User Guide: Capturing Standard Output should include example of renaming output file

[golharam]

http://www.commonwl.org/user_guide/05-stdout/index.html

The example shows how to capture stdout for a single tool to be output.txt. This example should also show how to rename the output file, or else having multiple tools writing stdout to output.txt will overwrite the previous tool.

[mr-c]

Welcome @golharam and thanks for this issue.

CWL workflow engines should automatically rename files that have the same file name if they are written to the same final output directory. The CWL reference runner cwltool and related implementations do this for sure.

[kinow]

@mr-c do we need to add a note about it somewhere in the user guide? (if so where would be a good place?)

Tested with a workflow and it does rename the duplicate file automatically.

cwlVersion: v1.2
class: Workflow

inputs: []
outputs:
  step_a_output:
    type: File
    outputSource: step_a/output
  step_b_output:
    type: File
    outputSource: step_b/output

steps:
  step_a:
    run:
      class: CommandLineTool
      inputs: []
      outputs:
        output:
          type: File
          outputBinding:
            glob: 'output.txt'
      stdout: output.txt
      arguments: ["echo", "step_a"] 
    in: []
    out: ['output']
  step_b:
    run:
      class: CommandLineTool
      inputs: []
      outputs:
        output:
          type: File
          outputBinding:
            glob: 'output.txt'
      stdout: output.txt
      arguments: ["echo", "step_b"] 
    in: []
    out: ['output']

$ cwltool test-outputs.cwl 
INFO /home/kinow/Development/python/workspace/user_guide/venv/bin/cwltool 3.1.20220830195442
INFO Resolved 'test-outputs.cwl' to 'file:///tmp/test-outputs.cwl'
INFO [workflow ] start
INFO [workflow ] starting step step_a
INFO [step step_a] start
INFO [job step_a] /tmp/y7vfdm3u$ echo \
    step_a > /tmp/y7vfdm3u/output.txt
INFO [job step_a] completed success
INFO [step step_a] completed success
INFO [workflow ] starting step step_b
INFO [step step_b] start
INFO [job step_b] /tmp/_xviar_m$ echo \
    step_b > /tmp/_xviar_m/output.txt
INFO [job step_b] completed success
INFO [step step_b] completed success
INFO [workflow ] completed success
{
    "step_a_output": {
        "location": "file:///tmp/output.txt",
        "basename": "output.txt",
        "class": "File",
        "checksum": "sha1$b17af6d5f1b21cb562c429f8ea5f66d1b7848dd4",
        "size": 7,
        "path": "/tmp/output.txt"
    },
    "step_b_output": {
        "location": "file:///tmp/output.txt_2",
        "basename": "output.txt",
        "class": "File",
        "checksum": "sha1$7b6d4d56856aac0d78f715f2c71e7ba2dda0645f",
        "size": 7,
        "path": "/tmp/output.txt_2" # <------------
    }
}
INFO Final process status is success

[mr-c]

@kinow In https://www.commonwl.org/user_guide/faq.html#rename-an-output-file ?

[kinow]

@kinow In https://www.commonwl.org/user_guide/faq.html#rename-an-output-file ?

Ah, that nails it! I think it covers exactly what @golharam has asked in the issue.

I created the example above after reading your previous comment (was also a Today-I-Learned moment to me):

CWL workflow engines should automatically rename files that have the same file name if they are written to the same final output directory. The CWL reference runner cwltool and related implementations do this for sure.

I think this issue can now be closed. WDYT @mr-c?

[swzCuroverse]

@mr-c and @kinow should this ticket be closed?

[tom-tan]

off topic:

CWL workflow engines should automatically rename files that have the same file name if they are written to the same final output directory.

I have two questions about it.

• Are there corresponding descriptions in the spec? I roughly checked the spec of v1.2 and changelog but I could not find it.
• I guess the current cwltest cannot check this feature because it does not check that the Files or Directory in the output object are distinct. Is my understanding correct?