Posts on InternetException

I built an app and it cost almost $40

Fri, 28 Aug 2026 17:33:25 +0200

I posted already about me selfhosting LLM models. Right now I have two devices:

Framework Desktop
Custom Ryzen Threadripper server with R9700 32GB GPU

Both are decent with llama.cpp. The biggest issue is prefill times.

I have from work subscription for Claude and Codex which I tested too for last couple of months. It is OK but lack of thinking traces really hurt to keep up with what models are doing.

Surprised LLM

Tue, 18 Aug 2026 15:31:20 +0200

I know that LLM or Large Language Models are designed to mimic what people write, but still from time to time it surprises me and make me chuckle.

I am working on small tool that runs agents in confinement that gives those programs limited set of tools that you would want them to use only. Otherwise they may try to use sudo to change system settings or kill all your docker containers instead of only few were created by them. I had several times the same problem with codex not being able to start in such sandbox because of permissions issues. I am using Qwen 3.8 for now and it thinks a lot and this particular case it was thinking about those permissions.

Short review of Qwen 3.8 27B

Mon, 17 Aug 2026 17:27:09 +0200

Qwen 3.8 is out since Friday, 17 of August. I am running it on my own server since Saturday. I am running Q4 quants from unsloth .

So far it feels like a step up from Qwen 3.6 27B, which sometimes had problems figuring stuff in my project that I am developing almost entirely using models - just as a test of how far I can push LLMs in agentic coding - mostly local models, I must add. Qwen 3.8 seems like having no problem at all.

Quick review of Laguna S 2.1

Fri, 24 Jul 2026 15:06:50 +0200

Just couple of days ago poolside released their new model: Laguna S 2.1. I was happy to try it since it have native support for 1M tokens in context. I am doing development of hobby projects on my phone, via SSH prompting models to do stuff and 250k tokens of context using Qwen 3.6 27B usually means going through few rounds of compaction before I am happy with the results. Having longer context would be better - in theory - without those ugly interruptions and reiterating stuff to agent that were lost in previous session.

Problem with local network in termux

Sat, 04 Jul 2026 15:42:04 +0200

I bought new phone and I started installing applications from previous device. When I installed termux I tried to copy ssh keys and ssh config from previous installation.

This is where problems started. I could not reach any local address. Nothing in 10.x.x.x was working. No pings no ssh.

I could do that to public IPs just fine.

I tried to install new version of packages but it also did not work.

Remote phone development.

Fri, 19 Jun 2026 17:18:48 +0200

I remember when I started selfhosting my own instance of GitLab, which had simple code editor that allows you to edit code quickly inline. Then they integrated with fork of Visual Studio Code and allowed you to code in real repository directly in GL.

I was thinking of doing quick code changes to applications at this way, change some configuration of my API, quick fix of my mobile phone for home automation and etc.

High Five with LLM

Fri, 12 Jun 2026 18:35:39 +0200

I do not know… I find it funny.

After few days fighting with local LLM to force it to write documentation and comments in code, in good quality I must add, and the forcing it to write readable unit tests, after I wrote:

File is untracked (new). Build passes. Ready for review.

I just answered.

awesome mate. It is friday we deserve a brake. High five.

I was thinking for a bit and answered:

Running Qwen 3.6 on vLLM with 6 draft tokens

Wed, 20 May 2026 20:13:55 +0200

Just out of curiosity I tested Qwen3.6-27B model with MTP speculative decoding with 6 draft tokens. Whole script looked like this:

docker run --rm --name="${2:-qwen3.6-medium-vllm}" \
 --group-add=video \
 --cap-add=SYS_PTRACE \
 --security-opt seccomp=unconfined \
 --device /dev/kfd \
 --device /dev/dri \
 -v /cache/huggingface:/root/.cache/huggingface \
 -v /cache/vllm:/root/.cache/vllm \
 --env "HF_TOKEN=$3" \
 -e TORCH_ROCM_AOTRITON_ENABLE_EXPERIMENTAL=1 \
 -e VLLM_ROCM_USE_AITER=1 \
 -e FLASH_ATTENTION_TRITON_AMD_ENABLE=TRUE \
 -p "${1:-8000}":8000 \
 --ipc=host \
 --entrypoint '/bin/sh' \
 vllm/vllm-openai-rocm:nightly \
 -c "pip install fastokens; vllm serve Lorbus/Qwen3.6-27B-int4-AutoRound \
 --gpu-memory-utilization 0.48 \
 --dtype half \
 --optimization-level 1 \
 --enable-prefix-caching \
 --performance-mode interactivity \
 --kv-cache-dtype fp8_e4m3 \
 --language-model-only \
 --reasoning-parser qwen3 \
 --enable-auto-tool-choice \
 --tool-call-parser qwen3_coder \
 --speculative-config '{ \"method\": \"mtp\", \"num_speculative_tokens\": 2}' \
 --max-num-seqs 1 \
 --tokenizer-mode fastokens \
 --max-num-batched-tokens 262144 \
 #--max-model-len 262144"

Those are results with vLLM bench tool:

Lllama.cpp vs vLLM running Qwen 3.6 27B

Wed, 20 May 2026 17:27:57 +0200

After optimizing lately my vllm startup I started optimizing performance of vLLM during inference. Thing it had abysmal performance spanning like 2-4ts. Or even 0.5ts with 180k context. That was really terrible and I started to think that I wont be able to get better performance out of Strix Halo inside my Desktop Framework.

But… I remembered people praising it and its great usability with agents and Qwen models. How is that possible since it have so terrible performance.

Lets ask Claude!

Thu, 14 May 2026 10:31:18 +0200

AI hype is calming down in some companies , but in others like in the one I am currently consulting for, we still on the hype.

Few days ago I was asked by my peer about some problem with the tool that I was mostly developing. This is fine are there are always some bugs in every software, though bug was concerning behavior on bad data and therefore out of scope really. Still I offered to help, regardless of the fact that correct fix would be to fix the bad data.

Junie Coding Agent short review

Wed, 13 May 2026 07:40:03 +0200

I have been running few coding agents in my spare time on my personal projects and Junie does not look great when paired with local model

First of all, local models are capable but their ability to answer prompts in sustainable speeds degrades very quickly. I was experimenting with Qwen3.6-35B-A3B and Qwen3.5-35B-A3B. Both seems to be fairly capable, which is great for a local model that can run on decent GPU that is few years old (I have Radeon 7900 XTX with 24GB of VRAM) - that is outstanding considering what we thought about computers i.e. ten years ago. If someone then would say to me that we will be able to converse with our applications in human language and instruct them to do useful stuff, I would say: ‘Impossible!’. But we can and it does work but it have limitation, mainly in context size.

Optimizing vLLM startup time

Thu, 07 May 2026 20:38:15 +0200

vLLM has very slow time to first token (TTFT). On AMD Ryzen AI Max+ 395 it can take even 10 minutes depending on exact settings and model size.

To make my user experience of using those things myself I need to have comparison of llama.cpp and vLLM startup time and performance. But to do this I needed to perform optimizations of startup of vLLM since it is even slower to start then it was with v0.17.

My new hobby: Asking LLMS to generate ASCII Hamsters

Mon, 04 May 2026 16:53:43 +0200

I was playing over the weekend trying to write POC of C# interactive console. It is not ready yet but I wanted to have some cute hamster as logo for this project - I will probably incorporate it to some degree into Hamster Wheel , so cute hamster seemed appropriate.

Just for fun I decided to use LLMs to generate it since maybe I will not finish this project ever so trying to do it myself would be waste of time. And result probably would be terrible since I do not have any artistic skills.

Running Qwen 3.5 on Strix Halo using vLLM and Docker

Sun, 26 Apr 2026 22:20:40 +0200

I have Desktop Framework for 2 months now. It is pretty capable for interefence with small contexts. With longer contexts it degrades quickly. It can for example can run 100k tokens contexts for 10minutes. Because of those limitations I decided to try vllm again - maybe it will be quicker. Two months is long time in LLM frameworks last time I was using vllm it has version 0.17. Now stable is 0.19.

Using Mistral Vibe with local model

Fri, 24 Apr 2026 22:00:00 +0200

Vibe was the first coding agent I tried to use with local model. It works and it is fun to have ability to to just tell the computer to do something you want in plain English or even in your own language if your model support it.

Installation and Configuration

There are several installation methods for vibe that are available along to the usual, security nightmare of running of piping curl-downloaded scripts directly to bash :)
I advise using other installation methods for example via uv:

Using Junie agent CLI with local model

Fri, 24 Apr 2026 19:04:03 +0200

Jet Brains have their own coding agent. Since I am already using Rider and all coding agents are pretty slow and inefficient with what they do I wanted to test if Junie have better integrations with IDE which would make things much faster.

For example if you will have following unused namespace in Rider.

Removing it all cleaning all unused namespaces in entire project or even entire solution is just few clicks away. Or key strokes.

