cycle time is key

[Raoul Duke]

Things like Rust and C++ can create executables that can run fast.

But the development cycle is such a slow freaking nightmare?!!!!!

[Paul Tarvydas]

[a comment I made recently]:

It seems to me that for human-centric programming notation, we should be using something like a Forth/Lisp/whatever with embedded C/Haskell/Rust/etc instead of thinking of embedding some small language into a bigger language. Use something that encourages a Design and throw-it-all-away-and-start-again mentality. Our current workflow is too much like waterfall - we are encouraged to spend too much time up-front making decisions and wasting time preening the code before we have sufficiently explored the problem space. The current workflow encourages the code-is-precious attitude.

Kinda like, start each project with the source code for Sector Lisp, and install project-specific native functions into the source code to do "just enough" stuff like saving values. Iterate on the project until you're happy with it. Hand the whole thing over to a Production Engineer for speeding up the 5% that matters. Leave the rest coded in - memory efficient - Lisp.

[I, also, like the idea of writing the code more than once, except that the current bevy of languages don’t encourage one to do this.]

pt

[Mike Austin]

I also like exploring, prototyping and creating proof-of-concepts first, then when satisfied or time is up, start the implementation in a new sprint. It allows us to learn more about the problem, and to try different approaches and UI. When we're ready to implement, we have a much better idea of how long it will take to complete a feature.

[jido]

I agree with you, that is the best way to size up a job.