Docs / Workflows
Pipeline model
What a pipeline file contains, and why that matters more than it sounds.
What is in the file
| Cards | Which ones, at which versions |
|---|---|
| Parameters | Every value, as set, not as resolved |
| Connections | The graph |
| Layout | Positions on the canvas, so it looks the same when reopened |
It does not contain your data, or any path to it. A pipeline is portable because it describes the analysis and nothing about the machine it ran on.
Card versions are recorded
Opening a pipeline that uses a card you do not have gives SS-E040 naming the card, so you can install it. Opening one saved with a newer version of a card you do have gives SS-E041.
Why archive it
The pipeline file is what makes a published analysis checkable. Deposit it with your data and a reviewer can run exactly what you ran. See how to cite.
During the beta the file format can still change between
versions. Record the version you ran, from Help then About, alongside the
file.