Taskfile dynamic command variables

Taskfile is amazing and I use it as the task runner for all my projects.

Why?

• no .env pollution between operations; define envs per task
• concatenate many commands together, for instance, regenerate tailwind when files change and restart server
• simplify my life by reducing the amount of typing

There more reasons but those are the highlights.

One thing that I only learned a couple months ago is that you can supply dynamic variables to tasks.

As example, I have a bunch of python files that need various environment variables - each file needing a different one to the next. Thats really annoying to configure manually. Here’s a simplistic version using Taskfile.

Here’s how I’d do that in a Taskfile

version: "3"

vars:
  PG_PORT: 5672

tasks:
  benchmark:*:
  env:
    PG_HOST: foo.bar
    TRACEMALLOC: 1
  vars:
    ARG_1: "{{index .MATCH 0}}"
  cmds:
    - python -m {{.ARG_1}}

  production:*:
  env:
    PG_HOST: prod.foo.bar
    TRACEMALLOC: 0
  vars:
    ARG_1: "{{index .MATCH 0}}"
  cmds:
    - python -m {{.ARG_1}}

Now I can run my benchmark or production code for any python module by passing the name of the module like so:

task benchmark:pg_connection
# or
task production:pg_connection

Thats a really silly example, heres one I actually use.

version: "3"

tasks:
  nats:cname:*:
  desc: Publish a message to NATS to trigger a CNAME lookup
  vars:
    ARG_1: "{{index .MATCH 0}}"
  cmds:
    - nats pub scanners.dangling_cname.{{.ARG_1}}

Now I can publish messages to NATS easily in development to the queue I want with any domain name I want to target.

task nats:cname:google.com
task nats:cname:tesla

This has been a massive quality of life improvement. Its possible to use more than one ARG here too but for simplicity I’ve left it at just one.

Tags:

#taskfile