Permalink
Commits on Mar 25, 2014
  1. magit-get-current-tag: avoid using --dirty

    tarsius committed Mar 25, 2014
    Don't use the Git `--dirty' argument unless the function argument
    WITH-DISTANCE-P is the symbol `dirty'.  This makes `git tag' much faster
    on file-systems that don't properly cache access times.  Inside Magit we
    currently are not ever interested in knowing whether there is any dirt,
    but some third-party code does (well, some of my local hacks, that is).
Commits on Mar 24, 2014
  1. add ability to undo last revert

    tarsius committed Mar 24, 2014
    Close #79.
  2. Merge pull request #1294 from Silex/master

    tarsius committed Mar 24, 2014
    Fix git pull behavior with multiple remotes
  3. Merge branch 'fix-rewrite'

    tarsius committed Mar 24, 2014
  4. magit-rewrite-{set-commit-property,finish-step}: fix regressions

    tarsius committed Mar 24, 2014
    Make `magit-rewrite-set-commit-property' always return t so that the
    `magit-section-action' hooks `magit-{apply,cherry-pick,revert}-hook'
    run just the rewrite case, and not the default `commit' case too.
    
    In `magit-rewrite-finish-step', do not use `magit-cherry-pick-commit'.
    That function does a refresh before we had a chance to call
    `magit-rewrite-set-commit-property'.  If cherry-picking a commit fails,
    then say so.
Commits on Mar 23, 2014
  1. magit-run-{git-gui{,-blame},gitk}: use call-process

    tarsius committed Mar 23, 2014
    Start the subprocess using `call-process' with argument DESTINATION
    being 0.  I.e. start the subprocess and immediately return 0 without
    caring what happens to the subprocess afterwards.  Previously we used
    `start-file-process' but that had the disadvantage that the process
    object still existed when the subprocess was still running when Emacs
    was being quit, needlessly forcing the termination of the subprocess.
  2. magit-gitk-executable: locate executable at load-time

    tarsius committed Mar 23, 2014
    Locate the gitk executable at load-time when setting the default
    value of `magit-gitk-executable', instead of doing it every time
    `magit-run-gitk' is used.  This builds on a similar hack used
    when setting the default value of `magit-git-executable'.
Commits on Mar 22, 2014
  1. magit-wash-diff-section: revert accidental face change

    tarsius committed Mar 22, 2014
    Not that I think we need to use different faces here but stopping to do
    so was not intentional.
  2. Merge pull request #1290 from achitu/master

    tarsius committed Mar 22, 2014
    magit-revert-buffers: revert buffers in emacs 24.4
  3. magit-revert-buffers: revert buffers in emacs 24.4

    achitu committed Mar 22, 2014
    In Emacs 24.4, when `auto-revert-use-notify` is t, but auto-revert is
    not enabled for current buffer (auto-revert-notify-modified-p remains
    nil), `auto-revert-handler` will not revert the buffer. Setting
    `auto-revert-mode` to t for buffer to be reverted will make
     `auto-revert-handler` effective.
Commits on Mar 21, 2014
  1. Merge branch 'git-executable'

    tarsius committed Mar 21, 2014
  2. magit-git-executable: use git.exe on Windows

    tarsius committed Mar 21, 2014
    On Windows, explicitly use the "git.exe" executable instead of "git".
    If that fails accept any "git".
    
    It has been reported that this gives a considerable speedup.  I don't
    know Windows but it appears that when asking for a "foo" one might get
    a "foo.exe" or a "foo.cmd" or a "foo".  It also appears that "git.cmd"
    is earlier on the `exec-path' than "git.exe", or that the former is
    preferred for some other reason.
  3. magit-git-executable: use executable-find when initializing value

    tarsius committed Mar 21, 2014
    This hopefully give a slight performance bust.
    
    The downside is that the headers in the process buffer get longer
    and that the "interesting parts" might end up being truncated.
    
    Still I hesitate to add an option to hide the full path and the
    `magit-git-standard-options'; when instructing users to provide
    the information required to debug some issue there would be one
    more thing that we would have to make sure they actually did.
    
    If `executable-find' fails to find Git then fallback to "git".
Commits on Mar 20, 2014
  1. use official travis-ci button again

    tarsius committed Mar 20, 2014
    The button from shields.io takes to long to be updated.
  2. Merge branch 'faster-diff-2'

    tarsius committed Mar 20, 2014
  3. magit-wash-diff: absorb utility funtions

    tarsius committed Mar 20, 2014
    Yes, this is spaghetti code but having the spaghetti split
    across multiple pots actually makes it harder to turn this
    mess into a good meal.
  4. move definitions of functions and variables used for diff washing

    tarsius committed Mar 20, 2014
    In preparation of major refactoring move definitions of functions and
    variables used for diff washing.  The old location has never made
    much sense.
    
    Only move the functions and variables that implement "diff" and "hunk"
    washing, and leave the "raw diff" functions where they are now.  They
    are not going to be refactored.  This extra washing cycle will be
    removed without a replacement because it has no benefits but many
    downsides.
  5. magit-mode-setup: abort if not inside a repository

    tarsius committed Mar 20, 2014
    Previously we ended up setting `default-directory' to nil in
    `magit-mode-init'.  While doing that is always a bad idea in this
    case the result was fatal; for some reason it caused Emacs to enter
    an endless loop and neither C-g nor ESC where able to break it.
    
    Close #1255.
  6. remove dead function magit-wash-typechange-section

    tarsius committed Mar 20, 2014
    By the way, the functionality this intended to provided will be
    implemented for real soon.
  7. magit-read-stash: actually use PROMPT argument

    tarsius committed Feb 22, 2014
    (cherry picked from commit f9739c5)
    
    Conflicts:
    	magit.el
Commits on Mar 19, 2014
  1. Merge pull request #1284 from vanicat/t/this-command-in-magit-key-mode

    tarsius committed Mar 19, 2014
    Set this-command in magit-key-mode-command