วีดีโอ

I feel like juat using rust (that does not have null.poinyers by default) is$way better than all these tricks

"It's gotta be ascii", excuse me what? Have you never heard of utf8?

QA is a joke. Inevitably, the QA department is less skilled than the engineers writing code. You have to be smarter than the person writing the code to find the bugs in their code. Mechanically doing checklists is not QA.

I want to shout out to my colleagues who will celebrate this video after it was posted today ;) I think we can all relate ;)

Seems pretty woo-woo. A case for "parse, don't validatate"? Obviously ifs are not evil.

the presentation is not visible

Anyone watching this in 2024 when the state of the Agile Industrial complex is even worse ?

2:29 "By the same token, Agile requires constant feedback and constant interaction with users which means that an actual end-user of your software must be in the room while you are developing. If you're not doing that, you're not doing Agile." By this definition I would guess that every company that isn't a startup and is over 15 employees is not doing Agile. Also by this def I would guess that 90% of companies which fall into the other groups I mentioned are not doing Agile. There are maybe 3 or 4 companies in the world you've ever heard of who are doing Agile. The other ones are complete unknowns -- a couple of broke people sitting in a dark room churning out code that they want (one is the user & the other is the dev): That's Agile. 😁

27:03 "The process itself is very simple. You start off by talking to your customers to find out what they need. And then you break off a small chunk of that and you build it & you give it to them. That's it." This is where everything falls apart & it is the core issue of all software development. Let me explain why. Users do not want to talk to you (at all or for very long) about what they need / want. I've had internal "customers" who ask for very specific features (even including a drawing of the WinForm design). I have taken that drawing & reproduced it in code exactly. I then create usage videos & provide the working code. After that...crickets. I hear nothing. Then 6 months later, I get a call, "Hey, why didn't you ever do X?" (They're referring to this thing that I did and documented with video.) I say, "Well, you never even watched the video & in 6 months you've never even tried the functionality that you begged for? Here's the thing, you need to right-click to get to that functionality. It is in the product." A sheepish reply, "Oh, I didn't know the functionality was there." You see? Users don't even try the functionality which they design and they beg for. I've worked in IT for > 33 years & I have no idea what world exists where a User sits in a room with Devs & works with them. Besides for 99% of products (external) how would you ever get a user to be available like that? It's not possible. That's why the best products have a Owner (who wants the product herself) who knows what the users want and drives the product to the vision. For example, Steve Jobs wanted a music player in his pocket and drove ruthlessly towards the iPod. That only exists in rare companies (like Spotify, Netflix, Apple (in past)). It doesn't exist in 99% of companies who pay developers to write code. They just write code & don't really care how the product works. And that is true in every (paid) role (QA, Project Mgr, Support, etc.) within the company. This all explains why True Agile rarely exists and why so much software is created which fails the user.

Allen shows issues with single inheritance in Java, and alternatives using interfaces and delegation. Java fortunately does not support multiple inheritance of Classes and I think it would be hard to use without violating most of the SOLID principles anyway.

If you still do big upfront design with UML and need to develop a system driven entirely by User input, and want to transition to Tests Driven Development, DbC could be a good intermediate step. However, it directly models real world Roles with Classes which I think is problematic, and also assumes that the Users will want to write behavioral tests, whereas they actually never have time to red them. I think this is well worth watching - it certainly gave me a lot to think about - but before attempting it, I would review Allen's more recent output to see if he even still advocates DbC and if so, what changes he's made to it. Personally, I don't see a need for this front end to TDD.

@21:28 Allen recommends "TDD by Example" by Kent Beck to learn about TDD. Part I is the core of the book and it uses Java. Part II uses Python. Part III is patterns which I skimmed. I didn't get much out of it the first time, so I watched lectures by Kent, Uncle Bob (Clean Code) and Ian Cooper (TDD). Then I went back and read it again and found it to be excellent. For more TDD examples with code, there are many TH-cams in Java, Python or C#.

Time sheets and billing codes may not improve production but are necessary for billing the customer and accounting for costs. Billing, taxes, and tracking actual profits are important to keeping the company alive. Any company that does not track those things eventually goes out of business. However, the company can use time clock programs with the appropriate functions to expedite the time tracking.

Sounds a bit over the edge there, this guy. Too angry, clearly overcompensating. Apart from any other statement here, a lot of which are plain wrong, his stance on documentation, I think, is the main hint to his immaturity.

