Run history and branches

Science is iterative: you run a pipeline, change something, run it again, compare. SciStudio keeps two tools for that in the bottom panel — Run history, which remembers every run so you can go back to one, and Branches, which lets you keep several variants of a pipeline side by side.

Run history

Every time you run a workflow, SciStudio records it. The Run history tab (bottom panel) lists your past runs, newest first. Each entry shows its status (completed / failed / cancelled / running), when it ran, the workflow, how long it took, and how many blocks it had.

Looking inside a run

Click a run to open its detail. You see every block that executed, and expanding a block shows:

the exact parameters it ran with (resolved, including defaults),
its inputs and outputs (which data objects, of which types),
any error if it failed.

So a run is a complete, inspectable record of what happened — not just the result, but the settings and data that produced it.

Going back to a previous run

This is the point of keeping history: you can return to any past run.

Restore workflow. Restores the workflow to exactly how it was when that run executed — the graph and every block's parameters. Use it when a later change made things worse and you want to get back to a version that worked.
Re-run. Creates a fresh run from a past run's recorded workflow and parameters. Before it does, SciStudio checks whether anything has drifted — an input file that changed on disk, or a different package version in your environment — and warns you, so you know whether the re-run will reproduce the original result. The new run is linked to the one it came from.

Note what is and is not restored: the workflow and its parameters are restored from the record; your input data files are whatever is on disk now (SciStudio checks them and warns if they changed, but does not roll them back).

Export methods

From a run you can Export methods — a ready-to-read Markdown description of the workflow and the parameters it used, to copy or download. It is meant for writing up what you did (a methods section, a lab note) without retyping it.

Branches

A run records the past. A branch lets you keep more than one present.

The common situation: you have a pipeline, and you want to try it two ways — two batches of data, two sets of processing parameters, or the same blocks in a slightly different order — and switch between them easily without losing either. That is exactly what branches are for. (If you have used git, these are git branches; you do not need to know git to use them.)

How it works

The Branches control lives in the Git tab of the bottom panel, showing your current branch.

Create a branch. From the branch menu, Create branch…, give it a name (e.g. batch-2 or stronger-smoothing). It branches from where you are now and switches you onto it. Changes you make now — different parameters, a reordered pipeline — live on this branch.
Switch branches. Pick another branch from the menu to switch to it; the canvas reloads to that branch's version of the workflows. If you have unsaved changes when you switch, SciStudio saves them for you automatically (and tells you, so you can recover them) — you never lose work by switching.
Compare and combine. When a variant works out, you can merge a branch into your current one; the menu has Merge into current.

What a branch covers

A branch tracks the things that define your pipeline: the workflow files, your custom blocks, and your project notes. It does not touch your past runs (those are kept permanently in run history, independent of branches) or your input data files. So switching branches changes the pipeline, while your run history and your data stay put.

A worked example

You ran a pipeline on batch 1 and it looks good. Batch 2 needs a higher smoothing setting and an extra normalization block.

Create a branch batch-2 and switch to it.
On batch-2, raise the smoothing parameter and add the normalization block.
Run it. Now main has the batch-1 pipeline and batch-2 has the batch-2 pipeline; switch between them anytime to compare or re-run either.
If batch-2's changes turn out to be the better default, merge into current to bring them back to main.