Start

A list

A stage starts one job for each item of a list or of a stream.

One job for each item

each: gives a stage a list. The stage starts one job for each item, in parallel, and the job reads its item as ${item}. The script is the one from First machine.

# titles.yaml

flow: fetch-titles
input: { urls: list }

stages:
  get_title:
    each: ${input.urls}
    job:
      script: ./fetch-title.js
      input: { url: "${item}" }
      allow: [net:example.com, net:www.iana.org]

output: ${stages.get_title.output}

corbie test titles.yaml --input '{"urls": ["https://example.com", "https://www.iana.org/help/example-domains"]}'  output: [{"item":"https://example.com","output":{"title":"Example Domain","url":"https://example.com"}},{"item":"https://www.iana.org/help/example-domains","output":{"title":"Example Domains","url":"https://www.iana.org/help/example-domains"}}]

The stage's output is a list in the order of the items.

A stream of items

each: can also name another stage. That stage's script calls emit once for each item, and a job starts for each item as it arrives. --watch prints the run's record as it is written. Its lines are shown here without their time and run number.

# paged.yaml

flow: paged-titles
input: { pages: number }

stages:
  list_pages:
    job:
      script: ./list-pages.js
      input: { pages: "${input.pages}" }
  get_title:
    each: { emitted: list_pages }
    job:
      script: ./fetch-title.js
      input: { url: "${item}" }
      allow: [net:example.com]

output: ${stages.get_title.output}

// list-pages.js

for (let n = 1; n <= input.pages; n++) {
  emit("url", "https://example.com/?page=" + n)
}

corbie test paged.yaml --input '{"pages": 3}' --watch  step 1  -> job list_pages: a script, on runner  step 2  <- url from list_pages "https://example.com/?page=1"  step 2  -> job get_title.0: a script, on runner

The first get_title job starts at step 2, while list_pages is still emitting.

Next: Machines together runs a flow from a statechart and from an agent.