Andraž Bajt's blog

How I use AI to help me prepare talks

Wed, 10 Dec 2025 00:00:00 +0000

I’ve recently changed my workflow for preparing talks. I prepared 3 with this approach so far and I’m quite happy with the process so I decided to document it.

Inspired by Gary

I love Gary Bernhard’s How to prepare a talk. But it’s a lot!

tl;dr: improvise a bunch of times and the structure will crystallize. As a bonus you’ve now practiced a lot and are well prepared now.

I’ve tried this but it’s hard to find time (well, motivation really) to spend that much time on initial ad-lib runs. So I tried my own twist.

GraphQL ❤️ SQLite

Sat, 06 May 2023 13:20:31 +0200

If you’ve done anything nontrivial with GraphQL you’re probably familiar with how “N+1 select problem” sneaks up on you. If not, this is how gqlgen docs explain it:

Imagine your graph has query that lists todos…

1

query { todos { user { name } } }

and the todo.user resolver reads the User from a database…

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17


func (r *todoResolver) User(ctx context.Context, obj *model.Todo) (*model.User, error) {
	res := db.LogAndQuery(
		r.Conn,
		"SELECT id, name FROM users WHERE id = ?",
		obj.UserID,
	)
	defer res.Close()

	if !res.Next() {
		return nil, nil
	}
	var user model.User
	if err := res.Scan(&user.ID, &user.Name); err != nil {
		panic(err)
	}
	return &user, nil
}

The query executor will call the Query.Todos resolver which does a select * from todo and returns N todos. If the nested User is selected, the above UserRaw resolver will run a separate query for each user, resulting in N+1 database queries. e.g.

(Re)Learning AWK with ChatGPT

Fri, 31 Mar 2023 16:12:00 +0200

awk is the Swiss army knife of the command line. If you have a text-shaped problem then awk can probably be a solution. Here’s a fabulous talk by Brian Kernighan (the k in awk) on why this may be the case.

But it’s also a tool that I use just rarely enough that I keep forgetting details. So everytime I reach for awk it’s a painful experience of not remembering and having to cobble together 3 different StackOverflow answers to get something resembling the script I want. Only to then realize I’ve been “holding it wrong” and there’s a better approach 🤦.

Hypermedia driven micro frontends

Wed, 29 Jun 2022 00:00:00 +0000

Micro frontends? Huh?

extending the microservice idea to frontend development – https://micro-frontends.org/

or: how to scale frontend development once you have too many engineers and they start stepping on each other’s toes. Assuming of course your frontend is an SPA (grug might have something to say about both points).

But of course splitting an SPA into multiple smaller applications and having it all play together brings along a ton of additional complexity. Just like splitting a monolith into micro services 😉.

How I fell in love with low-js

Fri, 28 Jan 2022 00:00:00 +0000

About 5 years ago I was working on a project written in style still common for today:

python backend serving a REST API (public btw)
SPA using framework-of-the day

And it was horrible. Not to mention that the framework-of-the-day did not actually take off and we ended up using something very fringe - there were real performance and complexity problems. And to this day I believe all our problems stemmed from the base architecture. There were just too many layers, each contributing overhead. Both in execution speed (aka “app is slow”) and development speed (“our velocity is too low”).

Tailwind CSS vs Pico CSS

Tue, 18 Jan 2022 00:00:00 +0000

tl;dr it’s a tradeoff so it depends.

With that out of the ways let’s start with the fact that this is not an article about technical merits and which is best. It about my experience and my beliefs about which kinds of projects are a good fit for which tool.

So what did I do? I have a (useless) pet project I keep fiddling with to try out new things. At one point I decided my eyes hurt and I want to make the frontend a bit prettier. It is paramount to explain that I have indeed been living under a rock for a few years and haven’t really touched any CSS. So I first did some exploration to see what has the world moved on to.

Speed is a feature; thoughts on modern web performance

Wed, 29 Dec 2021 00:00:00 +0000

I’m pretty sure that as computers got faster they got slower. As chips got faster the software slowed down. I’m acutely aware that we do much more now but just the user interfaces really have no reason for slowing down.

Why do I care? Because a fast web is a delight to browse. A fast app is a delight to use. The tool feels like an extension of your body. Where as a sluggish tool get’s in the way. Feels like you’re battling it instead of using it. This is one of the main reasons I still prefer vim over more “modern” IDEs.

