正在加载视频...

视频加载失败

Morning bathrobe rant: 20X.

206,662 次观看 • 3 个月前 •via X (Twitter)

33 条评论

Artificial Shitposting Intelligence 的头像
Artificial Shitposting Intelligence3 个月前

Can someone train an AI model whose sole job is to transform this guy's videos into normal talking

David Stevens 的头像
David Stevens3 个月前

I love the sound of uncle bob in the morning. sounds like victory.

Vincent Mayeski 的头像
Vincent Mayeski3 个月前

Yeah of course but what’s your point

Uncle Bob Martin 的头像
Uncle Bob Martin3 个月前

My point was to use the time you’ve been given very wisely.

Perry E. Metzger 的头像
Perry E. Metzger3 个月前

This is the way.

amine laariane 的头像
amine laariane3 个月前

i will make them write code so they can write code 100X faster and so on , i will be coding billions of light years ahead of everybody that this time tomorrow i will trigger the end of the world earlier. because i created everything that will ever be and that ever was .

Repojournal 的头像
Repojournal3 个月前

Is the ability to torture it and mutate it and heck even know what is happening not ingrained in you because you could read it and write it. If you get to a point where you can neither read it or write it will you be good in any of this? Losing the muscle memory to read and write code is losing the ability to do all of that. Even worse for those who never invested years doing all this to then say to them, don’t read or write means we will in the years to come not know anything about what computers are doing

Dhravya Shah 的头像
Dhravya Shah3 个月前

Yes, and?

Trevor Attema 的头像
Trevor Attema3 个月前

What a character! Love the performance.

jaimin 的头像
jaimin3 个月前

so should I write TESTS or not?

Uncle Bob Martin 的头像
Uncle Bob Martin3 个月前

I have my agents write unit tests. I don't read them. I have my agents write acceptance tests and QA procedures. I either read, or spot check them depending on the criticality of the project. I watch the agents execute and infer a mental model of their approach. I interrogate the agents to determine the internal structure and modular decomposition. I have the agents refactor structures that I don't like. I remain skeptical and austere. I work in short increments. At the end of each I run a manual test of what they have produced. I often find issues that I have them repair, and invent more tooling to detect such failures in the future. This is a never ending battle. I watch them like a hawk. I never trust anything they do. I trust the tools that I have given them, but do not trust them to use those tools. So I _manage_ the agents. I direct them, and then verify they follow my direction. It's a lot of work; but it pays off.

Northon Torga 的头像
Northon Torga3 个月前

right, so why bother with high level langs if we won’t be reading any of it? let’s drop to the bare min universal lang striving for closest to zero abstractions and dependencies

Pastor Soto 的头像
Pastor Soto3 个月前

I have seen this multiple times for many great people. Test driven development sounds like a great path to take!

Matt Wesney 的头像
Matt Wesney3 个月前

Yes sir

Bradford Ferguson 的头像
Bradford Ferguson3 个月前

Followed

ezenwa 的头像
ezenwa3 个月前

and by God, you will have test coverage 🤣

Ale𝕏 的头像
Ale𝕏3 个月前

Don't forget e2e test chief 👌

MEGA 的头像
MEGA3 个月前

The passion. Astounding morning motivation. 🤝

Dan White 的头像
Dan White3 个月前

We need AI generated uncle bob bathrobe rants

Chukwudi Nwobodo 的头像
Chukwudi Nwobodo3 个月前

Content is funny and educational at the same time 🤣

SloppySintax 的头像
SloppySintax3 个月前

🤪

John Connor 的头像
John Connor3 个月前

How old are you again? Give it up boomer.

Mike Garcia 的头像
Mike Garcia3 个月前

But are you running this systems in prod @unclebobmartin or just for your local env? 😂

Kurious dev 的头像
Kurious dev3 个月前

do you write tests after coding or before?

Jacob Enders 的头像
Jacob Enders3 个月前

Just to be clear, I will not be doing anything. The agents will be doing it.

Peter Dimov 的头像
Peter Dimov3 个月前

@unclebobmartin the other half outside of tests:

Kris Shkodrani 的头像
Kris Shkodrani3 个月前

Thats the correct attitude to use the extra time. I would love to see you build a harness if possible. I found it tricky to have agents write tests for nondeterministic cases. Sometimes the agents cheat 😄, so torturing the code and tests is the right thing to do.

Benjamin Smith 的头像
Benjamin Smith3 个月前

I've been wondering about all those legendary software ideas that have been talked about for decades, but never materialized. Modern day examples of the car that ran on water, or lightbulbs that last forever. Has Ted Nelson tried AI to finish Xanadu?

henry 的头像
henry3 个月前

You underestimate how proficient I am at having agents write awful tests that don't actually confirm any of the actual behavior I want my code to have.

Sasser 的头像
Sasser3 个月前

It is really funny how many people are mad that you can in fact ise your time as an engineer... engineering, instead of being nothing but a code monkey.

Hidden Layers 的头像
Hidden Layers3 个月前

The best code is code I understand. Code I understand needs less tests. For example my multithreaded architecture formally excludes race conditions. Absolutely, by definition. That's a lot of tests I don't have to write. But you, keep writing tests.

Igor Varga 的头像
Igor Varga3 个月前

Most people will use that extra time to doom scroll here. Rant - I couldn't describe this better myself. I like the Clark Kent vibes too.

Codewithoutai 的头像
Codewithoutai3 个月前

A world without AI is a sad one where engineers have to write their own code and rm -rf their own $HOME directories.

相关视频