I don't get why they need the lines; LogiSim let's you just label connections between components without a visible line connecting them, making the layout cleaner
LogiX in NeosVR has a similar ability but could be a whole subgraph if you want as well to essentially make function bricks. It's a hard medium to translate so boxes with bezier curve cables is where everyone starts because seeing a line to follow the flow of execution is the major value to programming this way.
There is no automated unit testing in visual flow. You can see every event/step and the state of data to know how it works and debugs itself real time and usually smaller than in the presentation. They usually have infinite loop/race detection/kill and auto disables/disconnects the wire in the flow to the node that started fatal error or visually points to it. Basically if you are allowed to wire it up it's a valid usage to the compiler even if it's not an efficient program just like real code but can have safety rails to keep from crashing the environment with cpu/mem limiters or validation/constraints baked into each node. Some are better than others and usually drive some virtual machine backed in another language. I like Logix from NeosVR's sim engine which is backed in C# and exposes their simulator engine API as well as websocket/http interfaces for external data i/o all programmed in sim running real time. They are usually color coded to be intuitively type safe but you are physically not allowed to connect things like an output that's a string to an input that requires an integer. You can read/manipulate the state of data real-time at any node's input/output ports as well as preview inside the node processing it on some systems that program like this. You can see quickly where a data pipeline is going wrong. Instead of adding logging or debugging wait hooks you just pull a read wire out of the data stream anywhere you want or alter input variables as it runs to tune a desired behavior. You can hook in anywhere on demand to benchmark the whole flow or a segment or a single node function to measure performance if needed. Having said all that it's a great way to rapid prototype or make a boilerplate generator or as user friendly control widgets for apps like Blender. It's a great medium to work out a problem and hand a working sample to a developer that can see the whole process as envisioned and refactor it in whatever language they need to make their product. The paradigm goes beyond programming and is starting to be used in network/data center systems or orchestrating microservice apps like Node Red.
I don't get why they need the lines; LogiSim let's you just label connections between components without a visible line connecting them, making the layout cleaner
LogiX in NeosVR has a similar ability but could be a whole subgraph if you want as well to essentially make function bricks. It's a hard medium to translate so boxes with bezier curve cables is where everyone starts because seeing a line to follow the flow of execution is the major value to programming this way.
Who got the reference at 30:01 ?
Testing is easier means what? I assume he's not talking about automated unit testing.
There is no automated unit testing in visual flow. You can see every event/step and the state of data to know how it works and debugs itself real time and usually smaller than in the presentation. They usually have infinite loop/race detection/kill and auto disables/disconnects the wire in the flow to the node that started fatal error or visually points to it. Basically if you are allowed to wire it up it's a valid usage to the compiler even if it's not an efficient program just like real code but can have safety rails to keep from crashing the environment with cpu/mem limiters or validation/constraints baked into each node. Some are better than others and usually drive some virtual machine backed in another language. I like Logix from NeosVR's sim engine which is backed in C# and exposes their simulator engine API as well as websocket/http interfaces for external data i/o all programmed in sim running real time.
They are usually color coded to be intuitively type safe but you are physically not allowed to connect things like an output that's a string to an input that requires an integer. You can read/manipulate the state of data real-time at any node's input/output ports as well as preview inside the node processing it on some systems that program like this. You can see quickly where a data pipeline is going wrong. Instead of adding logging or debugging wait hooks you just pull a read wire out of the data stream anywhere you want or alter input variables as it runs to tune a desired behavior. You can hook in anywhere on demand to benchmark the whole flow or a segment or a single node function to measure performance if needed.
Having said all that it's a great way to rapid prototype or make a boilerplate generator or as user friendly control widgets for apps like Blender. It's a great medium to work out a problem and hand a working sample to a developer that can see the whole process as envisioned and refactor it in whatever language they need to make their product. The paradigm goes beyond programming and is starting to be used in network/data center systems or orchestrating microservice apps like Node Red.
For the love of all that is holy, please...stop...going uuuuhhhh...all the time!
ehhh...uhmmmm ehhhh... uhmmmmehhh.
The way these ppl speak is very irritating. eeeehhh
You mean they have so much on their mind and knowledge that they can't just say the first dumb thing that floats to the top of their head?