Greenfield vs Brownfield: How AI Changed the Way I Build and Rescue Software

Greenfield vs Brownfield: How AI Changed the Way I Build and Rescue Software

I recently listened to an episode of the Merge Conflict podcast by James Montemagno and Frank Krueger where a topic came up that, surprisingly, I had never explicitly framed before: greenfield vs brownfield projects.

That surprised me—not because the ideas were new, but because I’ve spent years deep in software architecture and AI, and yet I had never put a name to something I deal with almost daily.

Once I did a bit of research (and yes, asked ChatGPT too), everything clicked.


Greenfield and Brownfield, in Simple Terms

  • Greenfield projects are built from scratch. No legacy code, no historical baggage, no technical debt.
  • Brownfield projects already exist. They carry history: multiple teams, different styles, shortcuts, and decisions made under pressure.

If that sounds abstract, here’s the practical version:

Greenfield is what we want.

Brownfield is what we usually get.


Greenfield Projects: Architecture Paradise

In a greenfield project, everything feels right.

You can choose your architecture and actually stick to it. If you’re building a .NET MAUI application, you can start with proper MVVM, SOLID principles, clean boundaries, and consistent conventions from day one.

As developers, we know how things should be done. Greenfield projects give us permission to do exactly that.

They’re also extremely friendly to AI tools.

When the rules are clear and consistent, Copilot and AI agents perform beautifully. You can define specs, outline patterns, and let the tooling do a lot of the repetitive work for you.

That’s why I often use AI for greenfield projects as internal tools or side projects—things I’ve always known how to build, but never had the time to prioritize. Today, time is no longer the constraint. Tokens are.


Brownfield Projects: Welcome to Reality

Then there’s the real world.

At the office, we work with applications that have been touched by many hands over many years—sometimes 10 different teams, sometimes freelancers, sometimes “someone’s cousin who fixed it once.”

Each left behind a different style, different patterns, and different assumptions.

Customers often describe their systems like this:

“One team built it, another modified it, then my cousin fixed a bug, then my cousin got married and stopped helping, and then someone else took over.”

And yet—the system works.

That’s an important reminder.

The main job of software is not to be beautiful. It’s to do the job.

A lot of brownfield systems are ugly, fragile, and terrifying to touch—but they deliver real business value every single day.


Why AI Is Even More Powerful in Brownfield Projects

Here’s my honest opinion, based on experience:

AI is even more valuable in brownfield projects than in greenfield ones.

I’ve modernized six or seven legacy applications so far—codebases that everyone was afraid to touch. AI made that possible.

Legacy systems are mentally expensive. Reading spaghetti code drains energy. Understanding implicit behavior takes time. Humans get tired.

AI doesn’t.

It will patiently analyze a 2,000-line class without complaining.

Take Windows Forms applications as an example. It’s old technology, easy to forget, and full of quirks. Copilot can generate code that I know how to write—but much faster than I could after years away from WinForms.

Even more importantly, AI makes it far easier to introduce tests into systems that never had them:

  • Add tests class by class
  • Mock dependencies safely
  • Lock in existing behavior before refactoring

Historically, this was painful enough that many teams preferred a full rewrite.

But rewrites have a hidden cost: every rewritten line introduces new bugs.

AI allows us to modernize in place—incrementally and safely.


Clean Code and Business Value

This is the real win.

With AI, we no longer have to choose between:

  • “The code works, but don’t touch it”
  • “The code is beautiful, but nothing works yet”

We can improve structure, readability, and testability without breaking what already delivers value.

Greenfield projects are still fun. They’re great for experimentation and clean design.

But brownfield projects? That’s where AI feels like a superpower.


Final Thoughts

Today, I happily use AI in both worlds:

  • Greenfield projects for fast experimentation and internal tooling
  • Brownfield projects for rescuing legacy systems, adding tests, and reducing technical debt

AI doesn’t replace experience—it amplifies it.

Especially when dealing with systems held together by history, habits, and just enough hope to keep running.

And honestly?

Those are the projects where the impact feels the most real.

The DLL Registration Trap in Legacy .NET Framework Applications

The DLL Registration Trap in Legacy .NET Framework Applications

If you’ve ever worked on a traditional .NET Framework application — the kind that predates .NET Core and .NET 5+ — this story may feel painfully familiar.

I’m talking about classic .NET Framework 4.x applications (4.0, 4.5, 4.5.1, 4.5.2, 4.6, 4.6.1, 4.6.2, 4.7, 4.7.1, 4.7.2, 4.8, and the final release 4.8.1). These systems often live long, productive lives… and accumulate interesting technical debt along the way.

This particular system is written in C# and relies heavily on COM components to render video, audio, and PDF content. Under the hood, many of these components are based on technologies like DirectShow filters, ActiveX controls, or other native COM DLLs.

And that’s where the story begins.


The Setup: COM, DirectShow, and Registration

Unlike managed .NET assemblies, COM components don’t just live quietly next to your executable. They need to be registered in the system registry so Windows knows:

  • What CLSID they expose
  • Which DLL implements that CLSID
  • Whether it’s 32-bit or 64-bit
  • How it should be activated

For DirectShow-based components (very common for video/audio playback in legacy apps), registration is usually done manually during development using regsvr32.

Example:

regsvr32 MyVideoFilter.dll

To unregister:

regsvr32 /u MyVideoFilter.dll

Important detail that bites a lot of people:

  • 32-bit DLLs must be registered using:
C:\Windows\SysWOW64\regsvr32.exe My32BitFilter.dll
  • 64-bit DLLs must be registered using:
C:\Windows\System32\regsvr32.exe My64BitFilter.dll

Yes — the folder names are historically confusing.


Development Works… Until It Doesn’t

So here’s the usual development flow:

  1. You register all required COM DLLs on your development machine
  2. Visual Studio runs the app
  3. Video plays, audio works, PDFs render
  4. Everyone is happy

Then comes the next step.

“Let’s build an installer.”


The Installer Paradox

This is where the real battle story begins.

Your application installer (MSI, InstallShield, WiX, Inno Setup — pick your poison) now needs to:

  • Copy the COM DLLs
  • Register them during installation
  • Unregister them during uninstall

This seems reasonable… until you test it.

The Loop From Hell

Here’s what happens in practice:

  • You install your app for testing
  • The installer registers its own copies of the COM DLLs
  • Your development environment was using different copies (maybe newer, maybe local builds)
  • Suddenly:
    • Your source build stops working
    • Visual Studio debugging breaks
    • Another app on your machine mysteriously fails

Then you:

  • Uninstall the app
  • The installer unregisters the DLLs
  • Now nothing works anymore

So you re-register the DLLs manually for development…

…and the cycle repeats.


The Battle Story: It Only Worked… Until It Didn’t

For a long time, this system appeared to work just fine.

Video played. Audio rendered. PDFs opened. No obvious errors.

What we didn’t realize at first was a dangerous hidden assumption:

The system only worked on machines where a previous version had already been installed.

Those older installations had left COM DLLs registered in the system — quietly, globally, and invisibly.

