Skip to content


Subversion checkout URL

You can clone with HTTPS or Subversion.

Download ZIP
Commits on Jan 27, 2013
  1. version bump

  2. @Pablo-Merino
Commits on Apr 9, 2012
  1. bumpity bump

  2. completely remove rubycocoa backend, as it hasn't worked correctly si…

    …nce 10.6 and rb-fsevent is a better solution regardless
Commits on Jan 6, 2012
  1. bumpity bump bump.

Commits on Jan 4, 2012
Commits on Dec 2, 2011
  1. minor changes that attempt to make backends a little more flexible

    1) autoload code as needed
    2) wait until the last possible moment to define FSSM::Backends::Default
    3) provide set_backend(symbol, value) to make it easier to opt out and use polling, as well as plug in your own custom backend or backend handling logic
    4) handling of platform specific dependencies via spec.extensions + rake
    I'm not completely convinced on point number four. Though, if gem install option args get passed to rake along the way, it'd be a non-issue to have it do nothing by default and only install/update backend support if asked.
Commits on Nov 5, 2011
  1. @mattbrictson

    Use RbConfig instead of deprecated Config

    mattbrictson authored
    This fixes the following warning under ruby 1.9.3-p0:
    `fssm/lib/fssm/support.rb:40: Use RbConfig instead of obsolete and deprecated Config.`
Commits on Apr 21, 2011
  1. version bump

Commits on Mar 21, 2011
  1. note to self: while productive, writing software between 2 and 3 in t…

    …he morning because you can't sleep may have unexpected consequences.
  2. add initial support for rb-fsevent multipath (pre-metadata). once ful…

    …l metadata support is added to rb-fsevent, these temporary hacks and lookups will be completely unnecessary. additionally, do NOT use rubycocoa on osx lion... it segfaults quite enthusiastically right now.
Commits on Jan 26, 2011
  1. fsevent is resolving symlinks in paths, which causes all kinds of cra…

    …zy things to happen if you're not expecting it. as a fix, we need to do the same and resolve the actual filesystem path without symlinks ahead of time.
  2. have inotify watch for close_write instead of modify. this should res…

    …olve compass issue 263 and FSSM issue 21.
  3. reformat

Commits on Jan 22, 2011
  1. bump version

Commits on Jan 20, 2011
  1. random whitespace OCD

Commits on Jan 15, 2011
  1. rb-fsevent support

Commits on Dec 21, 2010
  1. I need to figure out a real plan of attack for cleanly handling virtu…

    …al paths, like when a path exists but is within a jar file. Until then, lets at least error out early with a message that makes sense.
Commits on Dec 18, 2010
  1. oops.

Commits on Dec 11, 2010
  1. @Twisol

    Converted FSSM::Pathname to subclass Pathname instead of String, but …

    Twisol authored
    …using the same implemetation of #segments as before.
  2. quick fix for rubinius that's more of an ugly hack. it will, however,…

    … give something to profile against when investigating Jonathan Castello's solution: (
Commits on Aug 29, 2010
  1. @tkareine

    Use do/end block, so code in related methods looks similar

    tkareine authored
    So that the methods #created, #deleted, and #modified in Directory look
    similar to each other.
  2. @tkareine

    1.8.6 fix: avoid external enumerator in #tag_entries

    tkareine authored
    Ruby 1.8.6 does not support external enumerators. So, change the code in
    Directory#tag_entries to have the result hash in explicit variable and
    let the processing in #each_pair block to modify the hash via the
  3. @tkareine

    Sort.reverse is clearer than sort with a block

    tkareine authored
    Thanks to Nathan Weizenbaum (nex3) for the comment.
  4. @tkareine

    Avoid line-continuation, put dot in prev line

    tkareine authored
    Thanks to Nathan Weizenbaum (nex3) for the comment.
Commits on Aug 25, 2010
  1. @tkareine

    Pass file type argument to callbacks if :directories

    tkareine authored
    If :directories option is enabled, callbacks receive a third argument
    specifying the type of the file affected by the operation causing the
    callback. The file types are either :file or :directory.
    If :directories option is disabled, no third argument is passed to the
    callbacks; the monitored files are all non-directories anyway.
    With file creation and modification events, you can use
    `` to determine the file type. However, this is not
    possible with file deletion, since the file does not exist anymore. This
    is the motivation for the change.
  2. @tkareine

    Call callbacks in predictable order

    tkareine authored
    If a directory structure is modified, it is useful for the user of FSSM
    to receive the callbacks in predictable order.
    For example, let's consider an application that monitors files in a
    working directory and attempts to keep the directory contents
    synchronized to a remote host with SCP. If a subdirectory with a file
    gets deleted in the working dir, it is useful for the app to know that
    the file gets deleted first and the subdirectory second. This way, in
    the remote host, the app can remove the file first before attempting to
    remove the (non-empty) subdirectory.
Something went wrong with that request. Please try again.