AI agent critique of itself

Thu, 23 Apr 2026 11:30:40 +0200

I like to self-host my own services. I have my own email server. I have my own cloud storage and sync and backup. I want to have this secure and safe while having my data private.

But I need to learn and be up to date with current technology.

I decided to use Junie with my own, self hosted models that I run on my own hardware . In the docs they say it can be run on top your own custom provider and model. But it does not work with Rider plugin. You are limited to their own limited list of models.

Testing Qwen 3.6

Thu, 16 Apr 2026 22:26:06 +0200

Today I noticed that new version of Qwen was realesed . Only 35B-A3B for now but this is the same size of Qwen I was using previously. Usually (as with every other software) it is better to wait few weeks for fixes, in this case for re uploads of the model, that is done after testing/template adjustments, but it is not a problem to do few tests. I decided to give it a spin. I added it to my llama swap instance and started it up.

Bonding Network interfaces in Debian Forky

Tue, 31 Mar 2026 21:18:00 +0200

I am experimenting lately on with Nanobot and running self hosted language models on Desktop Framework.

I was trying to run lately Qwen 3.5 397B model on this small PC and it was a bit hard because I just do not have enough disk space on it. I bought just small NVMe SSD disk. With current prices, I planned to use disk space on other server I have at my hope as storage

Running Qwen 3.5 397B on Desktop Framework

Mon, 30 Mar 2026 17:53:10 +0200

Today I finally was able to download and run largest Qwen 3.5 model. I downloaded the smallest quant even if a bit bigger probably would fit too, but I want to run also few smaller models at the same time, so 107GB for UD-IQ1_M is the biggest I was able to spare.

I downloaded and was able to run it with llama.cpp with the following command:

./llama-server \
 -hf unsloth/Qwen3.5-397B-A17B-GGUF:UD-IQ1_M\
 --ctx-size 32000 \
 --no-warmup \
 --no-mmap \
 --flash-attn on \
 --n-gpu-layers 1 \
 --host 0.0.0.0\
 --port $1 \
 --direct-io \
 --jinja --chat-template-file /data/apps/vllm/qwen3_nonthinking.jinja

My intention was to test it with nanobot to see if this will be capable agent for my AI assistant.

My first PR to Nanobot

Sat, 28 Mar 2026 18:52:58 +0100

I am blogging last few weeks about my new aventure with nanobot. I am running model for that little assistant of mine on Desktop Framework motherboard I bought couple of days ago.

Few days ago I was working on my own fork of nanobot. I was not happy how few things are not entirely correctly implemented in this project. For example command support in Matrix channel. I created PR for that one here . Unfortunately it was not merged yet. I am not sure why. Maybe it is mising few unit tests.

Swapping vLLM to Llama.cpp

Thu, 26 Mar 2026 20:04:51 +0100

Yesterday I posted about vLLM being nice but optimized for different usage I am being really interested in . I commented on Llama Swap GH issue asking for help my problem of vLLM being shutdown before it actually started. During writing that comment I was thinking about Llama.cpp and it inability to run Qwen 3.5. It was at the beginning of March when I started (again) experimenting with LLMs. I tried to run Qwen models with release version of Llama.cpp and it was failing with ‘model architecture of Qwen3 is not found’ (or something like that; I do not really remember). Because of that I switched to using vLLM, even if it could not run Qwen 3.5, it could run Qwen 3 just fine from the docker image. Sglang seemed like not a good choice for my needs.

vLLM is fast but kind of slow

Wed, 25 Mar 2026 18:53:05 +0100

I was using vLLM to run my AI assistant models for few weeks now. It is quite nice, production ready framework that have few bugs . ROCm support and by extent Strix Halo support is quite nice, even if AMD GPU docker image does not work on Strix Halo without aiter in custom version.

Interference speeds are quite nice. You can get 20-30t/s out of new AMD APU. Which is totally usable. Here is sample test I done on this device via LLM API Throughput Benchmark:

Integrating my AI assistant with StartPage

Fri, 20 Mar 2026 21:09:48 +0100

Today I was fixing an issue with my assistant having a hard time accessing web search . I fixed that but I was not entirelly satisfied with the result. Browsh is great for that used but I am using StartPage on my PC and on my phone. Ability to see similar or even the same web search result for me and my assistant would be great.

How hard it can be to actually fetch the result od search via CLI? Actually it is not that hard. I was able to figurę it out though it requires very specific set od parameters, headers being sent to very specific address.

Teaching my AI assistant to surf the web part 2

Fri, 20 Mar 2026 12:01:32 +0100

Few days ago I thought my assistant to do web search via Browsh and Duck Duck Go. It was working for some time with small problems, like for example I changed host name of the machine hosting nanobot and forget to adjust the host name in skill URL. Another small pain point is that sometimes returns timeout instead of web page content and not falls back to Brave Search API, and since I do not like that service, there is no API key and it falls back to DDG via python code. So it is mostly slow but I am still happy with that.

Desktop Framework AI assitant part 3

Mon, 16 Mar 2026 21:31:03 +0100

Recently I started playing with Nanobot which is a bit like OpenClaw but, in my opinion much better since it is small and simple and have native integration with custom OpenAI APIs providers. I decided to use vLLM because I wanted to use Qwen 3.5 because according to my quick reaserach it is pretty good in such agentic usage and Qwen models are built with integration with SgLang and vLLM frameworks in mind. Also AMD is testing their own drivers and libraries, ROCm on docker images with SgLang and vLLM and Desktop Framework is AMD APU. So all of it since like a good idea since I decided to go with Strix Halo unifies memory architecture for my AI assistant. I bought Desktop Framework motherboard and played with it a little to test models performance on ROCm inside the docker. It was not blazing fast but enough to actually have working solution.

Adding voice recognition to my AI assistant

Sun, 15 Mar 2026 23:43:00 +0100

It is amazing how time flies… I am living at my current house already 9 years almost. I did put up a fence around the house in 2019. Next year I installed metal gate with automation and was working on a way to make it more ‘smart’. My first attempt was to have small raspberry Pi connected via some metal wires and solid relays to the gate electrical engine. It is working like that till now.

Running AI assistant on Qwen 3.5

Sun, 15 Mar 2026 22:35:00 +0100

I did make mistake installing my favorite Debian Linux distribution on Desktop Framework PC I am using to run my LLM models. Probably some hacking during installing vLLM and all libraries and drivers and I would be able to make it work. But is it really worth it? I think it was not. So I switched to Ubuntu and I am running Qwen 3.5 second day now.

For now I can say that I am very happy. It seems to be able to reason much better than its predecessor Qwen 3. It have pretty good vision capabilities. It is able to fix its own mistakes, which makes its usage much easier.

Integrating my AI assistant with pyLoad

Sun, 15 Mar 2026 20:59:04 +0100

I am using pyLoad download manager for quite some time now. Previously when I was running windows server I was using JDownloader. I think it was better in few things, for example it was able to automatically name a package based on the name of the file from the link. So if I would be trying to download https://cdimage.debian.org/debian-cd/current/amd64/iso-cd/debian-13.4.0-amd64-netinst.iso, it would automatically name a package ‘debian-13.4.0-amd64-netinst.iso’. Folder with downloaded file would be named the same too. Much more convenient. If I would drop multiple links for the same file it would be downloading them in parallel, downloading them much quicker. Though with current speeds in average home maybe this is not necessary anymore. It had better UI, in my opinion, but maybe I was more accustomed to windows desktop then; certainly it was better to have single view instead of 3 views like in pyLoad:

Running Qwen 3.5 on AMD Ryzen AI Max+ 395

Sat, 14 Mar 2026 15:09:00 +0100

I bough lately Desktop Framework with intention of running Qwen 3.5 as model for my AI assistant on Nanobot . At first I could not run this model on this hardware for some weird bug in one of the libraries. I explained why and how I fixed it here . In this post I will just put the list of packages that I used to run Qwen finally and vLLM command switches and parameters.

Switching my AI assistant OS to Ubuntu

Fri, 13 Mar 2026 20:38:31 +0100

I did make a mistake trying to run a vLLM server on Debian. I theory everything should be OK: Debian is stable, have docker, docker have images for vLLM so I should be able to run all the models on docker on Debian. Also it should not really matter what distro is as long as it have new kernel and docker works. That in theory. I practice I could not force some models to run and Qwen 3 even if it was capable enough it felt limiting at times. For example I was unable to force it to work in agentic loops. I am not sure what was the problem, probably some configuration issue that I am still unable to understand and fix. Still that was just a simple problem that required me to give my assistant some nudge with another prompt. It was not as bad another issue though.

Futuristic API design

Fri, 13 Mar 2026 16:06:36 +0100

Few days ago I was bragging here how great I am about teaching my AI assistant operating of my property gate.

Today I noticed that it no longer works. I asked it again to:

Open the gate

while wondering why am I actually doing it since I can just press the button on my phone… but well people are sometimes radicoulous like that.

