Release Notes: The missing source for the symharden tool was added.
Release Notes: CTRL-C is now caught to free all the duplicates discovered so far. Several changes improve platform portability, e.g. for Apple Mac Version 10.6.0 i386, where freedup was tested and works fine. Filesystem errors (e.g. in /sys) are now caught and warnings are printed instead of assertions. Several other bugfixes and contributions were applied, e.g. one to fix a memory allocation bug. A new minor tool called symharden was added to convert softlinks into hardlinks wherever possible.
Release Notes: This release removes leftover debugging statements. It also provides HTML documentation with the RPM file. Additionally, the installation preparation for the new Web interface can be tested. But be aware there is currently no usable Web interface to start freedup.
Release Notes: When linking files, it is unavoidable to change the modification and the access time of the directory the target belongs to. For cases where this leads to additional (unwanted) actions, use the new -T option to make freedup keep the modification times of the affected directories. This option is added and documented. Makefile.tests is now more easily readable and reduced in size.
Release Notes: This release no longer needs external commands to perform its tasks. If no "find options" are passed using -o, the internal tree scanning is used. It is supposed to run faster, but this could not be proven. The only visible advantage is a progress indicator.
Release Notes: Some variables were not initialized under certain circumstances, which could make freedup fail with SIGSEGV in interactive mode. In interactive mode since version 1.4, all identical files were displayed, whether already linked or not. This now obeys the rules, e.g. given by "-H". The linkage reports in interactive mode are now supposed to be correct (non-zero) again. Statistics for deleted files (only possible in interactive mode) are added, too. If using "extra styles" in interactive and batch mode, you may influence linking direction by file size (i.e. by existence of tags).
Release Notes: The major bugfix concerns the (n+m) assertion. Checks were implemented so that no endless loops with the extra style mp4 should occur. A minor one is to read and write the environment variable "globalkey" (-k) as letter not as ascii code. Please note that you have to correct existing environments manually. The new feature that lists the existing environments should help you on that.
Release Notes: You now have full capabilities to determine link direction, which is particularily helpful when more then two files need to be linked. Also, the limitation that not more than 256 files may be identical no longer exists. (For efficient memory usage, four identical files are assumed and if more exist, memory allocation is doubled each time the boundary is reached.) The "-H" switch allows you to treat files that are already linked as duplicates, too. Without "-H", depending on the linking directions, the results can vary.
Release Notes: The option '-k' allows you to select the linking direction between all identical files. The source may be the oldest, the newest, thrst given file or the one that has the most links already. These rules can also be specified for each link cluster individually in interactive mode. As a side effect, the space save counting works more exactly, and the performance was degraded by an average of about 10%. This will be compensated by the default, which avoids hash functions now. You no longer need to add directories manually to the config file.
Release Notes: This release offers a much more efficient usage of the internal hash function. This new default algorithm consumes a bit more memory, but takes, in the worst case, about the same time as without hash calculation. The file reads are significantly reduced and in a test case the wait time for about 8000 files (48MB average) was decreased from 5 to 2 minutes. This is about the performance of dupseek without the risk of wrong partial results. The documentation was updated and a bug in statistics was fixed, too. Please note that with this version the "-#" option now requires an argument.