This feature allows you to run fkill -v and then when searching for processes to kill, it will include the full proc command (including arguments)

This is useful when I have a daemonic process, like a dev web server, that I want to kill, but was launched with python start_web_server.py for example.

Fixes #13 Fixes #18

Fixes #62 .

Style and thresholds might be opinionated, but I tuned them to the point I was comfortable finding processes.

• processes are sorted by sum of cpu and memory usage(percentage), highest first
• in verbose mode memory and cpu usage is show when it is above 0.1% (minimal number we can get)
• in non-verbose mode memory usage is shown when it's above 0.5%, cpu usage shown when it's above 2%
• cpu usage is shown in dark red
• memory usage is shown in dark yellow

I have a node.js process listening on port 3000 and I can't kill it.

Ubunut 17.10 Node 8.9.3

Awesome stuff.

But Ctrl + C on my Mac leaves a residue of the fkill interface (which can only be cleared if I issue new commands or hit Enter for a few times to clear all those lines).

how to close fkill properly?

Forces a kill if the normal kill behavior fails. handleFkillError already includes functionality for handling the case of a failed kill, but fkill doesn't throw an error when it fails to kill a process, so I chose a different way to implement this feature.

Resolves #42

fkill takes about 150-300 ms to start up. Will be cool to see it under <100ms. (MacBook Pro 2016)

We need to kill a node process that runs a server (e.g. on port 8000) as part of our deployment script.

Using "fkill :8000" doesn't terminate the process, so we need "fkill -f :8000"

Unfortunately, fkill is returning an error code when there is no process running on port 8000. So we added the silent switch "fkill -f -s :8000"

This works, but gives an error message when no process has been found, which should not be the case using -s.

Is there a way to specify the signal to send to the process?

I understand that fkill is multi-platform but at least in Unix environment people regularly use both TERM and KILL, and possibly others like HUP

Fixes #49.

• I applied fuzzy search to verbose, name option both.
• I added a unit test for testing fuzzy search, I think it probably works in almost all cases, but not sure there is always no other such named process. In this case, the noopProcess could not be killed. Should I think about a better test method, or is it enough?

Will be cool to see CPU/mem usage as often a lot of same processes running with same name.

Will be cool to distinguish them.

Add option --force-timeout N/-t N, to force kill processes which didn't exit within timeout seconds.

In interactive mode,wait up to 3 seconds for processes to exit, then prompt user to force kill surviving processes.

Fixes #42 Requires sindresorhus/fkill#54

Fixes #34.

It seems working now but I think I need to a little more time to look into more error handling logics and test codes and lint (WIP). Thank you for all your efforts for maintaining awesome libs.

theres something wrong with the process search by name, in interactive mode fkill-cli can find and kill processes with BrowserStackLocal identifier but neither the cli or the npm module can kill that process as a variable

Having an issue resolving this warning of an unmet peer dependency when installing globally:

$ npm install -g fkill-cli inquirer@^6.0.0
/usr/local/bin/fkill -> /usr/local/lib/node_modules/fkill-cli/cli.js
npm WARN [email protected] requires a peer of inquirer@^5.0.0 || ^6.0.0 but none is installed. You must install peer dependencies yourself.

+ [email protected]
+ [email protected]
added 184 packages from 67 contributors in 6.88s

Installing the library, at least on Windows, throws the following warning:

warning fkill-cli > fkill > taskkill > execa > [email protected]: cross-spawn no longer requires a build toolchain, use it instead

I gather this is a package dependency deprecation warning, you may want to look into this.