1. 17 Jan, 2018 1 commit
  2. 07 Jan, 2018 1 commit
  3. 06 Jul, 2017 1 commit
    • Kyle Meyer's avatar
      magit-tests.el: require tramp-sh · 5a99e51b
      Kyle Meyer authored
      456a77d3 (magit-tests.el: require tramp, 2017-07-02) seems to be
      sufficient to get tests passing locally, but Travis still fails.
      The same with-editor.el commit that removed "(require 'tramp)" also
      removed "(require 'tramp-sh nil t)"; loading tramp-sh in the tests
      fixes the Travis builds.
      Note that we no longer need to use the NOERROR argument when loading
      tramp-sh because tramp-sh is present in Emacs 24.4, our minimum Emacs
  4. 02 Jul, 2017 1 commit
  5. 08 Apr, 2017 1 commit
  6. 04 Jan, 2017 1 commit
  7. 09 Oct, 2016 1 commit
  8. 17 May, 2016 1 commit
  9. 01 Jan, 2016 1 commit
  10. 06 Dec, 2015 1 commit
  11. 16 Oct, 2015 2 commits
    • Jonas Bernoulli's avatar
      magit-process-find-password-functions: new option · b9dfaa53
      Jonas Bernoulli authored
      Add new option `magit-process-find-password-functions' and advertise
      adding `magit-process-password-auth-sources' to its value, but do not
      actually do so.
      We don't use that function by default, because most users will be
      unaware of this functionality and that might cause issues.  For example
      a user may have configured `auth-source' at some point, and then stopped
      using it and also changed the password without also changing it in the
      source used by `auth-source'.  In such a case `auth-source' would return
      the wrong password and that would then be feed to git.  The user likely
      would not note that this is what is happening and report a Magit defect.
    • Damien Cassou's avatar
      Try to find the password before asking user · 1ae403bb
      Damien Cassou authored
      When git asks the user for a password, this patch first asks the auth-source library. If no password is found, prompt the user.
  12. 06 Sep, 2015 2 commits
    • Jonas Bernoulli's avatar
      magit-toplevel: add doc-string · c35f21dc
      Jonas Bernoulli authored
      Also add entry to release notes and a comment in the tests.
    • Jonas Bernoulli's avatar
      magit-toplevel: refactor and add tests · 52b5ef26
      Jonas Bernoulli authored
      The idea is to only follow symlinks if absolutely necessary.  The
      previous implementation was broken in that it usually failed when
      a symlink to some sub-directory of a repository was involved (and
      when that succeeded, then only by accident).
      This issue was reported in #2135.  A previous attempt (#2209)
      failed to address this issue, as it caused other cases to fail.
  13. 02 Sep, 2015 1 commit
  14. 30 Aug, 2015 1 commit
  15. 26 Aug, 2015 1 commit
  16. 23 Aug, 2015 1 commit
  17. 27 Jul, 2015 1 commit
  18. 23 Jun, 2015 1 commit
    • Jonas Bernoulli's avatar
      clarify copyright lines in library headers · 48e73138
      Jonas Bernoulli authored
      At the beginning of library headers use
      ;; Copyright (C) YEARS  The Magit Project Contributors
      instead of
      ;; Copyright (C) YEARS  The Magit Project Developers
      "Developers" might be read as to imply that only the core team hold the
      copyright.  But that is not the case, since nobody assigned their rights
      to us.  Everyone who ever contributed holds the copyright for their
  19. 13 Jan, 2015 1 commit
  20. 30 Nov, 2014 1 commit
  21. 27 Oct, 2014 1 commit
    • Jonas Bernoulli's avatar
      magit-{status,init}: fix and improve · 23f4f8dd
      Jonas Bernoulli authored
      When `magit-init' or `magit-status' is used to create a new repository,
      then directory chosen by the user likely does not exist.  In that case
      we cannot use `magit-get-top-dir' because that only works in existing
      directories.  Create `magit-toplevel-safe' to use when it is unknown
      whether the directory in question exists.
      In `magit-status' cleanly separate the "initialize" special case from
      the rest, previously it was not possible to create a new repository
      inside another using this command.  As a result this command is no
      longer suitable for non-interactive use and `magit-status-internal'
      has to be factored out.
      In `magit-init' create its parent directory before creating the new
      repository.  Refactor some more.
      Improve the doc-string.
  22. 09 Oct, 2014 3 commits
  23. 17 Sep, 2014 1 commit
  24. 03 Sep, 2014 1 commit
  25. 16 Aug, 2014 1 commit
  26. 29 Jul, 2014 1 commit
  27. 20 Jun, 2014 1 commit
    • Remi Vanicat's avatar
      Correct a typo in magit-list-remote-branch-names · 7817ca88
      Remi Vanicat authored
      match-string want an integer as first argument.
      Replacing the "and" by a "when" because the second value is not a truth
      Added also some test on the different function that should list branch.
  28. 09 Jun, 2014 1 commit
  29. 30 May, 2014 1 commit
  30. 02 May, 2014 1 commit
    • Jonas Bernoulli's avatar
      magit-stage-modified: new command, replacing magit-stage-all · ac175ea9
      Jonas Bernoulli authored
      The old command was badly named.  It uses `git add', which happens to
      have an argument `--all', but the command did not use that (except with
      a prefix argument), it used neither `--all' nor `--update'.
      The new command uses `--update', so now deleted files are also removed,
      in addition to changes to modified files being added.  Because the new
      command also takes a prefix argument which means to use `--all' it is
      not named `magit-stage-update' but `magit-stage-modified', which is
      sufficiently distinct from the names of the arguments to `git add',
      which might or might not be used.
  31. 11 Apr, 2014 1 commit
  32. 09 Apr, 2014 1 commit
  33. 04 Apr, 2014 2 commits
    • Jonas Bernoulli's avatar
    • Jonas Bernoulli's avatar
      clarify distinction between a section's type and value · cc0745d3
      Jonas Bernoulli authored
      Originally there were these three related but distinct slots: `type',
      `info', and `header'.  The last one was unnecessary and only existed
      because someone got confused about the distinction between the other
      two and was removed when I last refactored the section code.
      Now the time has come to make the distinction between the "type" and the
      "info" more clear.  We start by using the term "value" instead of
      "info", because that's what it is.
      There are two categories of sections: those that represent a thing, and
      those that serve as containers for other sections.  A section can fall
      into both categories.
      Every section has a type, a symbol.  Sections that represent a "thing",
      such as a file or hunk also have a value, usually a string (but that is
      not enforced).  Sections that only serve as containers do not have a
      value (i.e. the value is nil).
      Together the type and value of a section and the types of recursively
      all parent sections can be used to uniquely identify a section within
      a buffer.  For container-only sections the types are sufficient because
      there always is only one section of a particular container-only
      type.  (That is a convention and isn't enforced.)  Meanwhile there can
      be many e.g. `commit' sections and to tell them apart the values (here
      the commit hashes) also have to be compared.
      Sadly someone got confused along the way and started to use the just
      the values (then more ambitiously called "info") to identify sections.
      Since many sections don't actually have a value the result was that
      for sections that would otherwise not have a value, the type was stored
      in the `value' slot as well.
      This commit adds function `magit-section-ident' which returns a list of
      conses that can be used to uniquely identify a section as described.
      Function `magit-find-section' is changed to use that value, instead of
      using the value-except-when-it-actually-is-the-type.  For sections that
      have no value the value of the `value' slot now is nil.
  34. 30 Mar, 2014 2 commits