So when we deployed a new version without removing the old one:

  • Everything looked fine
  • No one suspected missing registrations
  • The system passed casual testing

The illusion broke the moment we tried a clean installation.

On a fresh machine — no previous version, no leftover registry entries — the application suddenly failed:

  • Components didn’t initialize
  • Media rendering silently broke
  • COM activation errors appeared only in Event Viewer

The installer claimed it was registering the DLLs.

In reality, it wasn’t doing it correctly — or at least not in the way the application actually needed.

That’s when we realized we were standing on years of accidental state.


Why This Happens

The core problem is simple but brutal:

COM registration is global and mutable.

There is:

  • One registry
  • One CLSID mapping
  • One “active” DLL per COM component

Your development environment, your installed application, and your installer are all fighting over the same global state.

.NET Framework itself isn’t the villain here — it’s just sitting on top of an old Windows integration model that predates modern isolation concepts.


A New Player Enters: ARM64

Just when we thought the problem space was limited to x86 vs x64, another variable entered the scene.

One of the development machines was ARM64.

Modern Windows on ARM adds a new layer of complexity:

  • ARM64 native processes
  • x64 emulation
  • x86 emulation on top of ARM64

From the outside, everything looks like it’s running on x64.

Under the hood, it’s not that simple.

Why This Makes COM Registration Worse

COM registration is architecture-specific:

  • x86 DLLs register under one view of the registry
  • x64 DLLs register under another
  • ARM64 introduces yet another execution context

On Windows ARM:

  • System32 contains ARM64 binaries
  • SysWOW64 contains x86 binaries
  • x64 binaries often run through emulation layers

So now the questions multiply:

  • Which regsvr32 did the installer call?
  • Was it ARM64, x64, or x86?
  • Did the app run natively, or under emulation?
  • Did the COM DLL match the process architecture?

The result is a system where:

  • Some things work on Intel machines
  • Some things work on ARM machines
  • Some things only work if another version was installed first

At this point, debugging stops being logical and starts being archaeological.


Why This Is So Common in .NET Framework 4.x Apps

Many enterprise and media-heavy applications built on:

  • .NET Framework 4.0–4.8.1
  • WinForms or WPF
  • DirectShow or ActiveX components

were designed in an era where:

  • Global COM registration was normal
  • Side-by-side isolation was rare
  • “Just register the DLL” was accepted practice

These systems work, but they’re fragile — especially on developer machines.


Where the Article Is Going Next

In the rest of this article series, we’ll look at:

  • Why install-time registration is often a mistake
  • How to isolate development vs runtime environments
  • Techniques like:
    • Dedicated dev VMs
    • Registration-free COM (where possible)
    • App-local COM deployment
    • Clear ownership rules for installers
  • How to survive (and maintain) legacy .NET Framework systems without losing your sanity

If you’ve ever broken your own development environment just by testing your installer — you’re not alone.

This is the cost of living at the intersection of managed code and unmanaged history.

ConfigureAwait(false): Why It Exists, What It Solves, and When Context Is the Real Bug

ConfigureAwait(false): Why It Exists, What It Solves, and When Context Is the Real Bug

Async/await in C# is often described as “non-blocking,” but that description hides an important detail:

await is not just about waiting — it is about where execution continues afterward.

Understanding that single idea explains:

  • why deadlocks happen,
  • why ConfigureAwait(false) exists,
  • and why it *reduces* damage without fixing the root cause.

This article is not just theory. It’s written because this exact class of problem showed up again in real production code during the first week of 2026 — and it took a context-level fix to resolve it.

The Hidden Mechanism: Context Capture

When you await a task, C# does two things:

  1. It pauses the current method until the awaited task completes.
  2. It captures the current execution context (if one exists) so the continuation can resume there.

That context might be:

  • a UI thread (WPF, WinForms, MAUI),
  • a request context (classic ASP.NET),
  • or no special context at all (ASP.NET Core, console apps).

This default behavior is intentional. It allows code like this to work safely:

var data = await LoadAsync();
MyLabel.Text = data.Name; // UI-safe continuation

But that same mechanism becomes dangerous when async code is blocked synchronously.

The Root Problem: Blocking on Async

Deadlocks typically appear when async code is forced into a synchronous shape:

var result = GetDataAsync().Result; // or .Wait()

What happens next:

  1. The calling thread blocks, waiting for the async method to finish.
  2. The async method completes its awaited operation.
  3. The continuation tries to resume on the original context.
  4. That context is blocked.
  5. Nothing can proceed.

💥 Deadlock.

This is not an async bug. This is a context dependency cycle.

The Blast Radius Concept

Blocking on async is the explosion.

The blast radius is how much of the system is taken down with it.

Full blast (default await)

  • Continuation *requires* the blocked context
  • The async operation cannot complete
  • The caller never unblocks
  • Everything stops

Reduced blast (ConfigureAwait(false))

  • Continuation does not require the original context
  • It resumes on a thread pool thread
  • The async operation completes
  • The blocking call unblocks

The original mistake still exists — but the damage is contained.

The real fix is “don’t block on async,”
but ConfigureAwait(false) reduces the blast radius when someone does.

What ConfigureAwait(false) Actually Does

await SomeAsyncOperation().ConfigureAwait(false);

This tells the runtime:

“I don’t need to resume on the captured context. Continue wherever it’s safe to do so.”

Important clarifications:

  • It does not make code faster by default
  • It does not make code parallel
  • It does not remove the need for proper async flow
  • It only removes context dependency

Why This Matters in Real Code

Async code rarely exists in isolation.

A method often awaits another method, which awaits another:

await AAsync();
    await BAsync();
        await CAsync();

If any method in that chain requires a specific context, the entire chain becomes context-bound.

That is why:

  • library code must be careful,
  • deep infrastructure layers must avoid context assumptions,
  • and UI layers must be explicit about where context is required.

When ConfigureAwait(false) Is the Right Tool

Use it when all of the following are true:

  • The method does not interact with UI state
  • The method does not depend on a request context
  • The method is infrastructure, library, or backend logic
  • The continuation does not care which thread resumes it

This is especially true for:

  • NuGet packages
  • shared libraries
  • data access layers
  • network and IO pipelines

What It Is Not

ConfigureAwait(false) is not:

  • a fix for bad async usage
  • a substitute for proper async flow
  • a reason to block on tasks
  • something to blindly apply everywhere

It is a damage-control tool, not a cure.

A Real Incident: When None of the Usual Fixes Worked

First week of 2026.

The first task I had with the programmers in my office was to investigate a problem in a trading block. The symptoms looked like a classic async issue: timing bugs, inconsistent behavior, and freezes that felt “await-shaped.”

We did what experienced .NET teams typically do when async gets weird:

  • Reviewed the full async/await chain end-to-end
  • Double-checked the source code carefully (everything looked fine)
  • Tried the usual “tools people reach for” under pressure:
  • .Wait()
  • .GetAwaiter().GetResult()
  • wrapping in Task.Run(...)
  • adding ConfigureAwait(false)
  • mixing combinations of those approaches

None of it reliably fixed the problem.