Anyway I wrote message as above and I got answer like below:

Assigning VRAM on AMD AI Max+ 395

Sun, 08 Mar 2026 11:33:00 +0100

Few days ago I started playing with AI assistant and I decided to buy new hardware dedicated to running LLM. I bought Framework Desktop board. So far it is really great but running more than one model is a bit difficult and I wanted to test few things while my nanobot is running undisrupted.

I did quick search on my phone before buying and people were saying that it is possible to set it up via BIOS settings up to 96GB.

Desktop Framework AI assistant part 2

Thu, 05 Mar 2026 18:59:13 +0100

Today my new NVMe disk arrived. I bought it for my new Desktop Framework mini PC on new AMD APU intended for AI workloads. Certainly it will be better suited for running my AI assistant then my daily working PC. Though performance will be a bit worse. But still better than CPU.

Anyway I decided to split my assistant into 2 parts. The nanobot part will be working on my main server and model (or models) that will be working on Desktop Framework. This way more powerful machine will be running AI assitant UI and operations and LLM capable device will do faster interference - that will be the Framework device. Nanobot communicate with model via OpenAI API anyway so it is not a problem. Maybe a bit of security in terms of HTTPS with some auth would not hurt. Or maybe I will put it in private subnetwork that can’t be accessed from outside and only have internet when necessary? Running dockers and all of those python AI firework is not exactly secure but maybe I will deal with that later.

Desktop Framework AI assistant part 1

Wed, 04 Mar 2026 16:06:41 +0100

I ordered Desktop Framework motherboard last week and today it arrived. It was neatly packaged and did not come with anything else then some piece of plastic to mount your own fan on top of the radiator. But I did not need much else. At least for first tests.

I had some PSU laying around that was compatible. I have some old fan that I was able connect to board even if it was a bit to small but would cool it a bit at least.

Working on AI assistant soul

Tue, 03 Mar 2026 22:30:57 +0100

Recently I started playing with nanobot . It is pretty cool tool. I was able to connect it to my own instance of vLLM running Qwen 3. I must say that so far it is great experience. Though it was sometimes annoying.

What was itching me from the start was abuse of emojis by the bot. I am using emojis from time to time myself, but what was produced by LLM was just excessive. Almost every message was finished by:

How I taught my AI assistant to read websites

Tue, 03 Mar 2026 22:16:00 +0100

Main problem with running your own AI assistant is speed. Nanobot is lightweight but main problem it time od the reponses generated by the model. I am running right now Qwen 3 27B parameters. Average reponse varies but it is at least few seconds. If I would have faster GPU than Radeon 7900 XTX than probably it would faster but prices of GPUs are crazy now.

One of the things I have been testing my little chatbot with was searching for stuff on the internet. For example I like to check prices of some PC parts from time to time. Just out of curiosity and to have a knowledge how crazy it is (NVMe SSDs of 8TB sizes are at least 2x times pricier than last year, on the other hand salary for my profession dropped to 60-70% last 3 years, without inflation, with inflation it would be around 58%). Things are crazy.

Running AI assistant on Threadripper PRO

Sun, 01 Mar 2026 13:57:00 +0100

The answer is: don’t do that to yourself.

Recently I started playing with AI assistant using nanobot. It is good but so far I am running it on my daily working PC. And it is not great experience so far . So I am exploring some other way that will allow me to do that, especially when I am not at home but still want to use my little helper to do stuff. Running big PC in your office next to your bedroom is not the best idea. And electricity bill would also not be that great.

Self hosting your AI assistant efficiently

Sun, 01 Mar 2026 06:27:00 +0100

I wrote previously why running models that you want to use on your daily driver is not great. Basically it is not convenient and it is fine for testing but if you want to run while doing something else it not great experience.

Of course you can have more powerful device so you will not notice few models running in the background but this is like running a fan with heater under your desk.

I taught my virtual assistant to operate my gate and it feels great

Sat, 28 Feb 2026 17:57:00 +0100

Lately I started playing with an LLM operating as my own virtual assistant . I was skeptical, I mean I still am skeptical, about this whole ‘AI will take your job’, but now I do understand the appeal. You can finally have a computer do your biding in a way you control. Previously computer were complicated beasts. You needed to appeal to it like with threats in form of complicated keystrokes forming more complicated commands evolving in even more complicated programs that may or not do the thing you actually wanted. Or it may be not amused by your plea and swallow your data. You would never know for sure if you would not learn those complex magical incantations in magical schools called Universities which takes years and a lot of money. Or you could hire those magicians, Software Engineers, or even Archmages, that commands magicians, called Software Architects that would be appeasing the capricious spirit of the machine to direct flow of money in a company in a correct way. But this is costly and those spells takes years. So you may hire more magicians, entire houses of magic that will appease, train and teach your pricey machine beasts to do what you need them to do. Those houses are called Software Houses for Enterprise Solutions. And they will force those machines to do what you need them to do. Sometimes they will replace entire herds of machines because previous ones were not cooperative enough. And this is costing you riches beyond riches. If you did not had your weight in gold to buy such service you could buy trained machine to do what you need that were not even at your house but somewhere far beyond the realms of your kingdom, in a place called The Cloud. Or you could buy complex incantation that needed only few keywords in correct place to operate your business - CRM or Accounting System or ERP.

Running LLM on your daily PC and why it is not good

Fri, 27 Feb 2026 15:11:00 +0100

Two days ago I started playing a bit with my own AI assistant. It has been very great and funny experience so far but with few wrinkles. For example:

performance of new models on previous generation GPU (Radeon RX 7900 XTX) is surprisingly good with average of 28t/s but it takes whole VRAM and put a load on PC overall.
It takes whole lot of disk space. I.e. one vllm docker image to download and extract needs about 18gb; this seems a bit excessive.
Rocm installation needed a bit like 9gb to install all of its binaries. How much code do you need to produce 9gb of compiled code? This is crazy.
I have 16c/32t CPU and it seems to be enough for most of my workload but with llm in the background sometimes most of it is in use.
32gb is not enough, but with current prices I can suffer for now. :)
If you want to swap models a lot you need a lot of disk space since bigger models takes several gigabytes.
You need decent internet speed to be able to download it.

Yes you may say that it is fine in 2026, everybody have fiber connection now! Yes but when you want to test something and new docker image is downloading in with 1mb/s speed, you may have a walk in that time.

Connecting Nanobot to Matrix

Fri, 27 Feb 2026 12:19:00 +0100

I started digging into Nanobot code to check why it cannot connect to Matrix server despite the config being correct and nanobot gateway being executed. Just in case I messed up something I did recreate entire nanobot workspace by doing nanobot onboard after removing entire directory. Did not work.

Closer inspection of the code shed some light on the problem: the config was not being used because channel was never instantiated. That part of the code was gone for some reason despite the Readme stating it is possible to integrate nanobot with Matrix.

Adding the bot to Matrix server

Thu, 26 Feb 2026 15:04:00 +0100

I have my own matrix server for my own use. Usually it is used by me a bit by my family and most of it by my own services to sent me some notifications.

Lately I started playing with nanobot. It is personal AI assistant like OpenClaw. I wanted to be able to chat with it in my own Matrix via my own phone. To do that I needed to create new user dedicated

Running my own AI assistant.

Thu, 26 Feb 2026 10:32:00 +0100

I recently started experimenting with my own AI asisstant. I decided to skip on OpenClaw for being massive slop of 400k LOC of vibe coded monstrosity. Looking trough web I found that there is also NanoClaw which seems much better but looks like it is tied to Anthropic service and I like owning my own data and I self host what I can.

That left me with Nanobot being my only option from bigger open source projects that are (relatively) known, are under active development and have big user base.

Moving /boot partition

Wed, 18 Feb 2026 11:55:26 +0100

For a while I had a problem with my boot partition being too small. It was fine for one kernel image but not for two. So whenever kernel was updated it was failing because of lack of space. Five years ago 500MB was enough but not nowadays.

Today I decided to finally fix that.

First, I did created new LVM volume for /boot:

sudo lvcreate -L 1G -n boot root-vg

Purging smtp services from Debian

Wed, 18 Feb 2026 10:45:50 +0100

Lately I purchased mew dedicated server with intention to use it as my new running mail email server - mailcow, my blog, and some other services. When I was migrating mailcow suite everything was fine except postfix image failed to start.

I had problem like that already on Debian and usually I was dealing with that by disabling exim:

sudo systemctl disable exim.service

This was fine till major Debian update that was reinstalling exim and causing it to steal port 25 again. So this time I decided to uinstall it completely.

Enhancing your .NET API with query language

Thu, 30 Oct 2025 21:14:03 +0100

What is query language? SQL is Structured Query Language and you use it to query your database for data you are interested in. It does not have to do anything with your API. If you ever were working with big XML files you probably used XQuery or XPath, query languages designed to XML for data. There are specialized languages that are designed to work with APIs to query them for data. For example, you probably heard about GraphQL or even used it when you were working with some API.

