Your options are an ugly wall that works or the beautiful lack of a wall.
Sure, though having gone through an entire monorepo refactoring of like half a million lines to basically destroy the codebase and switch from vue 2 to vue 3 among other things, it’s also possible to build the new, better designed wall right behind the old one, test like hell against that wall, and then shift that wall in when it’s ready in a planned release, ready for the issues that come because that wall isn’t quite like the old wall
This is a dangerous metaphor. Remove the old wall and it turns out the new beautiful wall was leaning against and supported by it.
I get what you mean, it’s just that the metaphor could support both perspectives.
Build the new wall airgapped from the old one
And keep both walls for redundancy.
Ah, yes: weaponizing cybersecurity requirements to trick - I mean “motivate” - higher management to do things “right.”
* For a very specific value of “works”. Offer is limited in time. No refunds. Warranty doesn’t cover damage caused by the elements, presence, or lack of gravity. Other conditions may apply.
“Look, guys, I vibecoded a wall!”
“It goes almost all the way to the ceiling. I just need you to connect the last layer into the structure, it should be a 15 minutes work, right? I already settled the deadline with your boss. Tanks; bye.”
That’s just the unreal engine
“Waterfall? Won’t the water break our computers?”
I’m not gonna lie, I like how that wall looks lol.
Me when Claude code writes bullshit I don’t need
With the difference that code actually is easily changeable, if you need to change a wall you actually need to tear it down, get rid of the waste, likely tear everything down that was supported by it, and completely rebuild it with new materials.
If you need to change the foundations of your code, yes, it’s not always super easy, but it is actually possible and I’ve done it before without super big headaches.
It’s just not comparable.
It absolutely is, you can partially tear down even a load bearing wall but it’s expensive, difficult, and should have been avoided from the start.
Kind of like building a whole infrastructure on spaghetti code.