Trying Out Remix

Sun, 19 Dec 2021 00:00:00 +0000

I came across Supporting Remix with full stack Cloudflare Pages on Cloudflare blog (yes I’m a CF fan) and Remix piqued my interest. Mostly because it addresses my concerns with “modern” web development: the wiring boilerplate is mostly gone, there is no 100 file setup to get a workable dev environment. And of course because it’s outside my comfort zone - so maybe I’ll learn something.

Getting started

There’s a code example right on the home page but I’m having hard time building a mental model from just that. Luckily there is also a bit Get started button which takes me to a tutorial. Great, let’s follow that.

Running amd64 docker images with Podman on Apple Silicon (M1)

Sun, 12 Sep 2021 00:00:00 +0000

tl;dr it works, scroll though and look for the code snippets

You might have heard about Docker (the company) changing the terms of use which effectively requires enterprise users to pay up. While I personally think that paying for critical infrastructure that supports your business is a good idea there was a considerable backlash among engineers and “alternative to docker” was the common them on HN front page for a few days.

Trying out Prometheus

Sun, 27 Jun 2021 00:00:00 +0000

This time I’m trying out a monitoring solution: Prometheus.

All I know about it so far:

pull based, server scrapes /metrics on workloads, human readable format
scales well because of this
popular in kuberetes space

Let’s try running the docker iamge based on the README.

1

docker run --rm -it -p9090:9090 prom/prometheus:main

We’ve got lift off! And it’s monitoring itself right out the gate:

Trying out Microk8s

Thu, 15 Apr 2021 00:00:00 +0000

In another “running kuberets” flavoured post I’ll be trying out microk8s

High availability K8s. Low-ops, minimal production Kubernetes, for devs, cloud, clusters, workstations, Edge and IoT.

In the past I’ve been using k3s for small (mostly single node) “clusters”. The pitch is very similar. The few differences I can point out now are:

microk8s is a bit heavier - I don’t think it will run on 1G vps, let’s try it out
microk8s has a better HA story - definately want to try this out
k3s comes as a single “run everywhere” binary whereas microk8s requires snap

Setting up

I’ll be using Hetzner Cloud to provision a few small VMs for my laboratory. But I’ll start with a single one.

Trying out Rancher

Wed, 14 Apr 2021 00:00:00 +0000

This time it’s a bit different: this post has very little code and a lot of screenshots as I was primarily working with a GUI.

Reading about (and working with) K3s is almost impossible to not hear about Rancher - the company behind K3s. But what had me slightly confused is their primary product: Rancher (the software).

The web page says

Managed Kubernetes Cluster Operations

but that doesn’t really tell me much. Neither do their presentations. So what better way to get a feel for it than taking it for a spin :) (there is some enterprise offerings but the base product is free).

Trying out Okteto Cloud

Sat, 27 Feb 2021 00:00:00 +0000

This is a part two of Trying out Okteto. I’ll be trying out okteto.com - “The Kubernetes development platform”.

At a first glance it looks like a hosted kubernetes solution + the CLI tool I already tested + some bells and whistles. Let’s find out what those are.

First impressions

Looks like I got a managed namespace with 8G ram, 4cpus and 5G storage. Quite generous for a free plan.

Trying out Okteto (CLI)

Mon, 22 Feb 2021 00:00:00 +0000

Today I’ll be giving Okteto a spin. From the docs

Okteto: Tool to Develop Applications on Kubernetes

Or my personal pitch:

Okteto delivers on the “my other computer is the cloud” trope

What does this mean? First of all: there is a distinction between

Okteto Cloud & Okteto Enterprise: this is a server side solution that provides managed k8s namespaces, CD and other bells and whistles - will look into this in the future
okteto - the tool: open source client to work with Okteto Cloud & Okteto Enterprise - but the core functionality works with any k8s cluster. This is the tool I’m trying out today.
Okteto - the company (behind all of this)

How it works

From the docs:

Trying out Buffalo

Sun, 07 Feb 2021 00:00:00 +0000

I’m trying out a new concept: my “I should check this out” list of tools and technologies has grown quite a bit so I got the idea that I could do short write ups of my experimentation and first impressions to give myself motivation to try these things out. Will see how this turns out, hopefully this will turn out a series :)

Today I’ll be trying out Buffalo - a Go web framework. I believe this one distinguishes itself by focusing on developer productivity - a kind of “Rails for Go”. But apparently it’s mostly some glue and tooling around a bunch of other established libraries. I very much like this approach. And yes, I agree that “just use standard library” is not (always) the right approach. Sometimes you just want to bootstrap a small project quickly.