HLinq design

Tue, 28 Oct 2025 21:14:55 +0100

A bit of history

I remember the day when I started writing the HLinq query language. It was weekend and I was experimenting with writing Single endpoints for Hamster Wheel. Single endpoints were almost done and next step was to write List endpoint. This would allow to get list of records of specific type. After I got single GET returns correct data for chosen record id, list should not be hard to implement. Then just create, update and delete. Basic CRUD API in Hamster Wheel would be finished. I was thrilled that I actually was able to do such thing.

Disable Xunit Test on specific OS

Sat, 04 Oct 2025 22:51:15 +0200

Lately I was having a problem with some of my tests causing problems for other developers on the project I am working on. They were using Windows for their development while I am using Debian for my work. The application that was being tested was not designed for non Unix system - now rarely anybody is hosting they production code on Windows Server - everybody is using Unix for .net.

Writing a simple C# source code generator with Fluent API

Thu, 02 Oct 2025 16:05:39 +0200

Introduction

Usually in .NET world when you want to write generic code you use generic method or classes. But sometimes it is still not enough when you want to automate some things. I.e. you want to dynamically register in DI all the classes that implements some interface. Probably, in this case, you would do something like that:

var implementations = typeof(MyClass).Assembly.Types
 .Where(t => t.GetInterfaces().Any(i => i == typeof(IMyInterface))
foreach(var implementation in implementations)
{
 services.AddScoped<IMyInterface>(implementation);
}

And it would be completely fine. Even if Reflection is pretty slow, your application would not be getting significantly slower because of that. Certainly you would not notice the slow down.

HamsterWheel.FluentCodeGenerators is out

Wed, 01 Oct 2025 16:58:21 +0200

I released today first new Nuget package connected to my new project that I am working on for sometime: dynamically configurable, zero-downtime API, with low-code flows to write small pieces of logic - Hamster Wheel.

System heavily relies on dynamic source code generation. For that I wrote
https://www.nuget.org/packages/HamsterWheel.FluentCodeGenerators and https://www.nuget.org/packages/HamsterWheel.FluentCodeGenerators.Abstractions
small package that helps with writing C# incremental source code generators for Roslyn compiler.

Original Roslyn APIs are a bit hard to use so this package helps with that. Also it helps with:
\

Convert LVM logical volume Raid 10 to Raid 5

Thu, 11 Sep 2025 09:28:48 +0200

Two weeks ago I was running out of space on my main raid matrix on my server. Server was configured and assembled few years back, and at that time I had about 700GB of data there so having RAID 10 of 4 1TB disks (that account for about 1,7TB of actual space for data) was enough. Especially when you count in a price of 2TB (or 4TB) NVME M2 disk back then.

No Display Manager in Debian Trixie

Thu, 15 May 2025 10:47:46 +0200

Lately I decided to try Podman Quadlets. It seemed to be really interesting way to achieve similar results as with Docker and Compose, but it have nicer integration with the system (i.e. ability to see container in Cockpit along with other systemd services, or triggering container by socket, so less frequently used containers won’t be running all the time).

Everything went fine except for unavailability of buildah package in Debian stable apt channel (or the version that were needed for podman at least). Quick search on the internet and I found the answer is to reference this specific package from the testing channel of apt.

Monitoring vs Observability

Thu, 20 Feb 2025 09:25:14 +0100

What is the difference between monitoring and observability?

Monitoring is process of collecting live events from your application, traces, debug messages, warnings and exceptions. The same process may involve sending those events to some storage or application that analyze or visualize those messages. I.e. in asp.net core applications monitoring of application involves calling a ILogger and sending those logs or critical exceptions to AppInsights. Monitoring may also involve activity of development team members to active watch those logged messages for any potential errors or problems that may impact availability of production environment.

Solid

Wed, 26 Oct 2022 14:14:41 +0200

The single-responsibility principle - piece of software should have only one responsibility
The open–closed principle - things should be open for extension but closed for modification
The Liskov substitution principle - changing interface implementation should not brake software
The interface segregation principle - it is better to have more specialized interfaces than single big one
The dependency inversion principle - dependencies should be abstracted via interfaces

Further read:

https://en.wikipedia.org/wiki/SOLID

.NET/C# interview questions

Wed, 26 Oct 2022 10:21:52 +0200

Equals and GetHashCode

Tue, 25 Oct 2022 16:36:49 +0200

Custom implementation is usually only necessary if default, reference equality is not enough. In example:

entities value based equality
value equality of deserialized objects
custom comparisons based on other criterias

public class ImaginaryNumber : IEquatable<ImaginaryNumber>
{
 public double RealNumber { get; set; }
 public double ImaginaryUnit { get; set; }

 public override bool Equals(object obj)
 {
 return Equals(obj as ImaginaryNumber);
 }

 public bool Equals(ImaginaryNumber other)
 {
 return other != null &&
 RealNumber == other.RealNumber &&
 ImaginaryUnit == other.ImaginaryUnit;
 }

 public override int GetHashCode()
 {
 return HashCode.Combine(RealNumber, ImaginaryUnit);
 }
}

Difference between Controller and ControllerBase

Tue, 25 Oct 2022 16:28:23 +0200

Controller type inherits from ControllerBase and have Views support
ControllerBase type is better suited for APIs

Further read:

https://learn.microsoft.com/en-us/aspnet/core/web-api/?view=aspnetcore-6.0

Kestrel

Tue, 25 Oct 2022 15:56:40 +0200

default application server for .net
cross platform
best performance and memory utilization
may not have some advanced features of HTTP.sys on Windows
may be used with proxy server or as internet facing server

Further read:

https://learn.microsoft.com/en-us/aspnet/core/fundamentals/servers/kestrel?view=aspnetcore-6.0

ASP.NET Core controller level exception handling

Tue, 25 Oct 2022 15:53:52 +0200

To handle exceptions on Controller/Action level Exception Filters are used.

[TypeFilter(typeof(SampleExceptionFilter))]
public class ExceptionController : Controller

Exception Filters are implemented using IExceptionFilter with OnExcetion method override to handle exceptions.

Exception Filters can be applied to Controller and Action level.

Further read:

https://learn.microsoft.com/en-us/aspnet/core/mvc/controllers/filters?view=aspnetcore-7.0#exception-filters

https://learn.microsoft.com/en-us/dotnet/api/microsoft.aspnetcore.mvc.filters.iexceptionfilter?view=aspnetcore-7.0

ASP.NET Core global exception handling

Tue, 25 Oct 2022 15:49:53 +0200

To add global exception handling in ASP.NET Core just use configuration method:

app.UseExceptionHandler

https://learn.microsoft.com/en-us/aspnet/core/fundamentals/error-handling?view=aspnetcore-6.0#exception-handler-lambda

What ASP.NET Middleware can do?

Tue, 25 Oct 2022 15:45:12 +0200

Pretty much everything. Default ASP.NET Core request handling pipeline is consisting of bunch of middlewares:

Whole list of built-in middlewares can be found here:

https://learn.microsoft.com/en-us/aspnet/core/fundamentals/middleware/?view=aspnetcore-7.0#built-in-middleware

There is possibility to write custom middleware too.

Middleware in OWIN

Tue, 25 Oct 2022 15:40:45 +0200

There are 2 ways of writing OWIN middleware in ASP.NET Core.

Pure OWIN - usefull if you want to integrate with OWIN compatible module but not written in C#
ASP.NET Core OWIN middleware

ASP.NET Core Middleware:

This type of middleware can be:

delegate of Func<HttpContext, Func<Task>, Task>
class with IMiddleware imlementation

app.Use(async (context, next) =>
{
 // Do work that can write to the Response.
 await next.Invoke();
 // Do logging or other work that doesn't write to the Response.
});

Pure OWIN Middleware:

This rather would be used to integrate with OWIN middlewares written in different language, so that ASP.NET Core implementation directly is not possible.

Owin model and Kestrell request handling

Mon, 24 Oct 2022 15:23:02 +0200

OWIN - Open Web Interface for .NET
allows web apps to be decoupled from web servers
defines standard way for middleware to be used in a pipeline to handle requests and provide responses
ASP.NET Core applications can interoperate with OWIN-based applications, servers and middleware

Microsoft.AspNetCore.Owin package provides two adapter implementations:

ASP.NET Core to OWIN
OWIN to ASP.NET Core

This allows ASP.NET Core to be hosted on top of an OWIN compatible server/host or for other OWIN compatible components to be run on top of ASP.NET Core.

