Honestly, that’s what most web API’s are. You are just pushing data around. The “hard” part is that everyone has their own opinions on how it should be formatted.
And of course the minor inconvenience of having to give the user a way to make data entry easy, convenient and consistent.
But deep down it’s all spreadsheets. The faster you can wrap your head around that the easier programming is for you.
Spreadsheets are a UI view of tabular structured data with terrible kludges to fake hashes, with no real support for tree structures and heterogenous data structure hierarchies. And if the API you’re using is complex, namespaced XML because vendors, UI tends to obscure more than help.
Postman and the like are expensive intermediate abstractions between you and the underlying calls that I find a hindrance. For end users, forms can help, but even then you have to choose between something so generalized to effectively be a DSL like the old XForms spec or else build separate bespoke forms for each use case.
From a programmers perspective it takes a grand total of 2 minutes to do a bash/python script or jupyter notebook that exposes a semantically adequate interface. Maybe 5 minutes, if instead of just doing something quick and dirty yourself you decide to get into an argument with an LLM.
Honestly, that’s what most web API’s are. You are just pushing data around. The “hard” part is that everyone has their own opinions on how it should be formatted.
And of course the minor inconvenience of having to give the user a way to make data entry easy, convenient and consistent.
But deep down it’s all spreadsheets. The faster you can wrap your head around that the easier programming is for you.
Spreadsheets are a UI view of tabular structured data with terrible kludges to fake hashes, with no real support for tree structures and heterogenous data structure hierarchies. And if the API you’re using is complex, namespaced XML because vendors, UI tends to obscure more than help.
Postman and the like are expensive intermediate abstractions between you and the underlying calls that I find a hindrance. For end users, forms can help, but even then you have to choose between something so generalized to effectively be a DSL like the old XForms spec or else build separate bespoke forms for each use case.
Lua figured that out back in '93, their main data structure is the table
but from a programmers perspective - shouldn’t it feel natural - you just keep filling up forms doing the same work again and again?
From a programmers perspective it takes a grand total of 2 minutes to do a bash/python script or jupyter notebook that exposes a semantically adequate interface. Maybe 5 minutes, if instead of just doing something quick and dirty yourself you decide to get into an argument with an LLM.
What do you mean by natural? Do you want stateful APIs?
Oh no.
I already hate apis that requires tokens. It’s awful to use manually
Why the fuck would you use an API manually?
In development. You may want to test the endpoint using curl or something like that
Scripts and aliases make this easy to set up.