Attaching your local machine to a Kubernetes cluster

Sat, 17 Oct 2020 00:00:00 +0000

Why on earth would you want to do that?! You may have a legitimate reason like pooling spare compute capacity. Oooooor you’re just goofing around in order to understand things. Like me.

Specifically I was playing with telepresence (a “VPN into the cluster”) and a had an idea: why not just run a part of the cluster locally. Then you can interact with the containers and containers can interact with your host (host port, host path etc), no magic required.

Surprising ease of gaming on Linux in 2020

Sun, 07 Jun 2020 00:00:00 +0000

During the quarantine my hobbies (and I imagine many others’) shifted to more indoors ones - possibly sedentary. Among other things I also started gaming after a hiatus of many years.

But as you can probably imagine gaming on a 13" laptop with integrated graphics is not the most gratifying of experiences. So I got myself a desktop with a proper graphics card.

And now comes the interesting part: since all the games I originally planned on playing have Linux support I decided to skip buying a copy of Windows and copy my laptop NixOS config and extend it as needed. Basically I wanted to do some tinkering again.

Rename rebase with git

Mon, 20 Apr 2020 00:00:00 +0000

Want to perform a git rebase in which you rename a file without pesky conflicts?

tl;dr

1
2


git filter-branch -f --tree-filter "git mv ORIGINAL_FILE_NAME NEW_FILE_NAME || true" -- $(git merge-base origin/master HEAD)..HEAD
git rebase origin/master

Ever wanted to rename a file you created/modified in a pull request but also wanted to keep the pristine history you worked hard for? Well tough luck you can either push a new commit to rename it (and keep old commits with the old file name) or resolve a bunch of pointless conflicts. There must be a better way…

Moving to Hugo

Wed, 15 Jan 2020 00:00:00 +0000

If you’ve been here before you might notice the site looks different. Indeed I moved my content to yet another platform (first it was blogspot -> hakyll).

I wanted to publish something but I ran into issues: my toolchain did not work. See, I’ve been doing less Haskell on my machine lately and negligence led into my setup rotting (as I did upgrade my machine quite a bit). Truth be told the infrastructure was not designed to last the test of time. Anyways after a few rounds of debugging I once decided to finally give up my complicated custom setup and move to something off-the-shelf so I can focus on writing.

Linear time sorting

Fri, 25 May 2018 00:00:00 +0000

Did you know that you can sort data in time linear with respect to the lenght of said data? Some people dismiss linear time sorts because they’ve learned that n log n is the bottom bound for sorting an arbitrary input.

But n log n is optimal!?

Let’s even sketch out a proof. Any function that sorts an array of elements will in fact figure out a permutation of elements into a sorted order. Even if your algorithm does not work in terms of permutations it can be viewed as a black box that computes a permutation.

Integrating Jira into Google Sheets

Thu, 25 Jan 2018 00:00:00 +0000

So I got a (recurring) task to compile some stats from Jira. You now, some filtering and agregation. Possibly a pivot table and a chart. Sounds like a perfect task for spreadsheets. But Jira doesn’t have spreadsheets - at least without addons, but that is not an options for me.

So I decided to use Google Sheets. And figure out a way to automatically fetch data from Jira because there’s no way in hell I’m doing that manually every time. Turns out there are some scripts you are supposed to paste into sheets and give them all the permissions. But I don’t trust that, not without properly reviewing the thousands of lines of code.

NixOS install & labor pains

Wed, 24 Jan 2018 00:00:00 +0000

The Back Story

My first foray into world of Linux happened with Red Hat Linux 6 (original, not the enterprise one). It was magical but it didn’t stick. See, I was a kid who liked video games. So it was back to Windows until I finally got serious about programming and discovered I the development environment on Linux. So I formatted my hard drive and installed Ubuntu. I quickly learned absorbed information and slowly got bored. When Unity came out it was time to switch. I dabbled a bit with Mint but ultimately landed on ArchLinux. I really like the RTFM or GTFO philosophy of Arch because it means you learn stuff and you learn it quickly. Because you just have to. Oh, and there is great documentation. And everything was great. Until I borked my system doing an upgrade or some other system package operation. A few times. Btrfs to the rescue! But there must be a better way. (I was doing manual snapshots of the root filesystem before any potentially destructive operation.

