AB
@alexbunardzic.bsky.social
110 followers 34 following 900 posts
https://alexbunardzic.com/ Looking for the communities of software developers, guitar players, visual artists, audiophile vinyl collectors, hi fi and turntable gear, French baguette makers. Bason tuner by day.
Posts Media Videos Starter Packs
Reposted by AB
ted.dev
I talk about this as “planning [designing] is important, but following it precisely even after we get new information (learn), is not the goal.” Or, it’s all about Continual Planning [Designing].
alexbunardzic.bsky.social
Can someone explain to me how is spec-driven development not waterfall?
alexbunardzic.bsky.social
It is very easy to use any tool in a harmful way. Only a shoddy workman blames the tools.
alexbunardzic.bsky.social
I have just received an interesting feedback on my talk at Devoxx Belgium 2025.
alexbunardzic.bsky.social
Everyone is unique, except me.
alexbunardzic.bsky.social
Keep twisting and turning, embracing the change, and focus all of our attention and effort on adaptability.
alexbunardzic.bsky.social
We live in a dynamic world, and any obsession with static, finite product, is a death wish.

Same applies to plans and planning. Every plan is a throwaway plan. The true value lies in our ability to keep planning.
alexbunardzic.bsky.social
For less experienced people, design is something that is precious, it is won by hard work and as such is to be cherished. But for experienced people, every design is a throwaway design. Experienced designers focus on the process, rather than focusing on the outcome.
alexbunardzic.bsky.social
It seems hard to intuitively grasp the adage "plans are useless, but planning is indispensable".

Maybe it would be easier to grasp the meaning of that adage if we were to reformulate it to read: "designs are useless, but designing is indispensable".
alexbunardzic.bsky.social
The most important aspect of a test is its fallibility. Without that, we would have no protection against regression.
alexbunardzic.bsky.social
Tests are only tests if they are fallible. If we have a test that asserts that 2 + 2 = 4, that's not a fallible test (there is no conceivable scenario under which that test will fail) and is therefore lame and phony - a quasi-test.
alexbunardzic.bsky.social
A plan is just a list of things that never happen.
alexbunardzic.bsky.social
Photos from my talk at Devoxx Belgium 2025 in Antwerp.
alexbunardzic.bsky.social
It's not AI's fault if your tests suck.
alexbunardzic.bsky.social
If all failing tests pass, you have a solid solution. If all tests pass and you still don't have solid solution, it's not the issue with the code, the issue is with the tests. Learn how to write quality tests. Don't blame AI if you don't know how to write solid tests.
alexbunardzic.bsky.social
AI cannot write the failing tests. AI is not equipped to do that. Only human experts should write failing tests.
alexbunardzic.bsky.social
"If light were to illuminate both itself and others, then certainly too darkness will conceal itself and others." Nagarjuna
alexbunardzic.bsky.social
Some people will definitely be squeezed out of a job if they decide that, nah, those XP practices are not something they'd be interested in. Fair enough, it's their prerogative and their choice. But it's not AI's fault that they choose that path.
alexbunardzic.bsky.social
Which includes XP, CI/CD, Trunk-Based Development, "Time to 'hello world' in production", walking skeleton, hexagonal architecture, limited work-in-progress, theory of constraints, Test Driven Development, mutation testing, etc., etc.
alexbunardzic.bsky.social
Just to be clear. AI isn't taking anyone's job in software. It's only that AI is now starting to apply serious pressure on the teams to pick up the slack and get serious about technical excellence.
alexbunardzic.bsky.social
History repeats itself. Same pattern played out 25 years ago, with the internet bubble burst following the dumb money fiasco.
alexbunardzic.bsky.social
It will be interesting to watch current AI naysayers start back paddling on their grandstanding after the current vacuous AI bubble bursts and the real AI-native businesses emerge from the ashes.
alexbunardzic.bsky.social
AI is augmentative (and yes, oftentimes it can be argumentative too; <eyeroll/>). AI basically augments whatever people bring to it.

If someone is disillusioned by AI, they are basically confessing that they are disillusioned by their own level of skill, insight, and knowledge.