Skip to content
Commits on Jul 3, 2009
  1. @bitprophet

    Cut 0.9 beta 1

    bitprophet committed Jul 3, 2009
  2. @bitprophet
  3. @bitprophet
  4. @bitprophet
  5. @bitprophet

    Big TODO update.

    Removed completed items, merged some others, clarified yet others.
    
    At this point, not seeing anything blocking a feature freeze in prep for
    0.9 beta 1.
    bitprophet committed Jul 3, 2009
  6. @bitprophet

    Minor language/clarity tweaks to docs.

    Most interestingly, removed mention of what the "current stable version"
    is so that dev docs aren't schizophrenic.
    bitprophet committed Jul 2, 2009
  7. @bitprophet
  8. @bitprophet
  9. @bitprophet

    Add "line_only" option to get_version().

    Also update test to match, and tweaked version numbers to be more up to
    date. Not like that really matters, of course, but there never was a
    Fabric 0.2...
    bitprophet committed Jul 2, 2009
Commits on Jun 22, 2009
  1. @bitprophet
  2. @bitprophet

    Attempted fix/reorganization of authentication code.

    Problem: output_thread()'s sudo password handling was not trying with a
    pre-entered password before prompting the user, leading to at least one prompt
    per system when used on multiple systems, even when the stored password would
    have worked fine on all.
    
    Gist of solution: go back to explicitly checking for both normal and "sorry,
    try again" sudo prompts, then always try sending a stored password, unless the
    latter prompt was found.
    
    Existing related tests all dealt with prompt_for_password's old behavior of
    checking output for the password prompts. That responsibility was moved outside
    the function, so the tests no longer apply.
    bitprophet committed Jun 21, 2009
Commits on Jun 21, 2009
  1. @bitprophet
  2. @brosner @bitprophet

    Handle carriage returns in output_thread correctly.

    Signed-off-by: Jeff Forcier <jeff@bitprophet.org>
    brosner committed with bitprophet Jun 10, 2009
Commits on Jun 17, 2009
  1. @bitprophet
Commits on Jun 16, 2009
  1. @bitprophet
  2. @bitprophet
Commits on Jun 14, 2009
  1. @bitprophet
Commits on Jun 13, 2009
  1. @bitprophet
  2. @bitprophet
Commits on Jun 8, 2009
  1. @bitprophet
Commits on Jun 7, 2009
  1. @bitprophet
  2. @bitprophet

    Fix race condition in run/sudo.

    See b9a7c9d for how the problem was (at least most recently) introduced.
    
    Since this change here is not quite a revert of that earlier commit (note that
    we kept the recv_exit_status check *before* the thread joins) I am hoping that
    this change does not introduce more bugs of its own; this part of the code has
    a very long history of back-and-forthing :(
    
    However, I verified that the race condition definitely existed beforehand
    (Fab task resulting in single line of output was run a few dozen times
    in a row; almost half of the time the line never showed up) and that this
    change fixes it (now every single invocation results in the expected output).
    
    Thanks to Evan Jones for catching this and suggesting the move of
    channel.close().
    bitprophet committed Jun 6, 2009
Commits on Jun 6, 2009
  1. @bitprophet
  2. @bitprophet
  3. @bitprophet
Commits on Jun 5, 2009
  1. @bitprophet
Commits on Jun 2, 2009
  1. @bitprophet

    Update upload_template to correctly handle backing up target director…

    …ies.
    
    This is related to the previous change to this function earlier today.
    bitprophet committed Jun 2, 2009
  2. @bitprophet
  3. @bitprophet
  4. @bitprophet
  5. @bitprophet
  6. @bitprophet

    contrib.files.append may now take a list of strings if desired.

    Also shuffles around where the warn/hide with statement is used (should really
    have been inside contains() to begin with) and updated to work correctly with
    "empty" append lines -- i.e. one may use append to tack on whitespace lines if
    desired.
    bitprophet committed Jun 2, 2009
  7. @bitprophet

    Update upload_template docstring, and back out some recent changes.

    For some dumb reason I had updated the behavior of upload_template so that it
    no longer worked as advertised. In retrospect it really should work as
    advertised, namely that the "destination" string is given verbatim to the call
    to "cp".
    
    This gives more power to the user, and the user may explicitly tack on
    a trailing slash if they want "cp" to copy a file into a folder, as normal.
    bitprophet committed Jun 2, 2009
Commits on May 25, 2009
  1. @bitprophet

    Handle exceptions with strerror attributes that are None instead of s…

    …trings.
    
    Thanks to Evan Jones for the catch.
    bitprophet committed May 25, 2009
  2. @bitprophet

    Minor cleanup to package init and setup.py.

    Tested with fresh virtualenvs doing setup.py install|develop|build. Seems to
    work OK in all of them.
    bitprophet committed May 25, 2009
Something went wrong with that request. Please try again.