I can't believe this is 9 f*cking years ago

And nobody ever mentions the significance of the first line of the agile manifesto: "We are uncovering better ways of developing software by doing it and helping others do it." This should be the main concern of the agile coaches. Uncovering better ways -> by listening to the teams, sprint the impediments and conducting experiments. And you have to be doing it, not floating about with PowerPoints and stakeholders. You need to be in the gemba.

I’ve been in the industry for more years than I can remember, and it puzzles me why people overlook the fundamental reason Agile was created. Initially, we had the Waterfall model, which involved working with a two-year timeline, a stack of requirements, and a strict deadline. However, software and solutions naturally evolve as you discover more, making development an iterative process. Waterfall represents one extreme, with rigid time slots and strict management. The other extreme is having no time slots, no accounting, and no management, which works fine until an auditor arrives, revealing the absence of teamwork. Unfortunately, we now have the worst of both worlds. Some believe that speed is paramount and that eliminating processes will achieve it, while management holds onto some form of structure. Both sides need to understand the core problem Agile addresses: coding blindly for two years, only to reveal software the customer dislikes, leading to frustration on both sides. Agile exists solely to prevent this. It has nothing to do with speed. We break down the code into small increments, show it to the customer, and proceed only if they are satisfied. This approach prevents us from developing for two years and finding it impossible to make changes. Agile is simply risk mitigation. It’s why Apple started with one small home-built hobby computer and released it incrementally to customers. It’s why McDonald’s didn’t spend ten years building 10,000 restaurants before opening but started with one. I appreciate this discussion to some extent, but I also dislike the notion that eliminating accounting, HR, and other departments would magically produce perfect software. We have these roles so developers can focus on coding without worrying about economics, audits, or legal issues

"In any sentence that you sentence where you see the word agile, if you can replace that word with the word Flexible, then you have a proper sentence" - I'm taking that quote, thank you very much. Still VERY relevant 9 years later.

Complain about grammatical errors. Proceed to make up a word like underabstraction and use the adjective uncohesive as a noun. 😱

Absolutely true, 2024 and it's still the biggest problem

Doesn't this mean that the null conditional access operator is the root of all evil? Actually it's null itself. This is why I like functional programming languages like F#, OCaml, Rust, ReScript, etc.

Evolution is not an efficient process - design wins.

For a little while now, I've been saying this with regard to comments: "Don't say 'what'... say 'why'!". I'm now thinking that this, specifically around the Dan North example, is also the way to go with naming too... Don't say "what"... say "why". My question with getters and setters is always... "if we've got a bunch of private properties and a bunch of getters and setters to read and write each one of them... why don't we just be done with it and have public (elpletive deleted) properties?" "When I split and infinitive, I expect it to stay split" - Raymond Chandler on editors.

I wish you'd re-edit this to show the right slides at the right time. :(

Tough crowd and extremely relevant material I Can hear them all screaming “not my data!” and “what about performance?”

So what do finance processes look like in an agile company? One of the things the finance team does is validate theres budget and validate where the money is going to (lots of scams exist to confuse people to send money to the wrong place).

8 fucking years later, the virus is still present

the moment you realise this guy is the director of QA

Flaw-lessss

At 6:40 Mr Holub makes a devastating observation: the whole organization has to be "Agile", not just the software developers. Coming from an Enterprise ISV that "forced marched" from "waterfall", classical development was painful, but Engineering did this within one year. What we discovered was that customers were not Agile. Their IT organizations could not deploy s/w developed in 90 day cycles. The back office end users rebelled against changes in the systems: their performance metrics degraded 15-20%. When one of your customers has 24,000 end users of your system, this is a problem. Their CEO calls your CEO. Lawyers get hired. The Business division owns the contract. The IT Division "owns" the operational success, but not the budget for it. Agile s/w delivery is seen as a solution to delivering CAPEX scope system changes within an OPEX budget. Customers wanted to pick and choose select feature changes as "system patches", not as full releases. Lastly, any Enterprise s/w system has to integrate with legacy and in-house s/e systems, hence the Corporate Customer's IT Division has to be Agile too.

How the tides have turned. C# is getting pretty popular nowadays. (WPF still sucks massively, though)

Cant believe the audience didnt laugh that much, "getLength, I did recieve a lot of spam suggesting this." LMFAO

Undisciplined

Very spot-on talk, I agree with the certification mills, agility of training and the dysfunctional tools points. The level of frustration in the voice tone most likely reflects actual experiences. ;)

Time traveling from 2024: At last an "Agile sux" video that makes sense. Agile doesnt suck, bad Scrum sucks. Allen pinpoints the problem - We need a culture of Agility. And Orgs are looking for a quick fix. And they love their extant culture. So change does not happen. What's most alarming is that this is 8 years old and still relevant.

