Oops ... the Archive function has a bug in TIMT, as it seems - it arcs itself into its own archive. I'll have to fix that before we start to get fractal files.
OK, I fixed it, but this is still somewhat unsatisfying. The problem occured when I introduced drag-and-drop support which forced me to drop my own implementations in favor of available Java mechanisms, because anything else would have been extremely ugly.
The problem is that I implemented archiving by creating an empty archive file first, then copy the selected files into the archive. By creating the archive file, the directory becomes longer by one element, but there is still a selection model behind, and it is not updated automatically. If the archive file happens to be sorted into the directory amidst the selected files, the loop over the selected elements will pick the newly created archive file. So what I did now is to first make a copy of the selected file references, then add the archive file, then use the copy of the references.
I noticed a similar phenomenon with other Java-based tools, for example, ProjectX (used for video editing). So maybe this is a Java-related issue that requires at least some more consideration.
The fix will soon be available; I intend to add some more features for the official 2.4 release.