Job artifacts serve two purposes:

• To publish generated files to the outside world.
• To pass files between two jobs within the same build.

And jobs can do two things with artifacts: save them or restore them.

Saving artifacts

Jobs that create artifacts in order for other jobs to use them, or to make them available for download, need to explicitly tell MonkeyCI where to find the artifact files in question, and give them an id. This id will be used to refer to them in other jobs. This is done using the save-artifacts function. For example:

(def create-file
  (-> (action-job
        "create-file-job"
	(fn [ctx]
	  (spit "test.txt" "This is an artifact file")))
      (save-artifacts (artifact "some-id" "test.txt"))))

In this example we use the Clojure standard spit function to write a string to a file. The artifact function is available to declare an artifact, with an id and a location. The location is relative to the job working directory, which is the checkout directory by default.

Using artifacts

Jobs that want to use an artifact that has been generated by another job, must indicate that to MonkeyCI. Both the id of the artifact, and where to restore it has to be specified. For this, the restore-artifacts function is available.

Building further on the previous example, we could write another job that uses the saved artifact like so:

(def use-file
  (-> (action-job
        "use-file-job"
	(fn [ctx]
	  (println "The artifact contents is:" (slurp "test.txt"))))
      (restore-artifacts (artifact "some-id" "test.txt"))
      (depends-on "create-file-job"))

We use the counterpart of spit, called slurp to read the file contents here. Of course, this is a contrived example, but it illustrates the point. Also note the explicit dependency declaration, more on that below.

Dependencies

Note that artifacts do not imply dependency. You still have to specify the job interdependencies explicitly. MonkeyCI will not warn you when a job wants to restore an artifact that is not being saved in the dependency chain (although we may add that as a feature later on). The artifact will simply not be available, and the job will probably fail.

So for MonkeyCI it's perfectly acceptable that an artifact is not available, it's up to the job itself to handle that situation.