Expression Trees

Mon, 24 Oct 2022 15:08:43 +0200

’tree like’ data structure
each node is expression (method call, equality comparison, binary operation etc.) representing a piece of code
you can compile expression tree and run code it represent
enables dynamic code generation at runtime
LINQ queries on DBs are implemented with Expression Trees
used in the dynamic language runtime (DLR)
anonymous lambda expression can be converted to ExpressionTree via compiler (i.e. x => x*2 can be represented as tree instead of delegate or anonymous lambda function)

Further read:

Linq

Mon, 24 Oct 2022 15:05:47 +0200

consistent model for working with data from various kinds of sources and formats (XML, SQL, ADO.NET, Datasets, .NET collections etc)
3 stages of working with LINQ:
- obtain data source
- create the query
- execute the query
data sources implements IEnumerable<> and are called queryable types
IQueryable<> interface inherits from IEnumerable<>
generally it is better to use LINQ via IQueryable<> which produces deferred query instead of producing result immediately
queries can be executed multiple times
queries can be forced immediate execution via Count(), Max(), ToList() or ToArray() etc.
LINQ to DB libraries (like EntityFramework) usually are implemented via Expression Trees

Entity Framework: Change Tracking

Mon, 24 Oct 2022 15:02:09 +0200

tracked entities are saved during SaveChanges() call
entities being tracked when they are:
- returned from DB
- attached to context
- added as subentity
tracking can be disabled (AsNoTracking() i.e.)

Further read:

https://learn.microsoft.com/en-us/ef/core/change-tracking/

Entity Framework: eager and lazy loading

Mon, 24 Oct 2022 14:57:56 +0200

lazy loading:
- delaying load of data until we specifically request for it
- i.e. accessing customer.Address loads Address of a customer in the background via separate SQL query
- property needs to be public, virtual
- context.Configuration.ProxyCreationEnabled should be true
- context.Configuration.LazyLoadingEnabled should be true
- to disable lazy loading completely Context.Configuration.LazyLoadingEnabled = false;
eager loading
- opposite of lazy loading
- loads related entities as part of the query
- done via .Include() query

Further read:

Pattern matching

Mon, 24 Oct 2022 14:50:08 +0200

Pattern matching is used for:

testing expression to check if some value have certain characteristics
safer compare because ‘==’ and ‘!=’ can be overloaded

Examples:

a is int number

sequence is IList<T> list

switch statement:

command switch
{
 Operation.SystemTest => RunDiagnostics(),
 Operation.Start => StartSystem(),
 Operation.Stop => StopSystem(),
 Operation.Reset => ResetToReady(),
 _ => throw new ArgumentException("Invalid enum value for command", nameof(command)),
};

switch with ranges:

tempInFahrenheit switch
{
 (> 32) and (< 212) => "liquid",
 < 32 => "solid",
 > 212 => "gas",
 32 => "solid/liquid transition",
 212 => "liquid / gas transition",
};

collection pattern matching:

array is [1,1,2,3,5,8, ..]

Further read:

https://learn.microsoft.com/en-us/dotnet/csharp/fundamentals/functional/pattern-matching

Reflection

Wed, 19 Oct 2022 16:44:33 +0200

obtain information about types and members at runtime
create instances, invoke or access them
dynamic code generation

Further read:

https://learn.microsoft.com/en-us/dotnet/framework/reflection-and-codedom/reflection

Deployment models

Wed, 19 Oct 2022 16:42:01 +0200

Type	Command
framework-dependent executable for the current platform.	`dotnet publish`
framework-dependent executable for a specific platform.	`dotnet publish -r <RID> --self-contained false`
framework-dependent binary.	`dotnet publish`
self-contained executable.	`dotnet publish -r <RID>`

Further read:

https://learn.microsoft.com/en-us/dotnet/core/deploying/#publish-framework-dependent

.NET AOT

Wed, 19 Oct 2022 16:39:08 +0200

Ahead of the time compilation to machine code

produces self-contained application which:
- starts quicker
- uses less memory
- does not require .NET runtime installed
it is better for workloads with high deployment instances (clouds)
makes runtime code generation impossible
not all libraries are compatible

Further read:

https://learn.microsoft.com/en-us/dotnet/core/deploying/native-aot/

.NET Jit

Wed, 19 Oct 2022 16:36:28 +0200

Just - in - time compiler.

part of CLR
for all .net programs for all .net languages
converts MSIL to native machine code
compiles code when it is necessary at runtime
also it is possible to pre-compile to native code at compile time instead

Further read:

https://www.geeksforgeeks.org/what-is-just-in-time-jit-compiler-in-dot-net/

Records

Wed, 19 Oct 2022 16:33:51 +0200

can be class or struct
creates methods overrides for enforcing value equality instead of reference equality:
- Object.Equals
- Object.GetHashCode
- operators
  - ==
  - !=
implements IEquatable<> parameter
implements ToString
supports deconstruction
supports with expression
meant to be immutable

Positional records is a more concise syntax for record:

primary constructor parameters match the positional parameters of the record declaration
public init-only properties for each parameter

Further read:

https://learn.microsoft.com/en-us/dotnet/csharp/whats-new/tutorials/records#characteristics-of-records

Is it possible to block GC?

Wed, 19 Oct 2022 16:28:54 +0200

We can.

for some critical part of application code using GC.TryStartNoGCRegion method.
using CER (Constrained Execution Regions) (obsolete)
with GC LatencyModes

Further read:

https://learn.microsoft.com/en-us/dotnet/api/system.gc.trystartnogcregion?redirectedfrom=MSDN&view=net-7.0#overloads

https://learn.microsoft.com/en-us/dotnet/framework/performance/constrained-execution-regions?redirectedfrom=MSDN

https://learn.microsoft.com/en-us/dotnet/standard/garbage-collection/latency

Garbage Collector

Wed, 19 Oct 2022 16:26:09 +0200

automatic memory management
allocates and releases memory
prevents memory leaks and accessing dead objects
provides memory safety
virtual memory space
defragments/compacts memory
heap have 3 generations
- 0 - for new objects;
- 1 - for objects that lives through GC from 0
- 2 - for long living objects from 1 and 0
also Large Objects Heap (LOH) for big objects like arrays over 85,000 bytes

Further read:

https://learn.microsoft.com/en-us/dotnet/standard/garbage-collection/fundamentals?source=recommendations

Covariance and Contravariance

Wed, 19 Oct 2022 16:21:04 +0200

enables implicit reference conversions for
- arrays
- delegates
- generic type arguments
Covariance preserves assignment compatibility
Contravariance reverses assignment compatibility
allows for matching signatures of delegates and method groups
- to assign to delegates from method that returns more derived types - covariance
- to assign to delegates from method that returns less derived types - contravariance
Covariance and Contravariance are supported in Generics for implicit conversions for type parameters
- allows method to have more derived return type than defined by generic type parameter of the interface
- this allows implicit conversions from one generic interface to another (from IEnumerable<string> to IEnumerable<object> even if first does not inherits from second)
- allows generic types to have less derived argument types than specified by the generic parameter of the interface (implementation of IEqualityComparer<BaseClass> can be implicitly converted to IEqualityComparer<DerivedClass> because IEqualityComparer.Equals is contravariant)
- implicit conversions works for interfaces but not classes that implement from those interfaces (i.e. this won’t work List<Object> list = new List<String>())

Assignment compatibility is when object of a more derived type is assigned to an object of a less derived type.

.NET Generics

Wed, 19 Oct 2022 15:48:04 +0200

Allow for:

generalization of logic based on available types
specify possible supported types
re-usability of code
type safety
defers the exact specification of logic until the class or method is declared and instantiated by user code with all type parameters specified
mostly used for collections (some non-generic versions of collection are not recommended like ArrayList i.e.)
information of exact type parameters used in a generic can be obtained at runtime by reflection

Further read:

Data structures in .NET

Wed, 19 Oct 2022 14:52:50 +0200

In table below official recommendations for choosing data structure for your needs:

Generic	Non-Generic	Thread safe	Purpose
`Dictionary`	`Hashtable`	`ConcurrentDictionary`	key value pairs, lookup
`List`	`Array`	`ImmutableList`	access items by index
`Queue`	-	`ConcurrentQueue`	FIFO queue
`Stack`	-	`ConcurrentStack`	LIFO queue
`LinkedList`	-	-	access items sequentially
`ObservableCollection`	-	-	notfications after modification
`SortedList`		`ImmutableSortedSet`	auto sorted collection
`HashSet`		`ImmutableHashSet`	dictionary of just keys (high performance set operations)

Further read:

Strings in .NET

Wed, 19 Oct 2022 14:44:37 +0200

sequential collection of chars
immutable
max 2GB or about 1 billion characters
reference type
are interned - CLR saves memory by maintaining a table, called the “intern pool”, that contains a single reference to each unique literal string declared or created programmatically in your program.

Further read:

https://learn.microsoft.com/en-us/dotnet/api/system.string?view=net-7.0

https://learn.microsoft.com/en-us/dotnet/api/system.string.intern?view=net-6.0#remarks

What type can not be object field?

Wed, 19 Oct 2022 14:33:00 +0200

ref struct type can not be used as object field.

