I was tired of cd'ing arround my filesystem. I could've just defined some shell aliases on my .shellprofile but they are awkward to mantain. As such I made a simple utility that allows one to move to recorded directories.

jmp keeps a list of files on a jumptable, when you run:

jmp somefile

It will fuzzily search for a file on that table by comparing the somefile string to the basename of the file. The comparison algorithm is such that you probabily only need to type som and it will match.

You can quickly add any path to your jumptable by running:

jmp -a path/to/your/directory/here

You can remove a path from the table with:

jmp -d path/to/remove

You can query the location of the table with:

jmp -T

On Linux the jumptable will be looked up on one of the following places:

$XDG_DATA_HOME/jumptable
$XDG_CONFIG_HOME/jumptable
$HOME/.jumptable

For other operating systems it depends on this library. In any case, you can always override the default path with:

jmp -t /path/to/table

And thats it.

The distance between two strings is the cheapest path, by deleting and inserting characters where, deleting a character at position k costs 1/k. Insertions cost the same as if you deleted the character you just inserted. This algorithm favors matching prefixes of a string, the distance between ta and table is just 0.78, but between ble and table it is 1.50 despite having a greater number of character matches.

If you want to test the comparison algorithm you can run

jmp -c string1 string2

Aditionally you can see how your pattern measures against your table by running

jmp -C patternhere

jmp jumps into the directory by execve()'ing your $SHELL in that directory. However the previous shell process will still be running, waiting for the new shell to exit. This means that once you jmp you can exit to return to where you started. You can query how deeply you have jumped with the JUMP_DEPTH environmental variable.