标签：#csharp

TL;DR — AI assistants are producing C# that looks correct and passes review, but reintroduces production regressions we spent years training out of teams. I'm trying to find out whether other .NET teams see the same patterns — and what's actually catching them before merge. More AI-generated C# is landing in pull requests. Most of it is fine. But a specific category keeps slipping through — and it's the dangerous one, because it compiles, tests pass, and a human skim says "looks good." The pattern The code compiles. Tests pass. Review approves. Production finds out. These aren't syntax errors. They're architectural intent violations — the kind of thing a senior dev would have caught in review before PR volume tripled. Five regressions I keep seeing 1. EF Core read paths without AsNoTracking() Fine in dev. Expensive on a hot read path in prod. // ❌ Looks reasonable. Tracks entities you never mutate. var orders = await _db . Orders . Where ( o => o . CustomerId == id ) . ToListAsync ( cancellationToken ); Fix direction: AsNoTracking() on read-only queries, or a team convention documented in CLAUDE.md / Copilot instructions. 2. Captive dependency (scoped service in a singleton) Compiles. Runs. Wrong state across requests. // ❌ Singleton lives forever; scoped dependency does not. services . AddScoped < IOrderRepository , OrderRepository >(); services . AddSingleton < ReportCache >(); // ctor takes IOrderRepository Fix direction: align lifetimes, or inject IServiceScopeFactory instead of capturing scoped services. 3. Dropped CancellationToken The method accepts cancellation. The downstream call ignores it. // ❌ Signature honours cancellation; body doesn't. public async Task RunAsync ( CancellationToken cancellationToken ) { await Task . Delay ( 500 ); // overload with token exists } Fix direction: forward cancellationToken to every downstream async call that accepts one. 4. Swallowed exception Failure disappears. Monitoring stays green. // ❌ "Handle errors gracefully" —

The first post went over some of its capabilities. Over the past week Edict went v1.0, adding cursors for reading projections after command dispatch (to close some eventual-consistency gaps), a new type of projection that holds state inside the Orleans grain directly instead of a table, saga timeouts, schedules, an improved skills package and MCP server that ships with Edict, and more. Edict has now grown to over 75,000 lines of code and more than 1000 tests, and contains several deep mechanisms that have been fixed, broken, and fixed again. It is well past the point where I can hold all of Edict in my head. This post is about working with AI on large codebases, which I expect to be the first problem most software engineers have to solve. The context problem Years ago I was talking to a PhD candidate whose area of research was Natural Language Processing (NLP). He explained to me that one of the most difficult NLP problems was context. If a colleague says they need to pop out to pick their kids up from school, a scene can form in your head: one with a school, the layout of the road, people waiting, walking, driving, the environs. You may never have seen the school your colleague mentioned, but you can form a rich scene from your accumulated experience and use it to drive the rest of the conversation with a shared understanding. LLMs ingeniously dodge this entire issue by making it your problem. Just a word-probability machine Strip away the chat window and a Large Language Model (LLM) is doing one thing: predicting the next token. Give it a run of text and it returns a probability distribution over what comes next, samples one, appends it, and repeats. Companies like OpenAI and Anthropic then beat it into shape using techniques like supervised fine-tuning and reinforcement learning, which tune those probabilities in meaningful ways. That is why Claude is always telling me "Good framing" or "You've spotted...". It even called me "Bold" on one occasion. The probabilit

