JSON Editor

The Editor is the text authoring lens of the Model Workspace's Build tab: the whole model as an editable OptimizationProblem JSON document. It is the fastest lens when you already know exactly what you want — pasting a model from elsewhere, bulk-editing bounds, or tweaking an expression by hand.

The document

The editor shows the same canonical model the other lenses edit, pretty-printed:

{
  "name": "production_plan",
  "variables": [
    { "name": "x", "type": "integer", "lower_bound": 0, "upper_bound": 100 },
    { "name": "y", "type": "continuous", "lower_bound": 0 }
  ],
  "constraints": [
    { "name": "capacity", "expression": "2*x + 3*y <= 240" }
  ],
  "objective": { "sense": "maximize", "expression": "50*x + 40*y" }
}

The schema is the same one the Solve API accepts — anything you can POST /api/v2/solve, you can paste here.

Live validation

As you type, the editor validates on two levels:

1. Syntax + shape — malformed JSON or a document missing the minimum structure shows an inline parse error immediately.
2. Semantic validation — a well-formed document is checked against the real validator (POST /solve/validate): undefined variables in expressions, empty bounds, and other model-level problems surface as non-blocking errors and warnings.

A Format button re-pretty-prints the document.

Invalid JSON never leaks

While the editor holds malformed JSON, the rest of the workspace keeps the last good model — the canvas doesn't go blank, and Analyze keeps its numbers. Solving and committing are blocked until the text parses again, so you can never run or version a model you only think you changed.

Large models

Very large models (past the canvas scale limit) are still fully editable here — the editor and the JModel lens are the intended way to work with them.