and re-enabling other plug-ins before and after showing
    (hiding) the Command-T match listing window.
    Closes: #200
  2. Fix unwanted flicker on opening/closing match listing in MacVim

    Noted by @jordwalke here:
    I hadn't noticed this flicker because I use terminal Vim and this only
    manifests in MacVim. The fix is to clear the prompt (by forcing a
    redraw) before restoring the settings rather than after.
Commits on Jan 25, 2016
  1. Fix tiny typo in docs

  2. Tweak gemspec

    We got rid of README.txt in b772049. Instead of just updating that ref,
    replace the dodgy blacklist-based code for selecting files with a
    whitelist based on `git ls-files`.
  3. Restore compatiblity with Ruby 0.8.7

    Ruby 1.8.7 doesn't like trailing commas in method invocation parameter
    Closes: #198
    Made sure there were no other instances of this by using this clowny
    Perl one-liner in Zsh (note the use of Zsh-style `**` file glob):
    perl -0pe 's/,\s*\)/XXXXXX/m' ruby/**/*.rb|less
    And searching for "XXXXXX".
Commits on Jan 19, 2016
  1. Provide improved README for GitHub

    Instead of a symlink to the voluminous Vim documentation, provide a
    short Markdown README that embeds a demo screencast gif.
    Related: #194
  2. Rakefile: print a warning if mechanize is not installed

    Check as part of "prerelease" task. Will save me from having the upload
    task explode in my face and having to re-run it every time I switch to a
    new machine, or a new Ruby version.
  3. Make "help" the default Rake task

    Also dropping the reference to mechanize in the help text, as I mean to
    roll a check into the "prerelease" task instead.
  4. Remove references to product pages and repo

    I'm not linking to those old product pages any more, and am not planning
    on making any updates to them either. Things have been living on GitHub
    for a while now and the docs should reflect that.
    If I'm doing that, I might as well update the repo references as well.
    Although the true canonical home continues to be, in
    practice everybody is cloning from GitHub anyway (even me; see:
    Related: #194
Commits on Jan 18, 2016
  1. Update specs to RSpec 3 syntax

    Done using:
    $ gem install transpec
    $ transpec -f # -f needed because I have some untracked files
    $ bundle exec rspec # verify that everything worked
    Closes: #193
Commits on Jan 17, 2016
  1. Avoid using deprecated/removed `be_` matchers

    To play nicely with RSpec 3 (even though I am using RSpec 2.14.7
    locally), which has removed the `be_true` and `be_false` matchers in
    favor of `be_truthy`/`be_falsey`, which is what the old matchers
    actually meant.
    Reported here: #190
    Using the alternative (`be true`, `be false` etc) recommended here:
    rspec/rspec-rails#976 (comment)
    Because my intention actually always was to assert strictly against
  2. Fix spurious test failure

    Commit 8e15beb added a new `g:CommandTRecursiveMatch` setting, which
    defaults to "true" in the context of Command-T running inside Vim (ie.
    preserving the previous behavior) but it does not default to "true" in
    the artificial context of the test suite, causing one of the specs to
    fail spuriously.
    Fix that by explicitly setting `:recurse => true` in the test suite.
    (Of course, maybe it _should_ be defaulting to true everywhere, for
    consistency, but I wanted to keep the fix minimal before looking into a
    deeper/riskier refactoring.)
    Fixes: #189
  3. Silence some Clang warnings

        match.c:142:31: warning: implicit conversion loses integer precision: 'VALUE' (aka 'unsigned long') to 'int' [-Wshorten-64-to-32]
            m.case_sensitive        = case_sensitive;
                                    ~ ^~~~~~~~~~~~~~
        matcher.c:184:39: warning: implicit conversion loses integer precision: 'long' to 'int' [-Wshorten-64-to-32]
                thread_args[i].thread_count = thread_count;
                                            ~ ^~~~~~~~~~~~
        matcher.c:185:39: warning: implicit conversion loses integer precision: 'long' to 'int' [-Wshorten-64-to-32]
                thread_args[i].thread_index = i;
                                            ~ ^
Commits on Jan 2, 2016
  1. Avoid unwanted scrolling when showing/hiding match window