A while back, I needed to integrate SMS into a .NET project. Giant SMS had a REST API, but no official .NET client. The only existing library was a PHP one from 6–7 years ago, and it only covered two methods: send and getBalance. So I built my own. It now has nearly 2,000 downloads on NuGet. Here's exactly how I did it. The Problem Wiring up raw HTTP calls to the Giant SMS API in every project gets repetitive fast: Manually setting Authorization headers Remembering which endpoints use token auth vs. username/password Deserializing responses every time Scattering credentials across your codebase I wanted something that felt native to .NET. Configure once in appsettings.json , register with DI, and just call a method. Designing the Public API The first decision was the interface. I wanted consumers to never touch HttpClient directly, and I wanted methods that mapped clearly to what the API actually does: public interface IGiantSmsService { bool IsReady { get ; } Task < SingleSmsResponse > SendSingleMessage ( string to , string msg ); Task < SingleSmsResponse > SendMessageWithToken ( SingleMessageRequest messageRequest ); Task < BaseResponse > SendBulkMessages ( BulkMessageRequest messageRequest ); Task < SingleSmsResponse > CheckMessageStatus ( string messageId ); Task < BaseResponse > GetBalance (); Task < SenderIdResponse > GetSenderIds (); Task < BaseResponse > RegisterSenderId ( RegisterSenderIdRequest senderIdRequest ); } Seven methods, the full surface of the API, no more, no less. The IsReady property is a small but useful addition. It lets consumers do a quick sanity check at startup rather than discovering a missing token on the first SMS send: csharp _isReady = !string.IsNullOrWhiteSpace(_connection.Token) && !string.IsNullOrWhiteSpace(_connection.Username); Handling Two Auth Methods This was the most interesting design challenge. The Giant SMS API uses two different authentication schemes depending on the endpoint: Token-based (Basic Authorization header) —

What can I say. Anyone claiming that F# is good mostly for finance and data processing and C# for everything else, has probably never written a single line of practical F# code. In previous two parts of the article, I tried to demonstrate that with F# you can achieve the same goals as with C#, but with less verbose, repetitive, structural code. How it started. At some point, developers realized that global state with unrestricted data access causes many side effects, producing insecure, error-prone, and hard-to-maintain code as software grows larger. That is when the idea emerged to bring data and the code operating on it together into a single unit, restricting direct access to the unit’s internal state and making software more secure and predictable. This is how data encapsulation was born. Alongside encapsulation, abstraction was introduced — the process of hiding how behavior works. Encapsulation ( hiding data ) and abstraction ( hiding behavior ) remain two foundational pillars of Object-Oriented Programming. And that is how OOP has worked ever since — developers bring data and behavior together ( classes ) and define abstractions for them ( interfaces ). For example, for C# developers — including myself — this has become a daily routine. And we rarely question it, because OOP languages like C# leave us little choice but to structure code this way. But if you ask yourself whether this repetitive routine is always necessary, the answer is — no. You don’t need OOP concepts to build stateless, streamlined request–response, data-processing pipelines, because in such systems there is no long-lived state to hide and protect. You have a request, and almost immediately you have a response. After that, everything is gone. That is what I tried to demonstrate in the first two parts of this article by applying FP concepts. And even if you have a classical desktop application, you don’t always need to approach it in an OOP way. Functional programming handles side effects no

This is a submission for the GitHub Finish-Up-A-Thon Challenge Originally, I didn't plan to join...

Overview PDF text extraction is a common pre-processing step in data pipelines — ingesting research papers, legal documents, or reports before embedding or indexing. Both pypdf and PdfPig are pure managed-code parsers: no native binaries, no OCR, no system PDF renderer. They implement the same PDF specification operations in their respective languages. This makes the benchmark unusually clean: the performance difference is entirely due to language execution speed, not library architecture differences. Benchmark Setup 200 recent arXiv PDFs (mixed technical papers, 5–40 pages each). Tested on subsets of 10, 50, 100, and 200 files. Both libraries extract all text from all pages; output is validated for page-count agreement and character-count agreement within 15% (pypdf and PdfPig decode whitespace and encoding tables slightly differently). Results PDFs Pages Python (pypdf) .NET (PdfPig) Speedup 10 ~120 ~0.9 s ~230 ms 3.9× 50 ~600 ~4.2 s ~810 ms 5.2× 100 ~1,200 ~8.5 s ~1.4 s 6.1× 200 ~2,400 ~17 s ~2.7 s 6.2× The speedup grows slightly with corpus size, suggesting pypdf has a per-document startup cost that compounds as PdfPig's JIT gets warmer. Why PdfPig Is Faster PDF parsing is byte-heavy: every page is a stream of PostScript-like operators (move, show text, set font, etc.). Each operator must be lexed, looked up in a dispatch table, and executed against a graphics state machine. In Python, each operator dispatch is a Python method call — the CPython bytecode interpreter has overhead per call regardless of what the method does. In .NET, the JIT compiles the dispatch loop to native code the first time it runs; subsequent pages pay only the cost of the actual work. Additionally, PdfPig's content-stream parser operates on ReadOnlySpan<byte> — zero-copy slicing through the raw page bytes with no intermediate string allocations. pypdf builds Python string objects for each token. Key Code // PdfPig — zero-copy span-based page extraction public Result Extract ( string path )