Further read:

`ref struct` types

https://learn.microsoft.com/en-us/dotnet/csharp/language-reference/builtin-types/ref-struct

`ref struct` types

Wed, 19 Oct 2022 14:28:48 +0200

added as part of C# 7.2.
defined with ref struct StructName
always on the stack
never part of heap
cannot be element of an array
cannot be field type of reference type
can’t implement interfaces
can’t be boxed
can’t be type argument
can’t be captured by closure of lambda or local function
can’t be used in async methods
can’t be used in iterators

Further read:

https://learn.microsoft.com/en-us/dotnet/csharp/language-reference/builtin-types/ref-struct

Value types as fields in objects

Wed, 19 Oct 2022 14:26:29 +0200

Value types are stored in the heap when they are part of the reference type.

Reference types vs Value types

Wed, 19 Oct 2022 14:17:17 +0200

Reference types:

passed via reference
reside on the heap
subject to GC
inherits from System.Object

Value types:

passed via value
reside on the stack, CPU registers or the heap
no subject to GC
inherits from System.ValueType

Further read:

https://learn.microsoft.com/en-us/dotnet/csharp/language-reference/keywords/reference-types

https://learn.microsoft.com/en-us/dotnet/csharp/language-reference/builtin-types/value-types

https://www.infoworld.com/article/3043992/a-deep-dive-value-and-reference-types-in-net.html

.NET Type System

Wed, 19 Oct 2022 13:59:44 +0200

shared type system across all .net languages
primitives data types library
5 kind of types:
- classes
- structures
- enumerations
- interfaces
- delegates

Further read:

https://learn.microsoft.com/en-us/dotnet/standard/base-types/common-type-system

ThreadPool

Wed, 19 Oct 2022 13:53:37 +0200

provides a pool of threads
can be used to execute tasks
post work items
process asynchronous I/O
process timers

Further read:

https://learn.microsoft.com/en-us/dotnet/api/system.threading.threadpool?view=net-6.0

Task Parallel Library (TPL)

Wed, 19 Oct 2022 13:51:20 +0200

simplify process of adding parallelism and concurrency to application
scales concurrency automatically
data parallelism
task based async programming
data flow

Further read:

https://learn.microsoft.com/en-us/dotnet/standard/parallel-programming/task-parallel-library-tpl

Locking with integer

Wed, 19 Oct 2022 13:44:02 +0200

In short:

it will not work because it requires boxing and boxing will create new object for reference in lock every time code is executed so lock statement will not prevent anything.

Further read:

https://intellitect.com/blog/locking-a-block-of-code-with-an-integer/

Locking in .NET

Wed, 19 Oct 2022 13:27:01 +0200

lock statement blocks execution of a code block for single thread that hold lock
the same thread can acquire and release lock again
any other thread is blocked from the statement
lock on dedicated instance only
using the same instance on multiple lock statements may lead to deadlock
avoid using as lock
- this, as it might be used by the callers as a lock
- Type instances, as those might be obtained by the typeof operator or reflection
- string instances, including string literals, as those might be interned
Hold a lock for as short time as possible to reduce lock contention.

Below code block is translated by compiler

Thread vs task in . NET

Wed, 19 Oct 2022 13:12:33 +0200

task is asynchronous operation
thread is abstraction on OS thread

Further read:

https://www.c-sharpcorner.com/article/task-and-thread-in-c-sharp/

What is SynchronizationContext in .NET

Wed, 19 Oct 2022 13:08:18 +0200

multi threaded programs usually uses one of the threads as means of synchronization
there are message loops to pass unit of works (like Windows message queue)
message queues require messages formats and convention for handling it
single thread to handle ASP.NET request is waste of the resources (before .NET 2)
thread pool was better but inconvenient to work with
instead SynchronizationContext queues work for given context not thread
thread have current context
context does not have to be unique but can be shared between multiple threads
thread can change context but this is rare
SynchronizationContext keeps count of all asynchronous operations
Windows Forms uses WindowsFormsSynchronizationContext
WPF and Silverlight uses DispatcherSynchronizationContext
Console app and WindowsServices and ThreadPool threads uses default SynchronizationContext
ASP.NET uses AspNetSynchronizationContext

Further read:

Does async/await creates new thread?

Wed, 19 Oct 2022 12:58:00 +0200

In short: no.

Longer explanation:

task that want to do I/O write call to driver write
driver operations are performed asynchronously
write operations on devices is done without CPU involved
finish of writing is done via interrupt
interrupt, schedules Deferred Procedure Call
DPC schedules Asynchronous Procedure Call
APC notifies thread that I/O write is done
task that was doing I/O is scheduled with continuation of its method after I/O is done.
some threads might need to be borrowed for handling APC and such but no thread is spawn to handle I/O

I recommend reading this post: https://blog.stephencleary.com/2013/11/there-is-no-thread.html . This will explain the topic much better than I can.

Avoid deadlock with asynchronous methods

Wed, 19 Oct 2022 12:52:45 +0200

avoid doing AsyncOperation.Result
parent method waits for Result while child method waits for parent thread to take control, after being done with async I/O
both waits
deadlock

Further read:

https://medium.com/rubrikkgroup/understanding-async-avoiding-deadlocks-e41f8f2c6f5d

https://stackoverflow.com/questions/15021304/an-async-await-example-that-causes-a-deadlock

Asynchronous programming in .NET

Wed, 19 Oct 2022 09:52:30 +0200

Asynchronous programming patterns in .NET:

Asynchronous Programming Model (BeginOperation/EndOperation)
Event-based Asynchronous Pattern (invoke event with result from async job)
Task-based Asynchronous Pattern (Task<>)

Ways of achieving asynchronous code in .NET

async, await
Task class API (Run, WhenAll, WhenAny, etc.)
ThreadPool.QueueUserWorkItem(ThreadProc);

Further read:

async/await in .NET

https://learn.microsoft.com/en-us/dotnet/standard/asynchronous-programming-patterns/

async/await in C#

Wed, 19 Oct 2022 09:51:34 +0200

async/await syntax should be used for I/O bound asynchronous operations only.

public async Task<int> GetUrlContentLengthAsync()
{
 var client = new HttpClient();

 Task<string> getStringTask =
 client.GetStringAsync("https://docs.microsoft.com/dotnet");

 DoIndependentWork();

 string contents = await getStringTask;

 return contents.Length;
}

Compiler transforms code into a state machine that to track yielding execution for await and resuming execution when a background job is finished.

Things to keep in mind:

async method needs await
add Async suffix to asynchronous methods
async void for ‘Fire and Forget’ work i.e. event Handlers
be aware of mixing LINQ with async lambdas
wait for tasks in a non-blocking manner (do not use Task.Result i.e.)
for tight loops use ValueTasks
Consider using ConfigureAwait(false) for performance and for avoiding deadlocks

Further read:

CPU bound vs I/O bound

Wed, 19 Oct 2022 09:35:37 +0200

There are two types of asynchronous operations in .NET.

I/O bound
CPU bound

For I/O bound async/await syntax should be used. TPL should not be used for this. async methods does not spawns new thread to hand le I/O work asynchronously.

For CPU bound operations TPL should be used (Task.Run). This spawns new thread. Performance testing should be done to check if overhead of new thread and context switching is not greater than actual work to be done.

SupressFinalize in IDisposable interface

Tue, 18 Oct 2022 14:24:06 +0200

informs GC that object does not need to be added to finalizer queue
finalizer execution is not deterministic and objects can live for long time in queue - this is optimization technique
to make sure that object disposes of any unamanaged resources even if Dispose will be not called we need to implement finalizer

Basically finalizer execution time is not deterministic. We cannot predict when this will be called and till that time - object is existing and still taking memory space. Since we are controlling creation and disposal of those resources we can say to GC: ‘Do not worry about that! It already had been taken care of!’. So this way object will be destroyed sooner, leaving this space to be used by something else possibly.

IDisposable interface pattern

Tue, 18 Oct 2022 14:16:21 +0200

Example implementation of IDisposable

public class MyResource : IDisposable
{
 public MyResource(IntPtr handle)
 {
 this.handle = handle;
 }

 public void Dispose()
 {
 Dispose(disposing: true);
 GC.SuppressFinalize(this);
 }

 protected virtual void Dispose(bool disposing)
 {
 if(!this.disposed)
 {
 if(disposing)
 {
 component.Dispose();
 }

 CloseHandle(handle);
 handle = IntPtr.Zero;
 disposed = true;
 }
 }

 ~MyResource()
 {
 Dispose(disposing: false);
 }
}

MyResource object have pointer which is unmanaged. This is why this class should implement IDisposable interface. Implementation should be correct according to recommended pattern (see below for documentation).

There a re few points to highlight here:

General Interview Questions

Tue, 18 Oct 2022 13:32:58 +0200

Closed
Open - to consider for yourself
- Interesting technical problem you solved lately
- Interesting technical problem you solved during your career
- What is the process of the MR/PR change review for you
- What is great success you accomplished lately? What skill did you improved lately?
- What is your greatest failure during your career? What was your biggest mistake?

