Eval outputs not captured if process exits early

Bug report

When running a test expression in a process that contains an eval output, I get an error.

Steps to reproduce the problem

The module file that causes the error is available at:
https://github.com/OlivierCoen/nf-chimera-detector/blob/2ed14de596506c016b62b11edb1f47736727d086/modules/local/find_chimeras/main.nf

Here is a test workflow that can reproduce the error.

process FOO {

    input:
    path file

    output:
    eval("ls")


    script:
    """
    [ -s $file ] || exit 0

    echo "I exist"
    """
}

workflow {

    main:

    ch_x = Channel.fromPath( 'myfile.txt' )
    FOO ( ch_x )


}

Program output

If myfile.txt exists and is not empty, I get no error.

Otherwise, I get:
ERROR ~ No such file or directory: <process workdir>/.command.env

Environment

nextflow info

  Version: 25.04.6 build 5954
  Created: 01-07-2025 11:27 UTC (13:27 CEST)
  System: Linux 6.1.0-37-amd64
  Runtime: Groovy 4.0.26 on OpenJDK 64-Bit Server VM 17.0.15-internal+0-adhoc..src
  Encoding: UTF-8 (UTF-8)

• Bash version: GNU bash, version 5.2.15(1)-release (x86_64-pc-linux-gnu)
  (I get the same with zsh 5.9)

Thanks in advance :)

Start by running the provided FOO workflow with an empty or missing input, then inspect the referenced modules/local/find_chimeras/main.nf example and the process-output handling it exercises. Compare the early-exit path with the successful path, focusing on the missing .command.env error. Done means an eval output is captured without an error when the process exits early.