At that point it stopped being a “missing await” story. It became a “the model is right but reality disagrees” story.

One of the programmers, Daniel, and I went deeper. I found myself mentally replaying every async pattern I know — especially because I’ve written async-heavy code myself, including library work like SyncFramework, where I synchronize databases and deal with long-running operations.

That’s the moment where this mental model matters: it forces you to stop treating await like syntax and start treating it like mechanics.

The Actual Root Cause: It Was the Context

In the end, the culprit wasn’t which pattern we used — it was where the continuation was allowed to run.

This application was built on DevExpress XAF. In this environment, the “correct” continuation behavior is often tied to XAF’s own scheduling and application lifecycle rules. XAF provides a mechanism to run code in its synchronization context — for example using BlazorApplication.InvokeAsync, which ensures that continuations run where the framework expects.

Once we executed the problematic pipeline through XAF’s synchronization context, the issue was solved.

No clever pattern. No magical await. No extra parallelism.

Just: the right context.

And this is not unique to XAF. Similar ideas exist in:

  • Windows Forms (UI thread affinity + SynchronizationContext)
  • WPF (Dispatcher context)
  • Any framework that requires work to resume on a specific thread/context

Why I’m Writing This

What I wanted from this experience is simple: don’t forget it.

Because what makes this kind of incident dangerous is that it looks like a normal async bug — and the internet is full of “four fixes” people cycle through:

  1. add/restore missing await
  2. use .Wait() / .Result
  3. wrap in Task.Run()
  4. use ConfigureAwait(false)

Sometimes those are relevant. Sometimes they’re harmful. And sometimes… they’re all beside the point.

In our case, the missing piece was framework context — and once you see that, you realize why the “blast radius” framing is so useful:

  • Blocking is the explosion.
  • ConfigureAwait(false) contains damage when someone blocks.
  • If a framework requires a specific synchronization context, the fix may be to supply the correct context explicitly.

That’s what happened here. And that’s why I’m capturing it as live knowledge, not just documentation.

The Mental Model to Keep

  • Async bugs are often context bugs
  • Blocking creates the explosion
  • Context capture determines the blast radius
  • ConfigureAwait(false) limits the damage
  • Proper async flow prevents the explosion entirely
  • Frameworks may require their own synchronization context
  • Correct async code can still fail in the wrong context

Async is not just about tasks. It’s about where your code is allowed to continue.

 

Structured RAG for Unknown and Mixed Languages

Structured RAG for Unknown and Mixed Languages

How I stopped my multilingual activity stream from turning RAG into chaos

In the previous article (RAG with PostgreSQL and C# (pros and cons) | Joche Ojeda) I explained how naïve RAG breaks when you run it over an activity stream.

Same UI language.
Totally unpredictable content language.
Spanish, Russian, Italian… sometimes all in the same message.

Humans handle that fine.
Vector retrieval… not so much.

This is the “silent failure” scenario: retrieval looks plausible, the LLM sounds confident, and you ship nonsense.

So I had to change the game.

The Idea: Structured RAG

Structured RAG means you don’t embed raw text and pray.

You add a step before retrieval:

  • Extract a structured representation from each activity record
  • Store it as metadata (JSON)
  • Use that metadata to filter, route, and rank
  • Then do vector similarity on a cleaner, more stable representation

Think of it like this:

Unstructured text is what users write.
Structured metadata is what your RAG system can trust.

Why This Fix Works for Mixed Languages

The core problem with activity streams is not “language”.

The core problem is: you have no stable shape.

When the shape is missing, everything becomes fuzzy:

  • Who is speaking?
  • What is this about?
  • Which entities are involved?
  • Is this a reply, a reaction, a mention, a task update?
  • What language(s) are in here?

Structured RAG forces you to answer those questions once, at write-time, and save the answers.

PostgreSQL: Add a JSONB Column (and Keep pgvector)

We keep the previous approach (pgvector) but we add a JSONB column for structured metadata.

ALTER TABLE activities
ADD COLUMN rag_meta jsonb NOT NULL DEFAULT '{}'::jsonb;

-- Optional: if you store embeddings per activity/chunk
-- you keep your existing embedding column(s) or chunk table.

Then index it.

CREATE INDEX activities_rag_meta_gin
ON activities
USING gin (rag_meta);

Now you can filter with JSON queries before you ever touch vector similarity.

A Proposed Schema (JSON Shape You Control)

The exact schema depends on your product, but for activity streams I want at least:

  • language: detected languages + confidence
  • actors: who did it
  • subjects: what object is involved (ticket, order, user, document)
  • topics: normalized tags
  • relationships: reply-to, mentions, references
  • summary: short canonical summary (ideally in one pivot language)
  • signals: sentiment/intent/type if you need it

Example JSON for one activity record:

{
  "schemaVersion": 1,
  "languages": [
    { "code": "es", "confidence": 0.92 },
    { "code": "ru", "confidence": 0.41 }
  ],
  "actor": {
    "id": "user:42",
    "displayName": "Joche"
  },
  "subjects": [
    { "type": "ticket", "id": "ticket:9831" }
  ],
  "topics": ["billing", "invoice", "error"],
  "relationships": {
    "replyTo": "activity:9912001",
    "mentions": ["user:7", "user:13"]
  },
  "intent": "support_request",
  "summary": {
    "pivotLanguage": "en",
    "text": "User reports an invoice calculation error and asks for help."
  }
}

Notice what happened here: the raw multilingual chaos got converted into a stable structure.

Write-Time Pipeline (The Part That Feels Expensive, But Saves You)

Structured RAG shifts work to ingestion time.

Yes, it costs tokens.
Yes, it adds steps.

But it gives you something you never had before: predictable retrieval.

Here’s the pipeline I recommend:

  1. Store raw activity (as-is, don’t lose the original)
  2. Detect language(s) (fast heuristic + LLM confirmation if needed)
  3. Extract structured metadata into your JSON schema
  4. Generate a canonical “summary” in a pivot language (often English)
  5. Embed the summary + key fields (not the raw messy text)
  6. Save JSON + embedding

The key decision: embed the stable representation, not the raw stream text.

C# Conceptual Implementation

I’m going to keep the code focused on the architecture. Provider details are swappable.