Overview PageRank is the canonical graph algorithm. NetworkX implements it in pure Python — its dict-of-dict adjacency representation means every power-iteration step dispatches millions of Python attribute lookups. When the graph has 1.8 million nodes and 28.5 million edges (Wikipedia category hyperlinks), those lookups dominate the runtime. The .NET replacement uses a CSR (Compressed Sparse Row) matrix — two flat int[] arrays for the graph structure — and TensorPrimitives for the SIMD-accelerated normalization step inside each iteration. Benchmark Setup Five SNAP datasets of increasing size: Dataset Nodes Edges wiki-Vote 7,115 103,689 soc-Epinions1 75,879 508,837 web-Stanford 281,903 2,312,497 web-Google 875,713 5,105,039 wiki-topcats 1,791,489 28,511,807 Algorithm: power-iteration PageRank, damping=0.85, tol=1e-6. Both implementations converge to identical top-10 node rankings. Results Dataset Python (NetworkX) .NET (CSR) Speedup wiki-Vote (103k edges) ~0.8 s ~100 ms ~8× soc-Epinions1 (508k edges) ~8 s ~600 ms ~13× web-Stanford (2.3M edges) ~120 s ~5 s ~24× web-Google (5.1M edges) ~5.5 min ~12 s ~28× wiki-topcats (28.5M edges) ~47 min ~60 s ~47× The speedup grows with graph size because NetworkX's Python dispatch cost scales with edge count, while the CSR inner loop is a tight JIT-compiled SIMD pass. Why CSR Beats NetworkX NetworkX represents each node's neighbors as a Python dict. Iterating the adjacency in one power-iteration step means: Calling G.neighbors(node) — a Python method call Iterating a dict — unboxing int keys, chasing heap pointers Accumulating a float into another dict value — another boxing step That happens for every edge, every iteration, roughly 50–80 times to convergence. CSR collapses the graph to two arrays: rowPtr[n+1] (where each node's neighbors start) and colIdx[edges] (the neighbor list). Iterating neighbors of node v is a tight C loop from rowPtr[v] to rowPtr[v+1] . No Python objects, no dict hashing, no pointer chasing. Key Code // P

As part of my game development journey, I recently created Hidden Collector , a Unity-based game where players explore levels and collect hidden items while progressing through different challenges. This project started as a way for me to improve my Unity and C# skills, but it quickly became an opportunity to learn about game design, UI systems, audio management, scene transitions, and player experience. What I Worked On While building Hidden Collector, I implemented: Player movement and interactions Collectible item systems Multiple game levels UI menus and game screens Audio and sound effects Progress tracking Game flow and scene management Challenges During Development One of the biggest challenges was making different game systems work together smoothly. Something as simple as collecting an item often required updates to UI elements, game state management, and progression systems. Debugging these interactions taught me a lot about organizing Unity projects and writing maintainable code. What I Learned This project helped me gain experience with: Unity Engine C# scripting Game architecture UI implementation Audio management Debugging and testing Most importantly, I learned that building complete projects teaches far more than following tutorials. Play the Game You can try Hidden Collector here: https://sinxcos07.itch.io/hiddencollector Screenshots What's Next? I'm continuing to improve my game development skills by building new projects, experimenting with different mechanics, and learning more about creating engaging player experiences. If you try the game, I'd love to hear your feedback. By Suryansh Sinha (sinxcos07) Connect With Me GitHub: https://github.com/sinxcos07 LinkedIn: https://www.linkedin.com/in/suryansh-sinha/ Play Hidden Collector: https://sinxcos07.itch.io/hiddencollector

There's a question that keeps surfacing across engineering teams — sometimes in architecture reviews,...