Implementing apply on tuples in Scala

Fri, 01 Sep 2017 00:00:00 +0000

One of the first things you learn as a newcomer to Scala is the difference between a list and a tuple: a list must be homogeneous but a tuple can be heterogeneous. That is all elements of a list must have the same types but a tuple can contain things of different types.

A direct consequence of this is that a list can define a by-index accessor and a tuple cannot. You can use list(3) but you can’t do tuple(3) - you need to do tuple._4 (and there is that pesky off-by-one).

Dockerless Services (with Nix)

Wed, 30 Aug 2017 00:00:00 +0000

So you want an “isolated” MySQL “service” but don’t want to (or are not able to) run Docker/rkt/LXC/whatever?

Here are the tl;dr steps:

1
2
3
4
5
6
7
8
9


cd /tmp
mkdir mysql-1
cd mysql-1
nix-shell -p mysql
mysql_install_db --datadir=$(pwd)
mysqld --datadir=$(pwd) --socket=$(pwd)/socket &
mysqladmin -uroot -h127.0.0.1 password <REDACTED>
mysql -uroot -p<REDACTED> -h127.0.0.1
...

Why?

My motivation for this was running a database on a machine with a specially tuned kernel that did not support containers.

TDD-ing a toy sized project

Wed, 21 Sep 2016 00:00:00 +0000

Just recently I was porting a toy sized parser combinator library (a proof of concept) from Haskell to Python. You know, for educational purposes. It turns out I’m not smart enough to keep the complicated types (and explicit laziness) in my head even for such a small project. So my solution was to do TDD. To clarify: I wanted to test happy paths through my functions to make sure at least types fit together.

Repairing a corrupt Git repo using a clone

Wed, 24 Feb 2016 00:00:00 +0000

Quite recently I managed to make myself a corrupt git repository due to a file system failure. See, git stores everything in content addressable blobs - the file name of something is it’s hash. Which lends itself nicely to checking repository integrity - it keeps out malicious attackers as well as my file system problems.

I already hear you saying: Why not just make a new clone, git is distributed anyway? Well, I wasn’t diligent enough to push everything. I had local commits that were quite important, so I spent some time fixing it.

Lazy unary numbers

Sun, 03 May 2015 00:00:00 +0000

We are used to encoding numbers on computers in binary. Binary is the “simplest” base that yields logarithmic length though it may not be optimal. But can we do simpler? How about unary?

Unary is often used with Turing machines where we don’t care for efficiency and I will assume this same stance. Let’s forget about efficiency and explore what can unary numbers do that binary can’t. Specifically lazy unary numbers as otherwise the systems are equivalent. I’ll be using Haskell as it is lazy by default and thus a good fit.

Cheap tagged types in Scala

Sat, 02 May 2015 00:00:00 +0000

Sometimes you want to distinguish between different types that have the same underlying representation. For example both UserId and ProductId could be represented by Long. The usual solution is to introduce wrappers in order to make the distinction safe.

1
2


case class UserId(id: Long)
case class ProductId(id: Long)

But this introduces runtime overhead of boxing and unboxing over and over which may add up in some cases. Luckily Scala 2.10 introduced value classes. We can ensure no runtime overhead by extending AnyVal (this can only be done with classes with one field).

Approaches to designing a Haskell API

Sun, 05 Oct 2014 00:00:00 +0000

Recently I’ve been thinking about the design of programming interfaces, especially in Haskell. But don’t let the title misguide you; this is not supposed to be a tutorial or a guide but simply an showcase of different styles. Feel free to tell me I’m wrong or missed something.

The problem

Let’s say we are writing an interface to RESTful web service. Our goal is to create type safe functions and descriptive models but all in all easy to use.

Static linking with GHC on ArchLinux

Sat, 03 May 2014 00:00:00 +0000

There are many reasons why to prefer dynamic linking to static but I’ll not go through them. Sometimes you just want static linking, period. In my case it was to show that Go’s static executables without dependencies are not something special and other languages can do it as good as well - Haskell included. My compiler of choice is GHC and I’m running ArchLinux. More on why this is important later.