Entities

public sealed class Activity
{
    public long Id { get; set; }
    public string RawText { get; set; } = "";
    public string UiLanguage { get; set; } = "en";

    // JSONB column in Postgres
    public string RagMetaJson { get; set; } = "{}";

    // Vector (pgvector) - store via your pgvector mapping or raw SQL
    public float[] RagEmbedding { get; set; } = Array.Empty<float>();

    public DateTimeOffset CreatedAt { get; set; }
}

Metadata Contract (Strongly Typed in Code, Stored as JSONB)

public sealed class RagMeta
{
    public int SchemaVersion { get; set; } = 1;
    public List<DetectedLanguage> Languages { get; set; } = new();
    public ActorMeta Actor { get; set; } = new();
    public List<SubjectMeta> Subjects { get; set; } = new();
    public List<string> Topics { get; set; } = new();
    public RelationshipMeta Relationships { get; set; } = new();
    public string Intent { get; set; } = "unknown";
    public SummaryMeta Summary { get; set; } = new();
}

public sealed class DetectedLanguage
{
    public string Code { get; set; } = "und";
    public double Confidence { get; set; }
}

public sealed class ActorMeta
{
    public string Id { get; set; } = "";
    public string DisplayName { get; set; } = "";
}

public sealed class SubjectMeta
{
    public string Type { get; set; } = "";
    public string Id { get; set; } = "";
}

public sealed class RelationshipMeta
{
    public string? ReplyTo { get; set; }
    public List<string> Mentions { get; set; } = new();
}

public sealed class SummaryMeta
{
    public string PivotLanguage { get; set; } = "en";
    public string Text { get; set; } = "";
}

Extractor + Embeddings

You need two services:

  • Metadata extraction (LLM fills the schema)
  • Embeddings (Microsoft.Extensions.AI) for the stable text
public interface IRagMetaExtractor
{
    Task<RagMeta> ExtractAsync(Activity activity, CancellationToken ct);
}

Then the ingestion pipeline:

using System.Text.Json;
using Microsoft.Extensions.AI;

public sealed class StructuredRagIngestor
{
    private readonly IRagMetaExtractor _extractor;
    private readonly IEmbeddingGenerator<string, Embedding<float>> _embeddings;

    public StructuredRagIngestor(
        IRagMetaExtractor extractor,
        IEmbeddingGenerator<string, Embedding<float>> embeddings)
    {
        _extractor = extractor;
        _embeddings = embeddings;
    }

    public async Task ProcessAsync(Activity activity, CancellationToken ct)
    {
        // 1) Extract structured JSON
        RagMeta meta = await _extractor.ExtractAsync(activity, ct);

        // 2) Create stable text for embeddings (summary + keywords)
        string stableText =
            $"{meta.Summary.Text}\n" +
            $"Topics: {string.Join(", ", meta.Topics)}\n" +
            $"Intent: {meta.Intent}";

        // 3) Embed stable text
        var emb = await _embeddings.GenerateAsync(new[] { stableText }, ct);
        float[] vector = emb.First().Vector.ToArray();

        // 4) Save into activity record
        activity.RagMetaJson = JsonSerializer.Serialize(meta);
        activity.RagEmbedding = vector;

        // db.SaveChangesAsync(ct) happens outside (unit of work)
    }
}

This is the core move: you stop embedding chaos and start embedding structure.

Query Pipeline: JSON First, Vectors Second

When querying, you don’t jump into similarity search immediately.

You do:

  1. Parse the user question
  2. Decide filters (actor, subject type, topic)
  3. Filter with JSONB (fast narrowing)
  4. Then do vector similarity on the remaining set

Example: filter by topic + intent using JSONB:

SELECT id, raw_text
FROM activities
WHERE rag_meta @> '{"intent":"support_request"}'::jsonb
  AND rag_meta->'topics' ? 'invoice'
ORDER BY rag_embedding <=> @query_embedding
LIMIT 20;

That “JSON first” step is what keeps multilingual streams from poisoning your retrieval.

Tradeoffs (Because Nothing Is Free)

Structured RAG costs more at write-time:

  • more tokens
  • more latency
  • more moving parts

But it saves you at query-time:

  • less noise
  • better precision
  • more predictable answers
  • debuggable failures (because you can inspect metadata)

In real systems, I’ll take predictable and debuggable over “cheap but random” every day.

Final Thought

RAG over activity streams is hard because activity streams are messy by design.

If you want RAG to behave, you need structure.

Structured RAG is how you make retrieval boring again.
And boring retrieval is exactly what you want.

In the next article, I’ll go deeper into the exact pipeline details: language routing, mixed-language detection, pivot summaries, chunk policies, and how I made this production-friendly without turning it into a token-burning machine.

Let the year begin 🚀

“`

RAG with PostgreSQL and C# (pros and cons)

RAG with PostgreSQL and C# (pros and cons)

RAG with PostgreSQL and C#

Happy New Year 2026 — let the year begin

Happy New Year 2026 🎉

Let’s start the year with something honest.

This article exists because something broke.

I wasn’t trying to build a demo.
I was building an activity stream — the kind of thing every social or collaborative system eventually needs.

Posts.
Comments.
Reactions.
Short messages.
Long messages.
Noise.

At some point, the obvious question appeared:

“Can I do RAG over this?”

That question turned into this article.

The Original Problem: RAG over an Activity Stream

An activity stream looks simple until you actually use it as input.

In my case:

  • The UI language was English
  • The content language was… everything else

Users were writing:

  • Spanish
  • Russian
  • Italian
  • English
  • Sometimes all of them in the same message

Perfectly normal for humans.
Absolutely brutal for naïve RAG.

I tried the obvious approach:

  • embed everything
  • store vectors
  • retrieve similar content
  • augment the prompt

And very quickly, RAG went crazy.

Why It Failed (And Why This Matters)

The failure wasn’t dramatic.
No exceptions.
No errors.

Just… wrong answers.

Confident answers.
Fluent answers.
Wrong answers.

The problem was subtle:

  • Same concept, different languages
  • Mixed-language sentences
  • Short, informal activity messages
  • No guarantee of language consistency

In an activity stream:

  • You don’t control the language
  • You don’t control the structure
  • You don’t even control what a “document” is

And RAG assumes you do.

That’s when I stopped and realized:

RAG is not “plug-and-play” once your data becomes messy.

So… What Is RAG Really?

RAG stands for Retrieval-Augmented Generation.

The idea is simple:

Retrieve relevant data first, then let the model reason over it.

Instead of asking the model to remember everything, you let it look things up.

Search first.
Generate second.

Sounds obvious.
Still easy to get wrong.

The Real RAG Pipeline (No Marketing)

A real RAG system looks like this:

  1. Your data lives in a database
  2. Text is split into chunks
  3. Each chunk becomes an embedding
  4. Embeddings are stored as vectors
  5. A user asks a question
  6. The question is embedded
  7. The closest vectors are retrieved
  8. Retrieved content is injected into the prompt
  9. The model answers

Every step can fail silently.

Tokenization & Chunking (The First Trap)

Models don’t read text.
They read tokens.

This matters because:

  • prompts have hard limits
  • activity streams are noisy
  • short messages lose context fast

You usually don’t tokenize manually, but you do choose:

  • chunk size
  • overlap
  • grouping strategy

In activity streams, chunking is already a compromise — and multilingual content makes it worse.

Embeddings in .NET (Microsoft.Extensions.AI)

In .NET, embeddings are generated using Microsoft.Extensions.AI.

The important abstraction is:

IEmbeddingGenerator<TInput, TEmbedding>

This keeps your architecture:

  • provider-agnostic
  • DI-friendly
  • survivable over time

Minimal Setup

dotnet add package Microsoft.Extensions.AI
dotnet add package Microsoft.Extensions.AI.OpenAI

Creating an Embedding Generator

using OpenAI;
using Microsoft.Extensions.AI;
using Microsoft.Extensions.AI.OpenAI;

var client = new OpenAIClient("YOUR_API_KEY");

IEmbeddingGenerator<string, Embedding<float>> embeddings =
    client.AsEmbeddingGenerator("text-embedding-3-small");

Generating a Vector

var result = await embeddings.GenerateAsync(
    new[] { "Some activity text" });

float[] vector = result.First().Vector.ToArray();

That vector is what drives everything that follows.