    Thanks to @jordwalke for pointing out this problem and sharing a gist
    showing that the unwanted displacement can be avoided by moving the
    cursor in each window away from the side on which a split is going to be
    This is pretty much a ghastly hack, so I'll want to bake it for a long
    time to be sure that it's worthy of landing on master. Note the
    unpleasant Vim APIs that make it impossible to get useful information
    from the windows without switching to them first. This requires us to do
    some awkward save/restore shenanigans and a bunch of manual focus
Commits on Jan 1, 2016
  1. Switch some defaults to more reasonable values

    As per discussion in:
  2. Merge branch 'pull/186'

    Originally submitted at: #186
    * pull/186:
      Update reference to 2.0a in readme
  3. @lencioni

    Update reference to 2.0a in readme

    lencioni committed with
    Now that 2.0 has been released, we can update this reference.
Commits on Dec 28, 2015
Commits on Dec 26, 2015
  1. Use input() to make sure deprecation messages get seen

    Quoting the rationale from what I wrote in a similar instance in the
    Ferret plug-in:
    " Use `input()` to show error output to user. Ideally, we would do this in a way
    " that didn't require user interaction, but this is the only reliable mechanism
    " that works for all cases. Alternatives considered:
    " (1) Using `:echomsg`
    "     When not using vim-dispatch, the screen is getting cleared before the
    "     user sees it, even with a pre-emptive `:redraw!` beforehand. Note that
    "     we can get the message to linger on the screen by making it multi-line and
    "     forcing Vim to show a prompt (see `:h hit-enter-prompt`), but this is not
    "     reliable because the number of lines required to force the prompt will
    "     vary by system, depending on the value of `'cmdheight'`.
    "     When using vim-dispatch, anything we output ends up getting swallowed
    "     before the user sees it, because something it is doing is clearing the
    "     screen. This is true no matter how many lines we output.
    " (2) Writing back into the quickfix/location list
    "     This interacts poorly with vim-dispatch. If we write back an error message
    "     and then call `:copen 1`, vim-dispatch ends up closing the listing before
    "     the user sees it.
    " (3) Using `:echoerr`
    "     This works, but presents to the user as an exception (see `:h :echoerr`).
  2. Show deprecation warnings for deleted functions

    Given that most people use Command-T via a plug-in manages they end up
    tracking `HEAD` of `master` and there is no nice way to warn about
    upcoming breaking changes (ie. there are no "releases" for these people
    and release notes are meaningless; furthermore, people may end up
    upgrading without any conscious decision to do so).
    So, add functions that show an error when a deprecated function is
    called, and then call through to the new function.
    Note that depending on the function being called, the screen may be
    cleared which means that the user may not see the error. I might add a
    prompt here to address this issue because many users may not be aware of
    the `:messages` command which can be used to display previous error
    messages if they scroll or otherwise move out of view.
Commits on Dec 25, 2015
  1. Add <leader>j mapping for :CommandTJump

    As usual, don't do it if a mapping already exists.
  2. Use <Plug> mappings as per standard best practises

    I am doing this in my other plug-ins, so it makes sense to do it here as
    well. This is kind of recommended in the Vim help ("used for mappings
    which the user might want to map a key sequence to").
Commits on Dec 18, 2015
  1. Future-proof copyright year ranges

    I used to like manually updating these whenever a particular file was
    touched because it gave me an indication at a glance of how "fresh" a
    file was (ie. I could see files which hadn't been touched in years).
    But now, laziness is starting to weigh more heavily than curiosity, so I
    am updating these to be future-proof, once and for all.
  2. Add g:CommandTCursorColor option

    This makes it possible to configure the color used for the cursor in the
    Command-T prompt.
    Closes: #184
Something went wrong with that request. Please try again.