Coroutines for free

Mon, 14 Apr 2014 00:00:00 +0000

Motivation

My first run-in with coroutines was with Python’s generators.

1
2
3
4
5


def ints():
 i = 0
 while True:
 yield i
 i += 1

This is a function that never returns. Instead it runs in an infinite loop over all natural numbers. But it yields every number. This means it actually stops and “returns” the number to the caller but is available for more execution. Kind-of like implementing an Iterator but the control flow is reversed. Now you don’t produce a value on demand but rather you run and output values at will while the runtime system stops and resumes you.

Practical Future[Option[A]] in Scala

Fri, 07 Mar 2014 00:00:00 +0000

Motivation

In real world concurrent code you often come across the Future[Option[A]] type(where A is usually some concrete type). And then you need to compose these things. This is not straightforward in Scala.

If you’ve done some Haskell just import scalaz._ and you can skip the rest of this article. Scalaz library defines a monad typeclass(and many others) that formally specifies what it means to be a monad(not “has a flatMap-ish thingy”). Then it’s easy to build abstractions upon this.

Comparing Haskell Web Frameworks

Sun, 23 Feb 2014 00:00:00 +0000

Intro

Lately I got sucked into Haskell. When I first saw it it looked like it might be a bit impractical for real-life projects but this prejudice faded away slowly. Now I’m at the point where I have an idea for a web application and I’d like to do it in Haskell. After a quick search I found many frameworks and libraries that I might use. So I decided to do some exploration and implement a bunch of stuff with different technologies. This way I don’t have to believe others and I can decide for myself what feels right to me.

Product and sum types (Go and Haskell)

Mon, 17 Feb 2014 00:00:00 +0000

Let’s take a look at Go-language functions. They let you return more than one result. Like in

1
2
3


func foo(n int) (int, int) {
 return n+1, n+2
}

This function foo returns two numbers. But does it really? Pairs(and tuples in general) aren’t a first class citizen in Go. Yes you can return them from function and you can even return a result of a function that returns the same signature as you but then you must immediately bind each result to separate variable.

Hakyll

Wed, 08 Jan 2014 00:00:00 +0000

Hello there. I think I’m gonna resurrect this blog as I feel like writing again. I kinda have a general dislike for PHP and don’t like Wordpress. I tried mucking about with Github Pages, but I’m not a fan of Jekyll either. And then I forgot about this.

Haskell

A few weeks ago I rediscovered Haskell. I tried learning it once before but gave up. And insisted in Scala. This time I believe I got past the hardest part of the learning curve. And it’s been awesome! I learned so much, even bought and start reading a book on abstract algebra. But most importantly, I’ve written some software. Nothing fancy. Some Project Euler problems and some homeworks. But I got a feel for the language and I love it.

Virtual machine in C++

Fri, 18 Oct 2013 00:00:00 +0000

This is not a tutorial. This post is a flashback I had today. It might be a bit fiction as my memory about events tends to be fuzzy at times. But I promise it at least resembles the real story. I was in elementary school and just found out about programming and was learning about c++. After reading “C++ na kolenih” by Goran Bervar I was empowered by knowledge and tried to do all sorts of projects. Mostly in console. Stuff like subtitle format converter - NIH syndrome. I was a bit frustrated because I couldn’t find any books about windows programming in the library. Yes, library was may primary source of information, because my English was not nearly good enough for technical stuff. I might add here I worked on Windows 98(and later XP) with DevC++. I found out about Visual Studio in a few years and did some Windows development. I digressed a bit. Then came the most optimistic idea. A virtual machine. Something quite high level(instruction to print) an eventually an assembler. I now realize I was always into language stuff. So a designed a machine language with just enough instructions to do Hello World, that is PRINT and END.

Setting up for better Scala development on Android

Wed, 04 Sep 2013 00:00:00 +0000

I did a tutorial how to set up everything using IntelliJ a while ago. I still think IntelliJ IDEA is awesome and you should use it(it has a free and open version) but I’ve found a better way. No more clicking around in wizards…it’s config file time. Relax - it’s quite simple. I recommend installing the Typesafe-stack. It gives you the two tools listed below in nice packaged form with updates. And that’s about it. I’m quite fond of package managers => one place to update everything.

750 words