Recompile task after runDeps to reload envs set by deps

[hori-ryota]

fix #1381

I would like to create an environment variable file using deps, read it with dotenv, and use it in cmds.

tested by followings

1. generate Taskfile.test.yml (copy from dotenv files are loaded before task dependencies are ready #1381 )

version: 3

tasks:
  make-env:
    cmds:
      - echo "MY_VAR=MY_VALUE" >> ./.env
  my-task:
    deps:
      - make-env
    dotenv:
      - ./.env
    cmds:
      - echo "MY_VAR = $MY_VAR"

2. run following

rm -f .env && go run ./cmd/task -t Taskfile.test.yml my-task

result:

(before):

task: [make-env] echo "MY_VAR=MY_VALUE" >> ./.env
task: [my-task] echo "MY_VAR = $MY_VAR"
MY_VAR =

(after):

task: [make-env] echo "MY_VAR=MY_VALUE" >> ./.env
task: [my-task] echo "MY_VAR = $MY_VAR"
MY_VAR = MY_VALUE

[hori-ryota]

@andreynering @pd93 If this approach doesn't seem optimal, I am considering setting up a custom system to reload dotenv and execute the tasks. I would appreciate your thoughts on whether this might be a viable direction.

[hori-ryota]

By the way, currently, dotenv: is checked before deps: and used in deps: executions, whereas env: seems to be used only in cmds: .

[hori-ryota]

@andreynering @pd93 What do you think about this pull request?

[hori-ryota]

fmmmmm.....

[andreynering]

Hey @hori-ryota, I know we've been slow on reviewing pull requests. We all have full time jobs and a real life happening. I added myself as a reviewer to remember to take a look when possible.

[hori-ryota]

Thank you! Please let me know if there's anything else I should do or improve. I really appreciate this excellent tool!

[andreynering]

@hori-ryota In the meantime, can you rebase this PR with main?

[hori-ryota]

rebased.

[andreynering]

This introduces a small performance penalty, and is something that is not needed 99% of the time.

I'm not so sure that it's worth doing for this rare edge case. In this case, it could be better to just run both tasks separately so you ensure that the .env file exists before running the actual task.

task setup && task actual-task

I suppose that your .env should be generated only once, right? I wouldn't expect a .env file to have to be generated on every run.

[hori-ryota]

Thank you for your feedback. In a large monorepo like ours, each module has its own Taskfile and setup. Being able to run a single task and have all necessary setups execute via dependencies greatly simplifies our workflow. Without this, we’d need to run separate setup tasks for each module, making management and documentation more complicated.

Additionally, our .env requirements change frequently as the team iterates. Running a lightweight, idempotent setup every time ensures our environment is always up-to-date and stable, reducing errors and confusion in the long run. This approach helps keep the entire project’s operations more efficient and maintainable.