I wonder who are these 40+ people that have downvoted this video. The discussion was constructive, focus on benefits and made by people with actual expertise. If you disagree, be courageous and produce some content to expose your counter arguments.
I was taught UML in college and I use it to express my ideas. But often I have found myself explain what UML is and what all of the lines an symbols mean. Some other times I have found myself scratching my head trying to remember what lines represent what type of association, even if I am the only one who understands the diagram. I think UML is great but I am very much interested in the next thing or tool that would help describe architecture.
On the obvious front: I think sometimes stating the obvious is valuable because sometimes even the obvious is overlooked or goes wrong. And the bit about the history is incredibly important: I'd be willing to guess that a lot of great ideas - in computing and elsewhere need several "attempts" before being entrenched: for example, functional programming is 60+ years old, but it wasn't until ~2010 that it escaped academia and the AI folks, it seems.
Academia itself switched away from functional. Marvin Minsky said they had very smart people who would write elegant self modifying functional code, but the industry needed more engineers. That meant they had to expand the circle to people who were not as bright. Imperative programming was easier to teach them, and matched the VN hardware.
@@-taz- I learnt discrete mathematics in college and found that functional programming to have that as a pre-requisite. Maybe imperative programming has faster learning curve to bring an output. FP has context to learn and master which takes time.
It'd have to be something with just enough detail to drive conversations, but not so detailed that you may as well write code. It's almost as if people like the idea of proving their code, but they won't like it enough to actually prove their code.
I wonder who are these 40+ people that have downvoted this video.
The discussion was constructive, focus on benefits and made by people with actual expertise. If you disagree, be courageous and produce some content to expose your counter arguments.
I was taught UML in college and I use it to express my ideas. But often I have found myself explain what UML is and what all of the lines an symbols mean. Some other times I have found myself scratching my head trying to remember what lines represent what type of association, even if I am the only one who understands the diagram. I think UML is great but I am very much interested in the next thing or tool that would help describe architecture.
"The" diagram? There are many kinds of UML diagrams. Some are even slightly useful.
Do take a look at sysml. It is the pragmatic younger brother of uml.
UML seems logical for small projects. But any real-world UML is too much and overwhelming
On the obvious front: I think sometimes stating the obvious is valuable because sometimes even the obvious is overlooked or goes wrong. And the bit about the history is incredibly important: I'd be willing to guess that a lot of great ideas - in computing and elsewhere need several "attempts" before being entrenched: for example, functional programming is 60+ years old, but it wasn't until ~2010 that it escaped academia and the AI folks, it seems.
Academia itself switched away from functional. Marvin Minsky said they had very smart people who would write elegant self modifying functional code, but the industry needed more engineers. That meant they had to expand the circle to people who were not as bright. Imperative programming was easier to teach them, and matched the VN hardware.
@@-taz- I learnt discrete mathematics in college and found that functional programming to have that as a pre-requisite. Maybe imperative programming has faster learning curve to bring an output. FP has context to learn and master which takes time.
Not only that, but sometimes you need to convince analysts or managers about what you're saying. This gives you validation to point to.
We could stand to have IDEs for architecture. And more than just a UML-based diagram making tool.
It'd have to be something with just enough detail to drive conversations, but not so detailed that you may as well write code. It's almost as if people like the idea of proving their code, but they won't like it enough to actually prove their code.
Structurizr kinda helps with providing sort of DSL/IDE
4:24 -> awesome answer!