⚠️ Embeddings Are Model-Locked (And Language Makes It Worse)

Embeddings are model-locked.

Meaning:

Vectors from different embedding models cannot be compared.

Even if:

  • the dimension matches
  • the text is identical
  • the provider is the same

Each model defines its own universe.

But here’s the kicker I learned the hard way:

Multilingual content amplifies this problem.

Even with multilingual-capable models:

  • language mixing shifts vector space
  • short messages lose semantic anchors
  • similarity becomes noisy

In an activity stream:

  • English UI
  • Spanish content
  • Russian replies
  • Emoji everywhere

Vector distance starts to mean “kind of related, maybe”.

That’s not good enough.

PostgreSQL + pgvector (Still the Right Choice)

Despite all that, PostgreSQL with pgvector is still the right foundation.

Enable pgvector

CREATE EXTENSION IF NOT EXISTS vector;

Chunk-Based Table

CREATE TABLE doc_chunks (
    id            bigserial PRIMARY KEY,
    document_id   bigint NOT NULL,
    chunk_index   int NOT NULL,
    content       text NOT NULL,
    embedding     vector(1536) NOT NULL,
    created_at    timestamptz NOT NULL DEFAULT now()
);

Technically correct.
Architecturally incomplete — as I later discovered.

Retrieval: Where Things Quietly Go Wrong

SELECT content
FROM doc_chunks
ORDER BY embedding <=> @query_embedding
LIMIT 5;

This query decides:

  • what the model sees
  • what it ignores
  • how wrong the answer will be

When language is mixed, retrieval looks correct — but isn’t.

Classic example: Moscow

  • Spanish: Moscú

  • Italian: Mosca

  • Meaning in Spanish: 🪰 a fly

So for a Spanish speaker, “Mosca” looks like it should mean insect (which it does), but it’s also the Italian name for Moscow.

Why RAG Failed in This Scenario

Let’s be honest:

  • Similar ≠ relevant
  • Multilingual ≠ multilingual-safe
  • Short activity messages ≠ documents
  • Noise ≠ knowledge

RAG didn’t fail because the model was bad.
It failed because the data had no structure.

Why This Article Exists

This article exists because:

  • I tried RAG on a real system
  • With real users
  • Writing in real languages
  • In real combinations

And the naïve RAG approach didn’t survive.

What Comes Next

The next article will not be about:

  • embeddings
  • models
  • APIs

It will be about structured RAG.

How I fixed this by:

  • introducing structure into the activity stream
  • separating concerns in the pipeline
  • controlling language before retrieval
  • reducing semantic noise
  • making RAG predictable again

In other words:
How to make RAG work after it breaks.

Final Thought

RAG is not magic.

It’s:

search + structure + discipline

If your data is chaotic, RAG will faithfully reflect that chaos — just with confidence.

Happy New Year 2026 🎆

If you’re reading this:
Happy New Year 2026.

Let’s make this the year we stop trusting demos
and start trusting systems that survived reality.

Let the year begin 🚀

Accessing Legacy Data (Fox pro) with XPO Using a Custom ODBC Provider

Accessing Legacy Data (Fox pro) with XPO Using a Custom ODBC Provider

One of the recurring challenges in real-world systems is not building new software — it’s
integrating with software that already exists.

Legacy systems don’t disappear just because newer technologies are available. They survive because they work,
because they hold critical business data, and because replacing them is often risky, expensive, or simply not allowed.

This article explores a practical approach to accessing legacy data using XPO by leveraging ODBC,
not as a universal abstraction, but as a bridge when no modern provider exists.

The Reality of Legacy Systems

Many organizations still rely on systems built on technologies such as:

  • FoxPro tables
  • AS400 platforms
  • DB2-based systems
  • Proprietary or vendor-abandoned databases

In these scenarios, it’s common to find that:

  • There is no modern .NET provider
  • There is no ORM support
  • There is an ODBC driver

That last point is crucial. ODBC often remains available long after official SDKs and providers have disappeared.
It becomes the last viable access path to critical data.

Why ORMs Struggle with Legacy Data

Modern ORMs assume a relatively friendly environment: a supported database engine, a known SQL dialect,
a compatible type system, and an actively maintained provider.

Legacy databases rarely meet those assumptions. As a result, teams are often forced to:

  • Drop down to raw SQL
  • Build ad-hoc data access layers
  • Treat legacy data as a second-class citizen

This becomes especially painful in systems that already rely heavily on DevExpress XPO for persistence,
transactions, and domain modeling.

ODBC Is Not Magic — and That’s the Point

ODBC is often misunderstood.

Using ODBC does not mean:

  • One provider works for every database
  • SQL becomes standardized
  • Type systems become compatible

Each ODBC-accessible database still has:

  • Its own SQL dialect
  • Its own limitations
  • Its own data types
  • Its own behavioral quirks

ODBC simply gives you a way in. It is a transport mechanism, not a universal language.

What an XPO ODBC Provider Really Is

When you implement an XPO provider on top of ODBC, you are not building a generic solution for all databases.

You are building a targeted adapter for a specific legacy system that happens to be reachable via ODBC.

This matters because ODBC is used here as a pragmatic trick:

  • To connect to something you otherwise couldn’t
  • To reuse an existing, stable access path
  • To avoid rewriting or destabilizing legacy systems

The database still dictates the SQL dialect, supported features, and type system. Your provider must respect those constraints.

Why XPO Makes This Possible

XPO is not just an ORM — it is a provider-based persistence framework.

All SQL-capable XPO providers are built on top of a shared foundation, most notably:

ConnectionProviderSql
https://docs.devexpress.com/CoreLibraries/DevExpress.Xpo.DB.ConnectionProviderSql

This architecture allows you to reuse XPO’s core benefits:

  • Object model
  • Sessions and units of work
  • Transaction handling
  • Integration with domain logic

While customizing what legacy systems require:

  • SQL generation
  • Command execution
  • Schema discovery
  • Type mapping

Dialects and Type Systems Still Matter

Even when accessed through ODBC:

  • FoxPro is not SQL Server
  • DB2 is not PostgreSQL
  • AS400 is not Oracle

Each system has its own:

  • Date and time semantics
  • Numeric precision rules
  • String handling behavior
  • Constraints and limits

An XPO ODBC provider must explicitly map database types, handle dialect-specific SQL,
and avoid assumptions about “standard SQL.” ODBC opens the door — it does not normalize what’s inside.

Real-World Experience: AS400 and DB2 in Production

This approach is not theoretical. Last year, we implemented a custom XPO provider using ODBC for
AS400 and DB2 systems in Mexico, where:

  • No viable modern .NET provider existed
  • The systems were deeply embedded in business operations
  • ODBC was the only stable integration path

By introducing an XPO provider on top of ODBC, we were able to integrate legacy data into a modern .NET architecture,
preserve domain models and transactional behavior, and avoid rewriting or destabilizing existing systems.