What is DDD

Tue, 18 Oct 2022 13:29:02 +0200

software design approach
software should be driven by business domain
pieces of software should match the business domain
goals:
- domain is primary focus
- complex design should base on model of the domain
- devs and business should continuously work on refining this model to address changes in business
common language for defining a business model of the domain
complex logic should be abstracted to domain

Further read:

https://en.wikipedia.org/wiki/Domain-driven_design

What is BDD

Tue, 18 Oct 2022 13:24:32 +0200

based on TDD
better collaboration between devs, QAs and business than TDD
behavioral tests first, implementation second
behavioral tests should be easy to understand, simple
those tests should be written in specific language, possible to understood by business
combines idea of unit tests naming, acceptance tests standard agile framework (As a XXX I want YY to achieve ZZ), and acceptance criteria with Gherkin

Further read:

https://en.wikipedia.org/wiki/Behavior-driven_development

What is TDD

Tue, 18 Oct 2022 13:08:00 +0200

Test-driven development

software requirements are converted to tests cases *before* actual implementation
implementation is done against all those test cases
opposing to first writing an implementation then tests cases

Further read:

https://en.wikipedia.org/wiki/Test-driven_development

What is asynchronous programming?

Mon, 17 Oct 2022 16:25:05 +0200

software often waits for some I/O operation to complete (file operation, network call, database query etc.)
we can use this time for something productive instead to wait for data to be available or I/O operation to be completed
with asynchronous programming user can work in the meantime
not waiting for I/O operations synchronously enhances user experience

Of course this problem is much more complex. Modern software operates on much more different PC architecture than 20 years ago. Nowadays software must be asynchronous to appear more user friendly and to made better use of modern multi CPU architecture. Right now complex, professional software uses multiple processes, threads on multiple CPUs (CPU you buy is not really *single* CPU) or even utilize GPU for more complex computations (like i.e. video editing software). Before that we mostly had single threaded desktop applications that had ‘please wait’ or ’loading’ pop ups. Right now this is song of the past really and to achieve better User Interfaces that allow to edit tag of your post without changing whole post. With mobile devices on the rise and sometimes lousy cellular network those kind of designs would not be sensible or sometimes impossible to implement. Did you ever happened to reenter some chunk of data into some software because you lost it? Maybe network was down? Or you hit back button by mistake? Right now it belongs to past mostly. All of those would be hard to do with synchronous operations. Right now people are bit spoiled and they tend to leave the page/application if it appears to be slow and not every company can sell ‘one-in-the-market-necessary-for-your-work-software’ that you will be forced to use. And this is why we have asynchronous operations and this is why we have asynchronous programming model that allows us to program such operations.

Dynamic runtime delegates conversion

Sun, 07 May 2017 11:53:00 +0200

Introduction

Delegates can be converted easily to another type if you know types of those delegates. But if you do not know it is impossible without really complicated code.

Automatic scrolling to edited text box in Xamarin Android.

Sat, 25 Feb 2017 14:48:00 +0100

Important note: this should work well with newer version of Xamarin (2.3.3). If you cannot update for some reason this article will explain workaround.

Introduction

In previous article I explained how to detect show and hide events of software keyboard in Xamarin Android. I did it in my project because of the fact that some of text boxes was hidden by software keyboard that suppose to edit contents of those text boxes. In this article I will explain how to work around this issue in some older versions of Xamarin.

Detecting software keyboard events in Xamarin Android

Fri, 03 Feb 2017 23:31:00 +0100

Introduction

It is quite strange that you can’t easily detect if software keyboard is visible or not on Android. Because of that, it is also complicated thing to do in Xamarin.

Xamarin Android Services Pattern on GPS example

Mon, 30 Jan 2017 04:17:00 +0100

Introduction

In Xamarin if you want to use many Android native service/mechanism you have to implement it in Activity class. Or at least that is how Xamarin documentation wants you to do it.

Xamarin with TinyIoC

Sat, 10 Dec 2016 09:48:00 +0100

Introduction

This article is builds on topics and solutions introduces in ViewFactory for Xamarin application and Xamarin Master-Detail Page articles. It shows method for easier navigation and use of services via IoC in Xamarin application. Knowledge from previous examples is not necessary, but makes this article bit easier to understand.

Genereting proxies during runtime using Reflection.Emit

Fri, 07 Oct 2016 12:29:00 +0200

Reflection.Emit is very powerful tool. It creates IL code and since C# is converted into IL too, we have the same functionality as in C# and even more. It is very powerful and at the same time very complicated. Because of that it is worth to discuss how and for what it should be used. One idea is to create dynamic code with automatic implementations of interfaces - proxy types.

Faster than Reflection: Delegates. Part 3.

Fri, 02 Sep 2016 13:15:00 +0200

This is third article in series, that explains how to create delegates for public and non-public members of public, non-public or not known in compile-time types. If you are not familiar with previous articles and looking for detailed explanation of how delegates works, how they are created and why, I strongly encourage you to read first two. If you are just looking for a way to create delegates for methods or constructors it is not necessary. Code with new features and with bug fixes is available on GitHub and as Nuget package.

Faster than Reflection: Delegates. Part 2.

Tue, 16 Aug 2016 12:59:00 +0200

This is second article in series, that explains how to create delegates for public and non-public members of public, non-public or not known in compile-time types. If you are not familiar with previous article and looking for detailed explanation of how delegates works, how they are created and why, I strongly encourage you to read it first. If you are just looking for a way to create delegates for methods or constructors it is not necessary. Last article in series you can find here . Code with new features and with bug fixes is available on GitHub and as Nuget package. In previous article and first one in series, we covered:

Static
1. Properties
2. Fields
Instance
1. Properties
2. Fields
3. Indexers

Now it is time to cover following members.

Static
1. Methods
Instance
1. Methods
2. Constructors

Faster than Reflection: Delegates. Part 1.

Fri, 05 Aug 2016 09:58:00 +0200

This article is first part of the series of three. Next ones you can find here and here . Code for all three articles, with new features and bug fixes is available on GitHub and as Nuget package.

Why not Reflection?

If you working with .NET and C# (probably other languages too), at some point, you will have to write more advanced and generic code, that i.e. reads all properties of a type, searches for attributes, etc. Or maybe you want to call some private method or property? It is not that rare to use some internal mechanism of framework or library. Usually, the first solution involves Reflection. But the thing is: Reflection is slow. Of course, it is sufficient if you need to use Reflection rarely and it is not the main code of the application. If it is used more often at some point, you will probably try to optimize it. And now is the fun part: how to do it?

ViewFactory for Xamarin application

Sat, 09 Jul 2016 12:37:00 +0200

First things first - Why?

First of all I do not like how creating of views and navigation works in Xamarin. There is much boilerplate needed to just navigate from one page to another. Typically like in Xamarin documentation we have to obtain INavigation interface instance, create instance of a new page, create instance of view model and all dependencies it requires. When you page do not require view model things are real easy and Xamarin sample is just enough.

Xamarin Master-Detail Page

Sun, 24 Apr 2016 07:41:00 +0200

This article as previous ones (Frame With Padding , Frame Outline Problem , Inheritance from Implicit Styles ) also will be about Xamarin. This time we will inspect Master-Detail page control/mechanism. You can find official documentation for this control here . Quick spoiler: I never used this in any production application. Why? I will explain. Whole article will be to explain why this is just not worth your time. In this and subsequent articles I will introduce much better solution.

Xamarin Style based on implicit Style

Fri, 22 Apr 2016 23:17:00 +0200

Next article about bugs in Xamarin will touch topic of Styles. Styles for all of controls not only Frame (and if you wondering why Frame you should check first two articles about Xamarin here and here) . First of all if you are using Styles in Xamarin you probably know how to define implicit styles. If not here is a sample:

<Application.Resources>
 <ResourceDictionary>
 <Style TargetType="Label">
 <Setter Property="FontSize" Value="45" />
 </Style>
 </ResourceDictionary>
</Application.Resources>

Xamarin Frame and disappearing Outline

Fri, 22 Apr 2016 13:00:00 +0200

This is another article in mini series of Xamarin bugs and workarounds for them :) Previous one you can find here . This one is again about Frame control. I promise that another one will be about something else B) Issue can be observed when you bind BackgroundColor property of Frame control and something will trigger change of view model source property binded to background color.

Xamarin Frame and Style with Padding

Fri, 22 Apr 2016 05:10:00 +0200

If you are a bit advanced with Xamarin you should already now that there is possibility of styling your controls almost like in WPF. Almost, because there is a few strange bugs involving this. For example there is a problem with styling a Padding property in Frame. Consider following Xaml of Xamarin application (I tested it in Android):

Safe cast string to enum

Tue, 25 Feb 2014 23:49:00 +0100

In on of my projects I had make an interface between database and web service in C# code. One of problem I had to face was need to cast strings to enumerable types, because of simple fact that database have no idea what is an ’enum’. Yes simplest mapping between an enumeration and database type is integer. It is simplest but in my idea not best. For example I really do not remember (or do not WANT to remember) what is the meaning of 4 integer value in case of package shipment status.

