Permalink
Commits on Jun 24, 2016
  1. @sorbits
  2. @sorbits

    Allow a commit summary of up to 72 characters before marking it invalid

    From ‘git help commit’ the recommendation is simply to make it ‘less than 50 characters’ which we still use as the limit for marking it as deprecated.
    
    I think GitHub used to truncate at 65 characters but currently they truncate at 72 which is a commonly used value for hard wrapping, so it makes sense for us to use this as well.
    sorbits committed Jun 24, 2016
Commits on May 1, 2016
  1. @arimbun @sorbits

    Fix error when deleting local branch containing slash character

    This fixes issue #40
    arimbun committed with sorbits Apr 30, 2016
Commits on Apr 27, 2016
  1. @arimbun @sorbits
Commits on Oct 31, 2015
  1. @jacob-carlborg

    Add support for automatically continue rebase after commit

    Requires TextMate 2.0-beta.8.2
    jacob-carlborg committed Oct 25, 2015
  2. @jacob-carlborg
Commits on Jun 11, 2015
  1. @jacob-carlborg
Commits on Nov 20, 2014
  1. @infininight

    Switch to using 'Hair Space' for menu selection control

    This changes the previously zero-width whitespace character used to control the menu item selected when typing a C with the menu open. The previous workaround became very visible on Yosemite, a hair space isn't absolutely invisible but almost so.
    infininight committed Nov 20, 2014
Commits on Sep 19, 2014
  1. @rdwampler

    Don't open revisions in TextMate asynchronously

    Previously, we would wait on TextMate to close before cleaning up the temporary file; however, we may want to avoid this since in the future TextMate may be able kill the process (before closing) and therefore leave behind the file. Also, let's use `TextMate.detach` instead of just `fork` to ensure we detach.
    
    This resolves an issue reported on the TextMate mailing list (http://lists.macromates.com/textmate/2014-September/037756.html).
    rdwampler committed Sep 18, 2014
Commits on Aug 23, 2014
  1. @kaiwood @infininight
Commits on Aug 19, 2014
  1. @kaiwood

    Add missing space to annotation matcher

    There is 1 additional space in the annotation format between
    the line number element and the parsed line. Without this,
    the output in the HTML popup for the blame indents 1 space to the
    right.
    kaiwood committed Aug 19, 2014
  2. @kaiwood
  3. @kaiwood

    Refactor annotation parser

    After #27, the old parse_annotation method was dead code
    nowhere used. This commit removes the dead part and renames
    the new method to the previous name, as it makes more sense
    and specs point to it.
    kaiwood committed Aug 19, 2014
Commits on Aug 18, 2014
  1. @gdsmith

    Support positive and negative timezone offset

    Currently breaks on positive timezone
    gdsmith committed Aug 18, 2014
  2. @infininight
  3. @infininight

    Follow moved files in Annotate/Blame history

    Cina Saffary committed with infininight May 21, 2014
Commits on Jun 18, 2014
  1. @infininight

    Point to ruby 1.8 shim

    Using a shim allows us to catch when 1.8 of ruby is not present and provide other options.
    
    #ignore
    infininight committed Jun 18, 2014
Commits on Apr 18, 2014
  1. @sorbits

    Include file name in commit window actions (revert and delete)

    Requires v2.0-alpha.9533
    sorbits committed Apr 13, 2014
Commits on Apr 16, 2014
  1. @rdwampler @infininight

    Refactor and update specs for PartialCommitWorker

    I wanted to include some additional specs with the new logic around amending, but ended up refactoring and adding more specs overall.
    rdwampler committed with infininight Apr 14, 2014
  2. @rdwampler @infininight

    When amending auto populate the commit window with last commit message

    Since we auto populate the textview with the last commit message, we eliminate the need to include the "-- update commit message --" item. This item was used to inform the bundle if the user intended to update the commit message (check) or just use the previous message (uncheck).
    