The Hidden Advantage: Modern UI and AI Access

Once legacy data is exposed through XPO, something powerful happens: that data becomes immediately available to modern platforms.

  • Blazor applications
  • .NET MAUI mobile and desktop apps
  • Background services
  • Integration APIs
  • AI agents and assistants

And you get this without rewriting the database, migrating the data, or changing the legacy system.
XPO becomes the adapter that allows decades-old data to participate in modern UI stacks, automated workflows,
and AI-driven experiences.

Why Not Just Use Raw ODBC?

Raw ODBC gives you rows, columns, and primitive values. XPO gives you domain objects, identity tracking,
relationships, transactions, and a consistent persistence model.

The goal is not to modernize the database. The goal is to modernize access to legacy data
so it can safely participate in modern architectures.

Closing Thought

An XPO ODBC provider is not a silver bullet. It will not magically unify SQL dialects, type systems, or database behavior.

But when used intentionally, it becomes a powerful bridge between systems that cannot be changed
and architectures that still need to evolve.

ODBC is the trick that lets you connect.
XPO is what makes that connection usable — everywhere, from Blazor UIs to AI agents.

ODBC: A Standard That Was Never Truly Neutral

ODBC: A Standard That Was Never Truly Neutral

When I started working with computers, one of the tools that shaped my way of thinking as a developer was FoxPro.
At the time, FoxPro felt like a complete universe: database engine, forms, reports, and business logic all integrated into a single environment.

Looking back, FoxPro was effectively an application framework from the past—long before that term became common.

Accessing FoxPro data usually meant choosing between two paths:

  1. Direct FoxPro access – fast, tightly integrated, and fully aware of FoxPro’s features
  2. ODBC – a standardized way to access the data from outside the FoxPro ecosystem

This article focuses on that second option.

What Is ODBC?

ODBC (Open Database Connectivity) is a standardized API for accessing databases.
Instead of applications talking directly to a specific database engine, they talk to an ODBC driver,
which translates generic database calls into database-specific commands.

The promise was simple:

One API, many databases.

And for its time, this was revolutionary.

Supported Operating Systems and Use Cases

ODBC is still relevant today and supported across major platforms:

  • Windows – native support, mature tooling
  • Linux – via unixODBC and vendor drivers
  • macOS – supported through driver managers

Typical use cases include:

  • Legacy systems that must remain stable
  • Reporting and BI tools
  • Data migration and ETL pipelines
  • Cross-vendor integrations
  • Long-lived enterprise systems

ODBC excels where interoperability matters more than elegance.

The Lowest Common Denominator Problem

Although ODBC is a standard, it does not magically unify databases.

Each database has its own:

  • SQL dialect
  • Data types
  • Functions
  • Performance characteristics

ODBC standardizes access, not behavior.

You can absolutely open an ODBC connection and still:

  • Call native database functions
  • Use vendor-specific SQL
  • Rely on engine-specific behavior

This makes ODBC flexible—but not truly database-agnostic.

ODBC vs True Abstraction Layers

This is where ODBC differs from ORMs or persistence frameworks that aim for full abstraction.

  • ODBC: Gives you a common door and does not prevent database-specific usage
  • ORM-style frameworks: Try to hide database differences and enforce a common conceptual model

ODBC does not protect you from database specificity—it permits it.

ODBC in .NET: Avoiding Native Database Dependencies

This is an often-overlooked advantage of ODBC, especially in .NET applications.

ADO.NET is interface-driven:

  • IDbConnection
  • IDbCommand
  • IDataReader

However, each database requires its own concrete provider:

  • SQL Server
  • Oracle
  • DB2
  • Pervasive
  • PostgreSQL
  • MySQL

Each provider introduces:

  • Native binaries
  • Vendor SDKs
  • Version compatibility issues
  • Deployment complexity

Your code may be abstract — your deployment is not.

ODBC as a Binary Abstraction Layer

When using ODBC in .NET, your application depends on one provider only:

System.Data.Odbc

Database-specific dependencies are moved:

  • Out of your application
  • Into the operating system
  • Into driver configuration

This turns ODBC into a dependency firewall.

Minimal .NET Example: ODBC vs Native Provider

Native ADO.NET Provider (Example: SQL Server)

using System.Data.SqlClient;

using var connection =
    new SqlConnection("Server=.;Database=AppDb;Trusted_Connection=True;");

connection.Open();

Implications:

  • Requires SQL Server client libraries
  • Ties the binary to SQL Server
  • Changing database = new provider + rebuild

ODBC Provider (Database-Agnostic Binary)

using System.Data.Odbc;

using var connection =
    new OdbcConnection("DSN=AppDatabase");

connection.Open();

Implications:

  • Same binary works for SQL Server, Oracle, DB2, etc.
  • No vendor-specific DLLs in the app
  • Database choice is externalized

The SQL inside the connection may still be database-specific — but your application binary is not.

Trade-Offs (And Why They’re Acceptable)

Using ODBC means:

  • Fewer vendor-specific optimizations
  • Possible performance differences
  • Reliance on driver quality

But in exchange, you gain:

  • Simpler deployments
  • Easier migrations
  • Longer application lifespan
  • Reduced vendor lock-in

For many enterprise systems, this is a strategic win.

What’s Next – Phase 2: Customer Polish

Phase 1 is about making it work.
Phase 2 is about making it survivable for customers.

In Phase 2, ODBC shines by enabling:

  • Zero-code database switching
  • Cleaner installers
  • Fewer runtime surprises
  • Support for customer-controlled environments
  • Reduced friction in on-prem deployments

This is where architecture meets reality.

Customers don’t care how elegant your abstractions are — they care that your software runs on their infrastructure without drama.

Project References

Minimal and explicit:

System.Data
System.Data.Odbc

Optional (native providers, when required):

System.Data.SqlClient
Oracle.ManagedDataAccess
IBM.Data.DB2

ODBC allows these to become optional, not mandatory.

Closing Thought

ODBC never promised purity.
It promised compatibility.

Just like FoxPro once gave us everything in one place, ODBC gave us a way out — without burning everything down.

Decades later, that trade-off still matters.

Oqtane Event System — Hooking into the Framework

Oqtane Event System — Hooking into the Framework

OK, I’m still blocked from GitHub Copilot, so I still have more things to write about.
In this article, the topic that we’re going to see is the event system of Oqtane.For example, usually in most systems you want to hook up something when the application starts.
In XAF from Developer Express, which is my specialty (I mean, that’s the framework I really know well),
you have the DB Updater, which you can use to set up some initial data.
In Oqtane, you have the Module Manager, but there are also other types of events that you might need —
for example, when the user is created or when the user signs in for the first time.

So again, using the method that I explained in my previous article — the “OK, I have a doubt” method —
I basically let the guide of Copilot hike over my installation folder or even the Oqtane source code itself, and try to figure out how to do it.
That’s how I ended up using event subscribers.

