grumpper Posts - Answer Overflow

grumpper

•Created by grumpper on 9/19/2024 in #help

What environment anad shell are task commands executed in?

I am trying to evaluate the result of a command which pipes through tee so that output of the command is logged to the STDOUT and saved in a file at the same time. The idea is to then invoke a script that logs the outcome of the command depending on the exit code in a file.

  task:
    cmds:
      - |
        pylint --recursive=y . | tee reports/python/pylint.log
        logger.sh $?

  task:
    cmds:
      - |
        pylint --recursive=y . | tee reports/python/pylint.log
        logger.sh $?

BUT: If there are pylint findings the exit code is 16 so a non-zero one but the end result of the pipe is 0 (due to the tee command)... Now how can I catch that in taskfiles? If I try with $? it's 0 ... in order to correctly catch it I must use set: [pipefail]. But if I use it, the task errors out and never reaches the script that logs the result (not to mention that it interrupts the whole task execution)... Ok so I will ignore the error instead using ignore_error: true. Still problem though as ignore_error is not actually ignoring anything - it is instead overriding the exit code to be 0 instead of 16. More appropriate name should have been override_error 🙂 Ok so let's use the appropriate way of handling this by utilizing ${PIPESTATUS[0]} (${pipestatus[1]} in zsh):

  task:
    cmds:
      - |
        pylint --recursive=y . | tee reports/python/pylint.log
        logger.sh ${PIPESTATUS[0]}

  task:
    cmds:
      - |
        pylint --recursive=y . | tee reports/python/pylint.log
        logger.sh ${PIPESTATUS[0]}

This way the execution will not fail but I will still catch that the exit code of the pylint command is non zero. Awesome! But whatever environment or shell this is running in does not have pipe status being set at all... In fact if you simply run set it outputs nothing... Soo... how can I handle that error handling mess?

4 replies

TTask

•Created by grumpper on 9/18/2024 in #help

Continue on error

Is there a functionality that could make a task continue execution even there is an error ? Context: I am trying taskfile to do CI checks. So I am currently doing python (pylint & bandit) & terraform (tflint & trivy) checks. I migrated the checks for each tech in an included taskfile like this:

includes:
  python_tasks:
    taskfile: ./tasks/python.yml
  terraform_tasks:
    taskfile: ./tasks/terraform.yml

includes:
  python_tasks:
    taskfile: ./tasks/python.yml
  terraform_tasks:
    taskfile: ./tasks/terraform.yml

So in each of these taskfiles each check is a separate task. For example the pylint checks can be executed as python_tasks:run_pylint, etc. Naturally I setup the default to depend on each check (example with the python taskfile):

  default:
    run: once
    deps:
      - run_pylint
      - run_bandit

  default:
    run: once
    deps:
      - run_pylint
      - run_bandit

Now I have the below tasks defined to triger the included ones:

tasks:
  python_checks:
    run: once
    cmds:
      - task: python_tasks:default

  terraform_checks:
    run: once
    cmds:
      - task: terraform_tasks:default

  default:
    run: once
    deps:
      - python_checks
      - terraform_checks

tasks:
  python_checks:
    run: once
    cmds:
      - task: python_tasks:default

  terraform_checks:
    run: once
    cmds:
      - task: terraform_tasks:default

  default:
    run: once
    deps:
      - python_checks
      - terraform_checks

So when I run the task I want all the checks to happen (regardless if one of them fails). Again example with the python tasks: if the run_pylint check fails because there are pylint findings, the run_bandit one should still execute. And the same logic for the parent taskfile: if the python_checks task fails the terraform_checks one still executes. What I have seen so far is the whole execution halts on error so even on python checks issues, the terraform checks also stop. Which defeats the whole purpose... Now I could work around all that by setting ignore_error: true on every task but then when this is all run in CI/CD everything will just exit with 0 (no matter the findings) so all tests will be marked successful (even if they weren't)... What are my options here? TL;DR: How can I make deps finish even when errors in one of them occurs?

1 replies

TTask

•Created by grumpper on 9/16/2024 in #help

dotnev variable value calculation

Hello ! I canot find anything in the docs so I decided to ask here: Is there some sort of prioirity in the dotenv declaraion? Like if I declare it to be ['.env', '$HOME/.env'] and have DEPLOY="true" in .env but DEPLOY="false" in $HOME/.env what will be the value of DEPLOY in the end? - true because .env is first so its values take precedence? - false because $HOME/.env is last so its value override the rest?

7 replies

Gaming

Programming