Computing On The Edge - 202332
Praise To The Headless
I've been maintaining for long that user interfaces are overhyped, they currently take much more work than they should.
I see a far future where there is an inventory of every possible user interaction, a standard, and boring libraries that shave off all frontend work.
But we're not there, obviously, and some blind faith in AI is even shifting the goal farther away into the future.
I think when we'll be there, every effort spent on user interfaces, that is non-business logic would be a total loss.
I no longer share the excitement whenever a brand new user framework pops up: we've seen so many of them already, it's just pushing a bit the ball further, it's probably no real progress.
And progress is what we need.
The progress we had in the ninenties was to actually gain a user interface of sort, the progress we need now is to avoid the need to code it altogether.
Dealing with portability, accessibility, localization, pixel-perfect round corners: Repetita Nocunt, this all has to stop somewhere, we need to focus on business logic, that is a desirable place where to spend our focus.
But while we wait for this bright future that, to be honest, I am not even sure we'll ever be able to experience in our own lifetime, the value of the efforts spent on user interfaces goes asymptotically to zero.
The time that will still be well spent in this post-apocalyptic world is the time on services and APIs that are reusable.
Headless services are in for the long run, they do not have the obsolescence of a node lib, they don't depend on your desktop background, on your font, on your operating system.
Implementing headless services is time well spent, because they will last.
There is much about the Microservices approach, you can also do them the wrong way, but one feature that they predominantly have is that they are headless. You don't make GUIs with microservices unless you are dominated by strange istincts.
When the frontend wars will be over, leaving place to some standard that is fully declarative and with a semantic approach, you will still be able to call your headless services, the good ones that make sense.
Therefore headless products and services are definitely worth investing time in. Maybe headless is not hip or fancy, but it is the real thing.
Reading Tips for the week
The Way To Go
Go structured logging with slog
Boring Security
Sign your git commits, no-nonsense
Authenticated encryption: why you need it and how it works
Unix Life
[git] [security] [api]