summaryrefslogtreecommitdiff
AgeCommit message (Collapse)AuthorFilesLines
2010-05-01Add dependency on python-progressbar.Lars Wirzenius2-1/+8
2010-04-30Use progressbar library instead custom code.Lars Wirzenius1-67/+80
2010-04-29Only investigate regular files, ignoring e.g. symlinks.Lars Wirzenius1-6/+9
Also, sort filenames, it's nicer for people.
2010-04-29Progress reporting fix.Lars Wirzenius1-0/+6
2010-04-29Add progress reporting when finding duplicates.Lars Wirzenius1-8/+21
2010-04-25New package version.Lars Wirzenius1-0/+6
2010-04-25Add note about automatic recursion into directories to manual page.Lars Wirzenius1-1/+3
2010-04-25Add missing file arguments to manual page.Lars Wirzenius1-0/+1
2010-04-25Bump Standards-Version. No other changes required.Lars Wirzenius1-1/+1
2010-04-25Declare Debian source package format.Lars Wirzenius1-0/+1
2010-04-25Install files properly.Lars Wirzenius2-0/+5
2010-04-25Run testit in "make check".Lars Wirzenius1-6/+2
2010-04-25Add preliminary Debian packaging.Lars Wirzenius7-0/+55
2010-04-25Add a rudimentary manpage.Lars Wirzenius1-0/+49
2010-04-25Remove old FIXME, add new one.Lars Wirzenius1-6/+7
2010-04-25Compare files chunk-by-chunk instead of checksums for whole files.Lars Wirzenius1-22/+53
This should be way more effective when there are a lot of files. We now stop comparing as soon as there is a difference. We can't handle it if too many files of the same size need to be opened, but that's a problem for another day.
2010-04-25Only report hardlinks as duplicates if there's another identical file as well.Lars Wirzenius2-1/+4
If foo and bar are hardlinks to the same inode, and foobar is a file with the same content, report all three. However, if there is no foobar to be reported, do not report anything.
2010-04-25Rename test case to be clearer.Lars Wirzenius1-1/+1
2010-04-25Do not follow symlinks when statting.Lars Wirzenius2-13/+23
Report all hardlinks to the same file as duplicates. This is probably stupid, but avoids a bug: if foo and bar are hardlinks to the same inode, and foobar is not, but has identical content, then previously it would be random whether foo or bar was reported as the hardlinks. Further, only one of foo and bar would be made into a hardlink with foobar. So the next run would report the other one as a duplicate.
2010-04-25Add test for symlinks.Lars Wirzenius1-0/+12
2010-04-25Replace hardcoded list of tests with some introspection.Lars Wirzenius1-5/+5
2010-04-25Add test for two hardlinks.Lars Wirzenius1-0/+13
2010-04-25Add rudimentary blackbox testing script.Lars Wirzenius1-0/+97
2010-04-18Report read speed.Lars Wirzenius1-4/+12
2010-04-18Do not read the whole file into memory when doing a checksum.Lars Wirzenius1-1/+9
2010-04-18Fix progress message wording by removing needless 'bytes' word.Lars Wirzenius1-1/+1
2010-04-18Improve progress reporting.Lars Wirzenius1-22/+32
2010-04-18Do not compute checksums when only one file has a given size.Lars Wirzenius1-13/+20
2010-04-18Bugfix: store dev/ino for each pathname, not just for first file with a ↵Lars Wirzenius1-8/+11
given size.
2010-04-18Make progress reporter's .finish() method really flush the output.Lars Wirzenius1-0/+1
2010-04-18Add some progress reporting.Lars Wirzenius1-1/+16
Add FIXME comment with idea for optimization. I am not going to implement it right now, since right now I can afford to wait for the script (it runs in the background, while I do other things).
2010-04-18Provide some progress reporting for the collection phase.Lars Wirzenius2-4/+36
2010-04-18Add option to replace names with hardlinks to duplicate content.Lars Wirzenius2-3/+32
Add Makefile to run 'make check'.
2010-04-18Initial import.Lars Wirzenius5-0/+73