I am still a believer in Scrum, but my job as a technology manager (despite many saying managers have no role -- I'm also a developer who still codes) is to protect the process and to constantly remind everybody (both on the team and in the business) of why we use it and what it is able to achieve, but only if we focus on doing it well and for the right reasons. One of the hardest things about Scrum is how to scale it, which is what inevitably led one company I worked at to embraces SAFe, which rubbed me the wrong way immediately. It was chock full of registered trademark symbols, and I saw through it immediately. But by this time, senior leadership had gotten a sniff of how effective my teams were, and they wanted that for everybody, and they had signed on the dotted line, and that was that. And that was not only the end of real agility in that department, but soon prompted me to leave. I'm still not ready to give up on Scrum, though, because I saw it work too well. I'm hopeful we can experience something similar at my current employer.

You can tell this individual has not had to deal with any form of tech archeology-figuring out what the fuck people in the past did. The whole not making adequate documentation is literally what happened with the developments of W76, W78 and W88 nuclear warheads via FOGBANK. Inadequate documentation cost billions upon tens of millions of dollars in wasted research that was forced to be done because they lost the means and methods to make the product in the first place. This is a common problem that happens particularly with long-standing technologies and it's one to clearly isn't addressed in this presentation. How can you be AGILE and also not leave a klusterfuck for future generations who now have to figure out what the fuck you and your peers were doing, how it works, and so on because documentation wasn't important because it "got in the way"/"was inefficient uses of time".

The US Air Force went through a similar thing with the Total Quality movement starting in the early nineties. The US Chief of Staff had heard that the Total Quality movement could work wonders in industry, and decided that he would apply it in the Air Force. It soon became clear that he only intended to use the terminology, while rejecting just about every substantive reform that TQ required. The whole effort was abandoned after he retired. There is now no indication in the Air Force that TQ was even attempted.

Yep, it could be true but... not every company is a software company and not every delivery process is a software build process. The basic mistake is that every company and every project inside IT corporations/companies will work with "software building" workflow and Agile methodology/frameworks. It will not. But this is all about very very very shallow understanding of how companies work.... It is like thinking that copying a file is a moving icon between folders. And when we make thing moving icon faster, the process will be faster.

Inconsistent audio level. Impossible to listen to in my car

The point about cloud aged like fine... Milk

Yes forget the word ‘Agile’ it’s really about the ‘Culture’. Unfortunately greed, money, unstable teams and poor social constructs can and will decimate the ‘Culture’. Until we find the solution to such human limitations we can only dream about the religion of Agile. In the meantime a Godless AI may supplant us all 😊😈

Nothing for a QA dept to do… oh sweet summer child. Actually a QA dept which can support dev teams whilst also maintaining system testing which tests the SYSTEM according to the system specifications or use cases known to be what the customer needs has obvious value. Even if you measure every grain, unless you’re taking a look at the actual system, you’ll probably end up with a heap and not a sandcastle. Unless your system is so small it fits into one small team where everyone knows everything…

"we do not have a hierarchy" a notion that is anathema to many business entities, not just mega corps. I once worked for a company that had maybe 40 employees. we in the technical staff were at the bottom of the org chart, and management (aka, the owners) wanted it that way. people with power will do anything to hold on to that power, and empowering their perceived underlings to take control of a major part of a business process is a threat to their power. the upper levels of the hierarchy will fight that tooth and nail. that's not some business school lesson, that's basic human nature.

Thank you very much, great class

Typical scrum project is like a waterfall without analysis and design phases. I think of it as a circle of poorly planned implementations followed by endless refactoring and bug-fixing. Scrum actually allowed and encouraged non-tech people to be directly involved in development process and it is one of the biggest problems with scrum. We can do scrum and it will work if we just replace non-tech SM and non-tech PO with Lead Dev who actually knows how development works and instead of having tons of useless meetings, just normally talk with our colleges during lunch or coffee. "Agile" should be just natural teamwork and cooperation.

Everything is true. Except the fact threat the real world complex project require integrated hardware and software which behaves unexpectedly and you tested but final system doesn’t work.

I use extended supers precisely because it has that brittle nature to it. Expecting subclasses to not have or have the wrong implementations and overrides. Essentially a glorified error generator and pattern enforcement tool (complexity enforcement). Downside is that those supers are fixed. They do not, nor should ever change only substituted and deprecated. Substitution of supers, while a bit consuming, is a very viable way to deal with it if you have well defined boundaries. That, if you still require the error machine.

Damn I didn't even do anything and I'm afraid.

Very good! Thanks!