Completely overhaul checkemv2.0

Remove error-prone inode detection and just do the size-then-checksum
algorithm, and use better data structure methods for the same.

1 files changed, 14 insertions, 12 deletions

diff --git a/README.markdown b/README.markdown
index d793349..795d0a5 100644
--- a/README.markdown
+++ b/README.markdown
@@ -2,23 +2,25 @@ Checkem
 =======
 
 Find duplicate files efficiently, using Perl on GNU/Linux, and maybe other
-systems. Requires Perl core include `Digest::SHA`.
+systems. Requires Perl core including `Digest::SHA`.
 
-Takes an optional parameter of the directory to check, otherwise assumes you
-mean the current directory.
+Requires at least one directory argument:
 
-    $ checkem
-    $ checkem /home/tom
+    $ checkem .
+    $ checkem ~tom ~chantelle
+    $ checkem /usr /usr/local
 
-`checkem-fixem`
----------------
+Caveats
+-------
 
-If you dump the output of `checkem` to a file, assuming you have no newlines in
-your filenames, you can run `checkem-fixem` over it to output some shell script
-to perform hard-linking to remove redundancies.
+This does not presently detect hard links, and will report them as duplicate
+files. The problem with doing this is inode integer collisions if your search
+includes multiple filesystems. I haven't yet figured out how to fix this in a
+way that should work on most Unix-like systems.
 
-Make sure you read its output really carefully before you pipe it into `sh`.
-The script itself requires Bash for `%q`.
+There used to be a script `checkem-fixem` included with this distribution that
+generated shell script to make hard links for you, but it wasn't safe because
+it couldn't know whether a hard link would even be possible.
 
 License
 -------