    So now when amending a commit, the commit window will show the last commit message in the textview and a list of potential files only if necessary (i.e., the table view will be empty in situations where the user wanted just to edit the commit message). This behavior is more consistent with the way amending a commit works when running "git commit --amend" from the command line.
    
    Also, we now check to see if there is nothing to amend (i.e., it is the initial commit).
    rdwampler committed with infininight Apr 11, 2014
  3. @rdwampler @infininight
  4. @rdwampler @infininight

    Add API to suppress/unsuppress (html) output

    This is useful when we would like to discard and not show any potential (html) output that may have been generated up to that point (e.g., canceling an action).
    
    Note that all output will be suppressed until "unsupress_output" is called. I.e., calling "flush" will not show the current output buffer.
    rdwampler committed with infininight Apr 14, 2014
Commits on Apr 11, 2014
  1. @infininight

    Update README

    #ignore
    infininight committed Apr 11, 2014
  2. @infininight

    Remove in favor of commit history

    #ignore
    infininight committed Apr 11, 2014
  3. @infininight
  4. @sorbits

    Revert "Fix for the diff action in the commit window"

    This reverts commit 12c87d5.
    sorbits committed Apr 11, 2014
  5. @sorbits

    Running diff from commit window will now show changes since HEAD

    By default, running “git diff” shows local changes that have not yet been staged, so if the user had already added (a version of) the file to the staging area, the diff button would only show changes done since that, which is misleading (as we commit to HEAD not the staging area).
    
    A minor issue exists for newly initialized repositories without a HEAD to diff against, here the diff command actually fails, but we ignore the return code and show a blank diff.
    sorbits committed Apr 11, 2014
Commits on Apr 10, 2014
  1. @rdwampler @infininight

    Only show output window after closing commit window

    When the commit window was a separate program and required siginificant time to launch, it made since to initially send some kind of information to the users via the output window. However, now that the commit window runs as part of the Textmate, it may be a better user experience to delay output until after the commit window is closed.
    rdwampler committed with infininight Apr 8, 2014
  2. @rdwampler @infininight
Commits on Apr 6, 2014
  1. @noniq

    Fix Branch#current_name for submodules.

    The fix is simple: Just use `Git#git_dir` instead of assuming that the git dir is `#{base_path}/.git` (which is not the case when working with submodules).
    noniq committed Apr 6, 2014
  2. @noniq

    Make parameter to Git#git_dir optional.

    `paths.first` seems to be a reasonable default value.
    noniq committed Apr 6, 2014
  3. @noniq

    Fix specs.

    Git commands like `status` and `log` are now called with additional parameters, so I updated `Git.command_response` mocks accordingly.
    
    Also there were some problems with redefined constants and nil strings that seem to be related to Ruby versions > 1.8 (fixes should be backward compatible, though).
    
    I commented out "should auto-expand the path when filtering to a relative path" (in `lib/commands/status_spec.rb`), as this feature was (accidentally?) removed in 8ecadc8.
    noniq committed Apr 6, 2014
  4. @noniq

    Add Gemfile to document RSpec version requirement.

    Specs need to be run with RSpec 1.
    noniq committed Apr 6, 2014
  5. @noniq

    Fix typo in Parsers#parse_status_hash

    Make spec "should parse a status document correctly" (`lib/commands/status_spec.rb`) pass again.
    noniq committed Apr 6, 2014
Commits on Mar 15, 2014
  1. @noniq @infininight

    Correctly determine working copy root when inside submodule.

    This fixes error messages like "fatal: '/path/to/file' is outside repository" when using the bundle with git submodules.
    
    `Git.path` assumed that the git dir would always be located directly below the working copy root. However, this is not (no longer?) true when working with submodules: The git dir of a submodule is nested inside the parent project’s git dir.
    
    Using `git rev-parse --show-toplevel` (available since Git 1.7.0) figures this out correctly, even for submodules.
    
    Similar incorrect code existed in `Compare Branches.tmCommand`. This was dead code however (`path` not being used anywhere afterwards), so I removed the line there completely.
    noniq committed with infininight Feb 22, 2014