Display existing hard links in result sets This is a breaking change in the plain text output, but somehow the list displayed in case of existing hard links was arbitrary, since the all the hardlinked filenames were not displayed. Here's a sample JSON result with this patch: { "file_size": 9216, "paths": [ "test_tree/f09-1_5.raw", "test_tree/f09-4_5.raw" ], "links": { "test_tree/f09-1_5.raw": [ "test_tree/f09-2_5.raw", "test_tree/f09-3_5.raw" ], "test_tree/f09-4_5.raw": [ "test_tree/f09-5_5.raw" ] } } Here the 5 files have the same contents, but there are two hardlink groups: "test_tree/f09-1_5.raw" "test_tree/f09-2_5.raw" "test_tree/f09-3_5.raw" are hard-linked, and "test_tree/f09-4_5.raw" "test_tree/f09-5_5.raw" are hard-linked. Here's the same set displayed With the regular text output: Group #5 (2 files * 9216 bytes): test_tree/f09-1_5.raw test_tree/f09-2_5.raw test_tree/f09-3_5.raw test_tree/f09-4_5.raw test_tree/f09-5_5.raw (The link file names are indented using 1 space character.)

Improve previous commit (JSON output)

Do not display Dupes title when using JSON output

Rename JSON size fields Make the field names more explicit...

JSON: Add fields 'total_size' and 'total_size_h'

JSON output: NumberOfSets was not set

Move formatSize() to output.go

Cosmetic changes

Some code refactoring Move logic out of main().

Add JSON output

Mention github mirror in the comments

Split goduf.go into several files

Add go.mod

Travis: Add testing with Go v1.11

Update README

Add one more example

Add README badges

Add Travis file

Add README.md and gitignore for the Github mirror

Add missing license file

Move comment

Add installation command line example

Add an assertion on constant variables

Experimental optimization This is actually an old patch; it tries to batch checksum computation, which is slightly helpful with high number of files and mecanical drives.

Improve file sorting

Better documentation

Remove duplicated code

Change comments so that build constraints still work

Add copyright notices to the arch-specific files

Sort results Sort groups by size of a group file, and files from a same group by path.

Cosmetics (go fmt and comments...)

Get rid of the now useless sizeClass structure

Code cleanup

Fix hard link detection The previous algorithm was assuming all files of a size group, or none of them, would be hard links. For size groups with both hard links and regular files, the detection could fail.

Sort files by path when sorting by inode number is not supported

Reduce code duplication related to computeSheduledChecksums()

Small optimization

Schedule checksum computations so that we reduce hard drive seeks The previous approach was not working.

Refactor checksum functions to reduce code duplication

Cosmetics - move function

Refactor the checksum part Instead of doing #1 partial checksums #2 full checksums when needed we do the full checksums as soon as partial checksums match. This should reduce disk seeks.

Update .hgignore

Add descriptions to exported functions

s/createSizeHash/initialCleanup/

Remove useless optimization

Goduf - initial version-controlled revision