Sawyer Shepherd's Blog

Can Version 0.1.0

I have released my program Can on RubyGems and the Arch User Repository.


I have a feeling this post is going to get hairy with a program name like Can!

Can is an idea for a program I’ve had for a very long time. As the trash specification says, “An ability to recover accidentally deleted files has become the de facto standard for today’s desktop user experience.” But what about a user’s command-line experience? I’m sure we’ve all accidentally annihilated a file or two or three or four with rm. I know I certainly have, and after irreversibly deleting an entire unpublished project of mine, I set out to make a tool that would bring the de facto experience of the desktop to the command line to prevent this from ever happening again.


Put simply, Can is nondestructive rm. Instead of unlinking files, Can moves them into a hidden trash directory (by default ~/.local/share/Trash/files) and creates a metadata file for later recovery.

Option parity with rm is a primary goal for the program. Unlike rm, however, Can is not intended for use in scripts. At the moment, it has a few blocking confirmation prompts to prevent data deletion–option parity only provides a smoother user experience.

While it is not intended for use in scripts, having similar options to rm allows for a smoother user experience.


My initial idea was to program Can in C. In theory, it’s a very simple program–just move a file to a trash directory and create a metadata file. In practice, the intricacies that come with making a tool with a user-friendly CLI made using C a headache.

I also experimented with the idea of writing Can in Bash by stringing together existing shell commands, but there was too much required functionality to implement easily. I would have to fill in some big cracks with my own scripting, and I hate writing in Bash. Due to my frustration, I shelved the program for many months.

During a school break, inspiration to finish the program struck. This time, I settled on Ruby because of my previous frustrations with slow development speed. With Ruby’s expressiveness, I quickly implemented basic features, achieved partial option parity with rm, and set up continuous delivery to RubyGems and the AUR (tutorial on how to do Ruby CD coming soon!).


Here are a few examples of how to use Can:

Move files to the trashcancan foo.txt bar.txt
Move files and directories to the trashcancan -r foo.txt bar.d
List files in the trashcancan -l
List files in the trashcan that match a regex patterncan -l '^foo'
View .trashinfo metadata of select files in the trashcancan -n foo.txt bar.d
Untrash filescan -u foo.txt bar.d
Empty select files from the trashcancan -e foo.txt bar.d
Empty everything in the trashcancan -e

Can has many safety nets built in to prevent data deletion. Trashing files of the same name has been safely handled to prevent overwriting data. Additionally, Can will not untrash a file to a path that already exists. And just like rm, Can has an -i option to wait for user confirmation before trashing a file.

To see the full usage options, run can --help.

Faults and Future Plans


A program like Can that is most likely to be run often on small groups of files benefits most from startup time optimization. Choosing Ruby may have been a good idea to boot programmer productivity, but it’s awful for startup times.

Frankly, I’m not if there’s any simple solutions to slow startup times. The nuclear option is to port the entirety of the program to Crystal. To put it simply, Crystal seeks to be compiled Ruby. My description does not do this very neat language justice, so please see the homepage for yourself!

Test-Driven Development

Can runs against no tests. I know, I know, it’s a terrible developer practice. I’ve been working to find a testing method I like. So far, the best option seems to be running Minitest tests with a Rake task on build.