Architecture is a set of trade offs you use to try to minimize the costs of building, or maintaining software. The moment you decide on one, you've precluded certain approaches, it's hoped those were the poor choices.
We're 50+ years into null terminated strings
We're 40+ years into Operating System architectures that favor ambient authority.
We're 30+ years into the Web, and HTML that doesn't let you mark up (annotate) existing hypertext without copy/modifying it.
We're 20+ years into trying to make everything work through a thin unreliable network connection onto a machine with a small glass screen and no keyboard
There's a civilization ending amount of debt, while your project is important to you, know that in the big scheme of things, it really doesn't matter, and you'll be forgiven. 8)
If both React and Angular are installed and actively controlling different parts of the UI (seen it) I think it’s worth spending the ~8 hours on the replatform.
If the main database is a mix of JSON saved in S3 and “the good data” is in mongodb atlas, it’s worth spending the ~12 hours on the migration.
But if it’s going from one fotm framework to another, or syntactic sugar type changes, I don’t think it’s worth sinking time into.
Should be a clear problem:solution to any architectural choice starting with the core problem.
These meme only makes sense for very small products.
After a certain threshold of complexity and size any software becomes a collection of trade offs and compromises, never something you might call a “clean architecture”. Its a red herring, myth, utopia.
It’s a thousand years old attempt to calculate and reduce the complexity of real world interactions and problems to a bunch of mathematically elegant metaphors and formulas.
The threshold is usually very low - first bunch of real world paying customers, real world usecases, startup market fit etc.
We're 50+ years into null terminated strings
We're 40+ years into Operating System architectures that favor ambient authority.
We're 30+ years into the Web, and HTML that doesn't let you mark up (annotate) existing hypertext without copy/modifying it.
We're 20+ years into trying to make everything work through a thin unreliable network connection onto a machine with a small glass screen and no keyboard
There's a civilization ending amount of debt, while your project is important to you, know that in the big scheme of things, it really doesn't matter, and you'll be forgiven. 8)
If the main database is a mix of JSON saved in S3 and “the good data” is in mongodb atlas, it’s worth spending the ~12 hours on the migration.
But if it’s going from one fotm framework to another, or syntactic sugar type changes, I don’t think it’s worth sinking time into.
Should be a clear problem:solution to any architectural choice starting with the core problem.
After a certain threshold of complexity and size any software becomes a collection of trade offs and compromises, never something you might call a “clean architecture”. Its a red herring, myth, utopia.
It’s a thousand years old attempt to calculate and reduce the complexity of real world interactions and problems to a bunch of mathematically elegant metaphors and formulas.
The threshold is usually very low - first bunch of real world paying customers, real world usecases, startup market fit etc.