Publishing Jobs

The publish feature makes it easy to share code and results. Published jobs are visible to all authenticated users.

How to Publish a job

Select a particular job that you would like to publish. On the page you should be seeing 3 buttons that say Cancel Job, Redo Job, Publish Job.

../_images/publish.png

Clicking the Publish Job takes you to the Publish page. Here you are required to fill out a meaningful name for the job and a short job description. At the minimum your job description must cover the following:

  1. What does the job do ?
  2. Is it CPU/RAM/IO intensive
  3. If the application is not multi-threaded/multi-processing note that.
  4. If your code accesses protected datasets, note that as a requirement.
  5. Describe each input file and expected format.
  6. Describe each output file and expected format.

Guidelines for Publishing

  1. Do NOT publish any results that are derived from a private dataset (e.g. Web of Science, JSTOR) without prior approval from the Administrators.
  2. Do NOT publish results or data that are going into a manuscript prior to publication.
  3. Add a complete and meaningful job description. When in doubt ask another human

Guidelines for Publishing

  1. Do NOT publish any results that are derived from a private dataset (e.g. Web of Science, JSTOR) without prior approval from the Administrators.
  2. Do NOT publish results or data that are going into a manuscript prior to publication.
  3. Add a complete and meaningful job description. When in doubt ask another human

Retracting a Published job

The owner/maintainer of a Published Job can retract a job from the Published jobs page by clicking the Retract button on the job info page. Please note that the Retract button merely takes the job off the Published jobs page. There are no group based controls on who can view your jobs, as a result if a job is published the unique job-id is known to the world and can be used to view the job even after it has been retracted. Therefore users are adviced to make sure that critical/sensitive results are not present in the jobs they publish.

Note

The job-id based access is purely security by obscurity. While the input datasets are protected by role based models, the outputs attached to jobs are treated as objects that require much less protection.