In one of my prototypes, what I needed to do was detect when the user is created and then create some records in a different system
using that user’s information. So I’ll show an example of that type of subscriber, and I’ll actually share the
Oqtane Event Handling Guide here, which explains how you can hook up to system events.

I’m sure there are more events available, but this is what I’ve found so far and what I’ve tested.
I guess I’ll make a video about all these articles at some point, but right now, I’m kind of vibing with other systems.
Whenever I get blocked, I write something about my research with Oqtane.


Oqtane Event Handling Guide

Comprehensive guide to capturing and responding to system events in Oqtane

This guide explains how to handle events in Oqtane, particularly focusing on user authentication events (login, logout, creation)
and other system events. Learn to build modules that respond to framework events and create custom event-driven functionality.

Version: 1.0.0
Last Updated: October 3, 2025
Oqtane Version: 6.0+
Framework: .NET 9.0


1. Overview of Oqtane Event System

Oqtane uses a centralized event system based on the SyncManager that broadcasts events throughout the application when entities change.
This enables loose coupling between components and allows modules to respond to framework events without tight integration.

Key Components

  • SyncManager — Central event hub that broadcasts entity changes
  • SyncEvent — Event data containing entity information and action type
  • IEventSubscriber — Interface for objects that want to receive events
  • EventDistributorHostedService — Background service that distributes events to subscribers
Entity Changes → SyncManager → EventDistributorHostedService → IEventSubscriber Implementations
                     ↓
               SyncEvent Created → Distributed to All Event Subscribers
  

2. Event Types and Actions

SyncEvent Model


public class SyncEvent : EventArgs
{
    public int TenantId { get; set; }
    public int SiteId { get; set; }
    public string EntityName { get; set; }
    public int EntityId { get; set; }
    public string Action { get; set; }
    public DateTime ModifiedOn { get; set; }
}
  

Available Actions


public class SyncEventActions
{
    public const string Refresh = "Refresh";
    public const string Reload = "Reload";
    public const string Create = "Create";
    public const string Update = "Update";
    public const string Delete = "Delete";
}
  

Common Entity Names


public class EntityNames
{
    public const string User = "User";
    public const string Site = "Site";
    public const string Page = "Page";
    public const string Module = "Module";
    public const string File = "File";
    public const string Folder = "Folder";
    public const string Notification = "Notification";
}
  

3. Creating Event Subscribers

To handle events, implement IEventSubscriber and filter for the entities and actions you care about.
Subscribers are automatically discovered by Oqtane and injected with dependencies.


public class UserActivityEventSubscriber : IEventSubscriber
{
    private readonly ILogger<UserActivityEventSubscriber> _logger;

    public UserActivityEventSubscriber(ILogger<UserActivityEventSubscriber> logger)
    {
        _logger = logger;
    }

    public void EntityChanged(SyncEvent syncEvent)
    {
        if (syncEvent.EntityName != EntityNames.User)
            return;

        switch (syncEvent.Action)
        {
            case SyncEventActions.Create:
                _logger.LogInformation("User created: {UserId}", syncEvent.EntityId);
                break;
            case "Login":
                _logger.LogInformation("User logged in: {UserId}", syncEvent.EntityId);
                break;
        }
    }
}
  

4. User Authentication Events

Login, logout, and registration trigger SyncEvent notifications that you can capture to send notifications,
track user activity, or integrate with external systems.


public class LoginActivityTracker : IEventSubscriber
{
    private readonly ILogger<LoginActivityTracker> _logger;

    public LoginActivityTracker(ILogger<LoginActivityTracker> logger)
    {
        _logger = logger;
    }

    public void EntityChanged(SyncEvent syncEvent)
    {
        if (syncEvent.EntityName == EntityNames.User && syncEvent.Action == "Login")
        {
            _logger.LogInformation("User {UserId} logged in at {Time}", syncEvent.EntityId, syncEvent.ModifiedOn);
        }
    }
}
  

5. System Entity Events

Besides user events, you can track changes in entities like Pages, Files, and Modules.


public class PageAuditTracker : IEventSubscriber
{
    private readonly ILogger<PageAuditTracker> _logger;

    public PageAuditTracker(ILogger<PageAuditTracker> logger)
    {
        _logger = logger;
    }

    public void EntityChanged(SyncEvent syncEvent)
    {
        if (syncEvent.EntityName == EntityNames.Page && syncEvent.Action == SyncEventActions.Create)
        {
            _logger.LogInformation("Page created: {PageId}", syncEvent.EntityId);
        }
    }
}
  

6. Custom Module Events

You can create custom events in your own modules using ISyncManager.


public class BlogManager
{
    private readonly ISyncManager _syncManager;

    public BlogManager(ISyncManager syncManager)
    {
        _syncManager = syncManager;
    }

    public void PublishBlog(int blogId)
    {
        _syncManager.AddSyncEvent(
            new Alias { TenantId = 1, SiteId = 1 },
            "Blog",
            blogId,
            "Published"
        );
    }
}
  

7. Best Practices

  • Filter early — Always check the entity and action before processing.
  • Handle exceptions — Never throw unhandled exceptions inside EntityChanged.
  • Log properly — Use structured logging with context placeholders.
  • Keep it simple — Extract complex logic to testable services.

public void EntityChanged(SyncEvent syncEvent)
{
    try
    {
        if (syncEvent.EntityName == EntityNames.User && syncEvent.Action == "Login")
        {
            _logger.LogInformation("User {UserId} logged in", syncEvent.EntityId);
        }
    }
    catch (Exception ex)
    {
        _logger.LogError(ex, "Error processing event {Action}", syncEvent.Action);
    }
}
  

8. Summary

Oqtane’s event system provides a clean, decoupled way to respond to system changes.
It’s perfect for audit logs, notifications, custom workflows, and integrations.

  • Automatic discovery of subscribers
  • Centralized event distribution
  • Supports custom and system events
  • Integrates naturally with dependency injection
Oqtane Silent Installation Guide

Oqtane Silent Installation Guide

OK, I’ve been wanting to write this article for a few days now, but I’ve been vibing a lot — writing tons of prototypes and working on my Oqtane research. This morning I got blocked by GitHub Copilot because I hit the rate limit, so I can’t use it for a few hours. I figured that’s a sign to take a break and write some articles instead.

Actually, I’m not really “writing” — I’m using the Windows dictation feature (Windows key + H). So right now, I’m just having coffee and talking to my computer. I’m still in El Salvador with my family, and it’s like 5:00 AM here. My mom probably thinks I’ve gone crazy because I’ve been talking to my computer a lot lately. Even when I’m coding, I use dictation instead of typing, because sometimes it’s just easier to express yourself when you talk. When you type, you tend to shorten things, but when you talk, you can go on forever, right?

Anyway, this article is about Oqtane, specifically something that’s been super useful for me — how to set up a silent installation. Usually, when you download the Oqtane source or use the templates to create a new project or solution, and then run the server project, you’ll see the setup wizard first. That’s where you configure the database, email, host password, default theme, and all that.

