Removing .NET Interactive Breaks a Critical Exploratory Workflow in C#

[nczsl]

Title

Removing .NET Interactive Breaks a Critical Exploratory Workflow in C#

Content

I understand the cost and complexity behind maintaining .NET Interactive and Polyglot Notebooks. However, this deprecation removes a uniquely important capability from the .NET ecosystem — one that currently has no real replacement.

---

1. This is not about notebooks — it is about execution model

What .NET Interactive provides is not just a notebook UI, but a fundamentally different way of working with code:

• Incremental execution
• Stateful evaluation
• Dynamic, cumulative compilation
• Step-by-step exploration via #load

This enables a workflow that is not achievable with traditional compile-run cycles.

---

2. Dynamic cumulative compilation is essential

In my workflow, I rely heavily on:

• Breaking code into parts using #load
• Executing and evolving logic incrementally
• Avoiding full project recompilation on every change

This is not just a convenience — it directly impacts:

• Development speed
• Iteration quality
• Willingness to explore and validate ideas

Without this model, the workflow degrades into:

modify → compile → run → repeat

For exploratory development, this is simply too slow and disruptive.

---

3. Real-world usage: dual-mode development

I currently maintain a dual-mode system:

• Build mode → standard dotnet build
• Exploration mode → .ipynb + #load (powered by Roslyn)

I even use Roslyn to switch between these modes.

This allows:

• Fast experimentation in notebook mode
• Clean transition back to production code

This workflow significantly improves both productivity and code quality.

---

4. File-based apps are not a replacement

The suggested alternative (file-based apps / scripts) does not provide:

• Incremental execution
• Persistent state across steps
• Structured, step-by-step evaluation

Even with scripting, the lack of cumulative execution makes it fundamentally different.

---

5. A minimal path forward exists

Even if Polyglot Notebooks are too costly to maintain, there is still a high-value middle ground:

A minimal, C#-only interactive kernel

Without multi-language support, this would still preserve:

• Stateful execution
• Incremental compilation
• #load-based modular exploration

This would retain the core value while reducing complexity.

---

6. Long-term concern

The current guidance suggests continuing to use .NET Interactive “as-is,” with the risk of future breakage.

This raises serious concerns:

• Should users stop upgrading .NET to preserve their workflow?
• Is there any supported long-term alternative?
• Are exploratory workflows no longer a goal for the .NET ecosystem?

---

7. Why this matters

This is not a niche scenario.

Exploratory programming is critical for:

• Designing systems
• Validating ideas
• Understanding complex logic

Removing this capability without a replacement reduces the effectiveness of C# as a tool for thinking, not just building.

---

Final request

If full support cannot be maintained, I strongly suggest:

• Keeping a minimal, supported C# interactive execution environment
  OR
• Providing an officially supported alternative with equivalent capabilities

Without this, an important part of the developer experience is lost.

[nczsl]

I would like to add an additional perspective that may not be fully captured above: the role of notebook-style workflows in learning, analysis, and visualization.

Jupyter-style environments are widely used not only for data science, but for:

• Learning complex systems step by step
• Exploring algorithms interactively
• Visualizing intermediate results
• Understanding behavior through incremental evaluation

In many cases, the value is not just execution, but the ability to:

• Inspect state at each step
• Visualize data (charts, tables, structured output)
• Iterate quickly without losing context

This is especially important when:

• Investigating non-trivial logic
• Prototyping ideas
• Teaching or explaining concepts

While C# is not the dominant language in data science, the lack of an interactive, notebook-style environment makes it significantly harder to use C# in these scenarios.

Even a minimal interactive environment would still enable:

• Step-by-step reasoning
• Structured exploration
• Integration with external visualization tools

Without this, C# becomes much less effective for exploratory and educational use cases.

I believe this aspect is an important part of the overall impact of this deprecation.

[sheng-jie]

Yes, I completely agree. It was a bad decision to abandon dotnet interactive. Polyglot notebook can accept, at least keep dotnet interactive

[jonsequitur]

You might be interested in #4195.