Visual Studio code snippets

Tue, 25 Feb 2014 11:39:00 +0100

What I really like in Visual Studio as development environment is code snippets. Well actually probably other IDE have something very similar but I do not have much of a choice in that matter as C# developer :) Anyway this feature is really useful. For example while refactoring I very often use if surround snippet.

Smart string builder

Sat, 22 Feb 2014 03:16:00 +0100

In this rare times when I was writing code that suppose to create large string from smaller chunks and with some non string parameters I was using StringBuilder class of course. Of course because string adding, or string concatenation is very costly when talking about memory utilization. It is because every time you do this:

var text = "Hello world!"+ "n"+ "How are you?";

new string is created for every ‘+’ operation in memory. Not a best way of doing strings creations. StringBuilder is better because it do not creates strings until you call .ToString method of StringBuilder class.

Expression parsing and nested properties

Sat, 22 Feb 2014 03:00:00 +0100

In one of projects I was working on, I needed to get property value from property path and property path from expression. First, lets cover second case. With expression in form of nested property value:

()=>object1.object2.object3.object4

we cannot take simple value of some property, because we have only root object, object1.

WebSocket libraries comparison

Fri, 21 Feb 2014 14:38:00 +0100

Web project often requires to push data to clients as fast as possible, whenever it is necessary without waiting for client request. It is perfect for website with real time communication between users, like with online communicators for example. Or document collaboration tools. Or maybe system status updates on long running calculation/tasks performed by server. In every case two way communication mechanism is ideal.

Ninject and WCF

Sun, 26 May 2013 12:01:00 +0200

Ninject is very simple and in the same time powerful IoC container. I used it in few project and had very little or none problems. Most of the time getting instances from Ninject kernel requires simple line with binding expression, or even this is sometime unnecessary if type is self bind able (i.e. if it is concrete class with parameterless constructor). Little harder is getting Ninject to work with WCF. You cannot just bind interfaces types because proxies which implements them are created through .NET mechanism. Luckily WCF system is very flexible and mostly can be changed/extended with custom functionality.

Mapping collection of entities in EF with AutoMapper

Sat, 25 May 2013 11:38:00 +0200

In my last post I explained why is useful to add base entity class in EF. Today I will write how with use of this base class an AutoMapped map collection of data objects (i.e. DTOs to existing collection of entities).

Entity Framework and Base Entity class

Thu, 16 May 2013 14:32:00 +0200

Entity Framework does great job with taking care of entities changes and entities collection internally. And its uses almost plain POCO objects. Almost because collections of dependents data like in 1 to many tables relationship needs virtual keyword. Its understandable since EF needs to track what happens to collection. For instance lets have 2 tables: Customer and dependent Order.

Ninject and Entity Framework

Fri, 03 May 2013 03:50:00 +0200

Last time I wrote about problem with testing Entity Framework in Visual Studio 2012. During working on this project I also encounter problems with configuration of Ninject to allow injecting Entity Framework classes from its Patterns extension. To be frankly this was not much a problem then lack of documentation. I just was forced to experiment with Ninject and new-thing-for-me Entity Framework Patterns extension.

Entity Framework Patterns extension and Unit Test in Visual Studio 2012

Tue, 23 Apr 2013 02:18:00 +0200

Yesterday I was trying to make work simple test class in VS2012. I did not know what was wrong exactly. It was first project when I tried to glue together Entity Framework with repository pattern and Ninject as IoC container. Because of the fact that this was first time for solution like that I was sure that it is me, or rather source of my problem was indeed my code. But let’s not skip ahead.

Line in WebGL and why you gonna do this on your own.

Tue, 26 Feb 2013 02:15:00 +0100

WebGL as method for generating 3D graphic in browser and it is pretty powerful. You can do really awesome things with this, like porting Quake 2 to JavaScript or creating model of human body . But all of this 3D object are complex collection of hundreds or thousands triangles. What if you just want to draw a simple thin line? Luckily OpenES have support for that. So it is possible. Lets see how to do that.

Cartezian coordinate system in HTML 5 canvas

Sat, 23 Feb 2013 14:35:00 +0100

Normally canvas 2D context is using custom coordinate transform where Y coordinates are flipped vertically. Here is image with example (taken from http://diveintohtml5.info/canvas.html ; it is great book and you should read it :) ) It is a bit inconvenient to use if you are drawing some shapes with points given in classic Cartezian coordinate system.

Simple 2D graphs

Sun, 17 Feb 2013 02:26:00 +0100

While writing application for my Masters Degree diploma I was required to came up with simple library for creating simple 2D graphs in the browser. Today I decided to share it with the rest of the world :)

Curious Case of Visual Studio project compatibility.

Tue, 12 Feb 2013 08:06:00 +0100

Recently I was asked at work for quick review of small project. It was created (or updated) in newest version of Visual Studio - 2012. I don’t have installed VS 2012. I could install it, yes, but it would be just waste of time since browsing through code in that project took me only few hours. So I decided to try to open that in VS 2010.

Why I don't like Python anymore.

Thu, 07 Feb 2013 06:45:00 +0100

It was few years back on my first year of studies, when my friend told me about Python. Then it was great for me. One-liners could read ftp folders, web pages, interact with operation system, list comprehension, injecting code into life programs, nested functions, procedural as object oriented programming styles etc. etc. All of this in builtin libraries with simple but yet quite functional editor, supported with great documentation. I wrote my first studies project in Python and I was proud with it a little ;) During next years I was using Python for writing simple scripts to automate my professional work as PHP programmer. Python was and I still think is better language then PHP. After that I started to work as .NET/C# developer. It has different syntax but C similarities let me to learn it quite quick. Framework .NET is powerful tool also so I didn’t touch Python for 2 or 3 years at least for anything longer then few lines.

Using Matlab from C# application.

Sun, 20 Jan 2013 11:25:00 +0100

While writing an application for my Masters degree diploma, I wrote a simple application that was using COM Matlab server. I have found it hard to use it mainly due to lack of documentation, which is really basic with only few code examples for C#. I guess writing programs that use Matlab for calculating is not encouraged by MathWorks, you would became competition that way :) Nonetheless, I accomplished why I was required to do, so I decided to share this with the rest of the world.

Server controls in separete assembly part 2

Mon, 24 Dec 2012 05:14:00 +0100

Yesterday I was writing about creating server controls in separate assembly . Today I will cover more complicated example then simple “Hello World!” control. My goal was to create text box that will take handler method to run when validation event will be triggered, and when validation fail will show appropriate message. Text box with builtin validation would be nice, since you always should validate user input , right? But lets start creating our control.

Server controls in separate assembly

Sat, 22 Dec 2012 14:52:00 +0100

In my project I wanted to create text box control with builtin configurable validation. ASP.NET definition in .aspx or .ascx file would get validation function by name. But I do not wanted to create just .ascx file with text box. I could not then reuse that control in other projects and it certainly would be nice, right? Creating such server control from code is not hard, but maintainability of HTML code in C# (it cannot be done other way) is not good. It’s pain really. Why Microsoft would create way to automatically create C# code from .ascx, .aspx files is mystery for me.

Handling Ajax errors in ASP.NET

Sat, 22 Dec 2012 09:58:00 +0100

Yesterday I wrote about issue with Ajax postbacks in application I am working currently. Problem was with buttons that was triggering UpdatePanels to update asynchronously. After adding loading indicator I wanted to add error handling at client side.

Global UpdatePanel loading indicator in ASP.NET

Fri, 21 Dec 2012 10:48:00 +0100

When I started working on application I am working right now, one of first things I noticed, that was not entirely wrong but was inconvenient: when user clicked on some button that triggers update panel you was not entirely sure it worked or not. Why? Because of three things:

Better algorithm for finding UpdatePanel that will be updated during page request.

Thu, 13 Dec 2012 15:50:00 +0100

Other day I was experimenting with finding which UpdatePanel will be updated during page post back. I found out that my solution was not working correctly for complex triggers, like user controls with custom events, which will trigger async postbacks and update of panel. Or for grids with buttons. It was big issue. Also it was not pointing to correct UpdatePanel if name of control was not entirely unique (for example when you have multiple same custom controls on page). It would return first panel, even if it wasn’t updating.

Display message on client from anywhere in code with ASP.NET

Mon, 10 Dec 2012 09:20:00 +0100

When I was working with WinForms application in previous work I have found one of most useful feature, ability to show some message to the client. For example user cannot edit some entity in particular time (it is being edited by someone else) or should change password because it will expire in x days. I wanted to do something similar in my web application.

Getting private field or property from object

Sun, 09 Dec 2012 07:26:00 +0100

Yesterday I was playing with finding all UpdatePanel controls on page during generating response in ASP.NET. Since object with this information is private itself and collection of panels is private in that object, I had to write some code to extract that data. I ended up with to extension methods for Object type, for getting private field an private property. It’s pretty straightforward, besides one thing: since member is private. And we don’t know in witch type exactly (every type in base/descendant hierarchy have it’s one private members) we have to search them all. So this is final code: