Re:Vim is touch-typing on steroids
by wruza on Friday January 13, 2023 @12:42PM
(#34371094)
read documentation and other technical resources, conceptualize what I want to happen, and construct the corresponding logic flow, while trying to anticipate potential problems and corner cases. In other words, most of what I do doesn't involve touching the keyboard at all.
In my case, this often creates an elegant and conceptual code that doesn’t match expectations or even reality. Reminds me these meetings when someone e.g. says “no-brainer, there’s API for that” and goes on with requirements, schedules and administrative tasks without checking first if it’s hooks, streaming, polling with a rate limiter, or a non-idempotent email-like protocol with a completely undocumented bullshit in its payload section. The correct remark is “hold your horses”.
More than half of my work (strange corners of fintech) is checking, testing, prototyping, fail-refine cycles, unrealistic expectation reports. Not only APIs, also hw, niche software, reverse engineering, a little bit of everything. 80% of it goes straight into the “Attic” repo after few hours. I have to type an unelegant and unconceptual code fast and reshape it at the speed of thought about a realtime problem, there’s no time for clever design if a business-critical decision or a paper-heavy process depend on these experiments.
Using a boring editor feels like you want to quickly visit a few places to see if they are good enough, but they are all in different cities, so you start to forget details while you drive. And relying on a documentation is like buying a house by pictures alone.
For me the structure comes from the actual code, after I can extract its true shape, after the requirements got satisfied in a fully functional prototype, which is also a sort of a knowledge base for a project. I’m able to plan it in advance, but only with 20% probability and 80% of unnecessary work.
In my case, this often creates an elegant and conceptual code that doesn’t match expectations or even reality. Reminds me these meetings when someone e.g. says “no-brainer, there’s API for that” and goes on with requirements, schedules and administrative tasks without checking first if it’s hooks, streaming, polling with a rate limiter, or a non-idempotent email-like protocol with a completely undocumented bullshit in its payload section. The correct remark is “hold your horses”.
More than half of my work (strange corners of fintech) is checking, testing, prototyping, fail-refine cycles, unrealistic expectation reports. Not only APIs, also hw, niche software, reverse engineering, a little bit of everything. 80% of it goes straight into the “Attic” repo after few hours. I have to type an unelegant and unconceptual code fast and reshape it at the speed of thought about a realtime problem, there’s no time for clever design if a business-critical decision or a paper-heavy process depend on these experiments.
Using a boring editor feels like you want to quickly visit a few places to see if they are good enough, but they are all in different cities, so you start to forget details while you drive. And relying on a documentation is like buying a house by pictures alone.
For me the structure comes from the actual code, after I can extract its true shape, after the requirements got satisfied in a fully functional prototype, which is also a sort of a knowledge base for a project. I’m able to plan it in advance, but only with 20% probability and 80% of unnecessary work.