Since I’ve been doing tons of prototypes, I’ve seen that setup screen thousands of times per day. So I downloaded the Oqtane source and started digging through it — using Copilot to generate guides whenever I got stuck. Honestly, the best way to learn is always by looking at the source code. I learned that the hard way years ago with XAF from DevExpress — there was no documentation back then, so I had to figure everything out manually and even assemble the projects myself because they weren’t in one solution. With Oqtane, it’s way simpler: everything’s in one place, just a few main projects.

Now, when I run into a problem, I just open the source code and tell Copilot, “OK, this is what I want to do. Help me figure it out.” Sometimes it goes completely wrong (as all AI tools do), but sometimes it nails it and produces a really good guide.

So the guide below was generated with Copilot, and it’s been super useful. I’ve been using it a lot lately, and I think it’ll save you a ton of time if you’re doing automated deployment with Oqtane.

I don’t want to take more of your time, so here it goes — I hope it helps you as much as it helped me.


Oqtane Installation Configuration Guide

This guide explains the configuration options available in the appsettings.json file under the Installation section for automated installation and default site settings.

Overview

The Installation section in appsettings.json controls the automated installation process and default settings for new sites in Oqtane. These settings are particularly useful for:

  • Automated installations – Deploy Oqtane without manual configuration
  • Development environments – Quickly spin up new instances
  • Multi-tenant deployments – Standardize new site creation
  • CI/CD pipelines – Automate deployment processes

Configuration Structure

{
  "Installation": {
    "DefaultAlias": "",
    "HostPassword": "",
    "HostEmail": "",
    "SiteTemplate": "",
    "DefaultTheme": "",
    "DefaultContainer": ""
  }
}
  
Key Purpose Required
DefaultAlias Initial site URL(s)
HostPassword Super admin password
HostEmail Super admin email
SiteTemplate Initial site structure Optional
DefaultTheme Site appearance Optional
DefaultContainer Module wrapper style Optional

SiteTemplate

A Site Template defines the initial structure and content of a new site, including pages, modules, folders, and navigation.

"SiteTemplate": "Oqtane.Infrastructure.SiteTemplates.DefaultSiteTemplate, Oqtane.Server"

Default options:

  • DefaultSiteTemplate – Home, Privacy, example content
  • EmptySiteTemplate – Minimal, clean slate
  • AdminSiteTemplate – Internal use

If empty, Oqtane uses the default template automatically.


DefaultTheme

A Theme controls the visual appearance and layout of your site (page structure, navigation, header/footer, and styling).

"DefaultTheme": "Oqtane.Themes.OqtaneTheme.Default, Oqtane.Client"

Built-in themes:

  • Oqtane Theme (default) – clean and responsive
  • Blazor Theme – Blazor-branded styling
  • Bootswatch variants – Cerulean, Cosmo, Darkly, Flatly, Lux, etc.
  • Corporate Theme – business layout

If left blank, it defaults to the Oqtane Theme.


DefaultContainer

A Container is the wrapper around each module, controlling how titles, buttons, and borders look.

"DefaultContainer": "Oqtane.Themes.OqtaneTheme.Container, Oqtane.Client"

Common containers:

  • OqtaneTheme.Container – standard and responsive
  • AdminContainer – management modules
  • Theme-specific containers – match the chosen theme

Defaults automatically if left empty.


Example Configurations

Minimal Configuration

{
  "Installation": {
    "DefaultAlias": "localhost",
    "HostPassword": "YourSecurePassword123!",
    "HostEmail": "admin@example.com"
  }
}
  

Custom Theme and Container

{
  "Installation": {
    "DefaultAlias": "localhost",
    "HostPassword": "YourSecurePassword123!",
    "HostEmail": "admin@example.com",
    "SiteTemplate": "Oqtane.Infrastructure.SiteTemplates.DefaultSiteTemplate, Oqtane.Server",
    "DefaultTheme": "Oqtane.Theme.Bootswatch.Flatly.Default, Oqtane.Theme.Bootswatch.Oqtane",
    "DefaultContainer": "Oqtane.Theme.Bootswatch.Flatly.Container, Oqtane.Theme.Bootswatch.Oqtane"
  }
}
  

Troubleshooting

  • Settings ignored during installation: Ensure all required fields are filled (DefaultAlias, HostPassword, HostEmail).
  • Theme not found: Check assembly reference and type name.
  • Container displays incorrectly: Use a container matching your theme.
  • Site template creates no pages: Ensure your template returns valid page definitions.

Logs can be found in Logs/oqtane-log-YYYYMMDD.txt.


Best Practices

  • Match your theme and container.
  • Leave defaults empty unless customization is needed.
  • Test in development first.
  • Document any custom templates or themes.
  • Use environment-specific appsettings (e.g. appsettings.Development.json).

Summary

The Installation configuration in appsettings.json lets you fully automate your Oqtane setup.

  • SiteTemplate: defines structure
  • DefaultTheme: defines appearance
  • DefaultContainer: defines module layout

Empty values use defaults, and you can override them for automation, branding, or custom scenarios.

My Journey Exploring the Oqtane Framework

My Journey Exploring the Oqtane Framework

Mental notes on architecture, learning by reading source, and what’s next.

OK — so it’s time for a new article. Lately, I’ve been diving deep into the Oqtane framework, and it’s been a beautiful journey. It reminds me of my early days with XAF from Developer Express—when I learned to think in software architecture and modern design patterns by simply reading the code.Back then, documentation was scarce. The advice was: “Look at the code.” I did—and that shaped a big part of my software education. It taught me that good source code is often self-explanatory.

Even though XAF is still our main tool at the office (Xari & BIT Frameworks), we’re expanding. We’re researching new divisions for Flutter and React, since some projects already use those fronts with an XAF backend. I also wanted to explore building client-server apps with a single .NET codebase that includes mobile—another reason Oqtane caught my eye.

Why Oqtane Caught My Attention

The Oqtane team is very responsive on GitHub. You can open a discussion and get thoughtful replies quickly. The source code is clean and educational—perfect for learning by reading. There are plenty of talks and videos on architecture and module development; some are a bit dated, but if you cross-check with the code, you’ll be fine.

I’ve learned there are two steps to mastering a framework: (1) immerse yourself in material (videos, code, docs), and (2) explain it to someone else. These notes do both—part research, part knowledge sharing.

Oqtane Video References

A Missing Clip Worth Finding

There’s one clip I couldn’t locate where Shaun Walker explains that .NET already provides the pieces for modern, multi-platform, server-and-client applications—but the ecosystem is fragmented. Oqtane unifies those pieces into a single .NET codebase. If I find it, I’ll make a highlight and share it.

On Learning and Time

I’m trying to publish as much as I can now because I’m about to start a new chapter: I’ll be joining the University of St. Petersburg to learn Russian as my second language. It’s a tough language—very different from Spanish or Italian—so I’ll likely have less time to write for a while. Better to document these experiments now than let them sit in my notes for months.

That’s it for today. I hope these clips and notes help you understand Oqtane the way they helped me. Stay tuned—and happy coding!