If my PM tells me we are only doing triangles, I go out of my way to make that part extensible to accept any shape. Never EVER trust a PM when they say “we are only going to support X” because they are only thinking about the MVP.
ow worse. you’ve been waiting for years to be able to finally get rid of the legacy circle logic that keeps you pinned on old versions of everything. you got the confirmation from the PM.
you explained to them carefully and patiently how much the spread over many shapes is slowing you down, and you also acknowledged that you can keep doing that, or you could invest a bit into architectural changes that would make it more bearable, and you got the response: i understand, it’s not worth it, let’s finally have balls and deprecate the circles once and forever.
so with the feeling of wind in your back, you clean up your abstractions, simplify your logic, streamline your pipelines. and you finally get around to implementing some of the features that were just too hard to do safely with the obsolete pieces in place.
and then… you see another fucking circle again and then you just die inside.
“Can you do it quicker if we assume it only has to support squares?”
“Yes but I’m not going to assume that”
“Why not?”
“Who has to do the work to support circles when a circle shows up: you or me?”
I have also tried “Okay we save 2 days by avoiding circles, but now we have to spend 1 day adding validation and messaging to make sure everyone knows circles are unsupported. And now, from here until eternity, if we spend just two 30-minute meetings with 8 people to discuss supporting circles, we will have spent more time than just supporting circles from the start.”
I’ve been around long enough,I know. It is indeed good advice to make things extensible, within reason. I have written some over engineered things that ended up being a pain to maintain.
My best advice is to learn the domain as much as possible, then you’ll be more aware of both the concepts involved and the potential pitfalls. In the OP comic, knowing that the machines are capable of making other shapes would have helped predict the problem even when management says “we’re a triangle shop”.
I’m the guy that goes “fuck, another exception to add to my logic” after the PM swore that “we’re only ever doing triangles”
If my PM tells me we are only doing triangles, I go out of my way to make that part extensible to accept any shape. Never EVER trust a PM when they say “we are only going to support X” because they are only thinking about the MVP.
ow worse. you’ve been waiting for years to be able to finally get rid of the legacy circle logic that keeps you pinned on old versions of everything. you got the confirmation from the PM.
you explained to them carefully and patiently how much the spread over many shapes is slowing you down, and you also acknowledged that you can keep doing that, or you could invest a bit into architectural changes that would make it more bearable, and you got the response: i understand, it’s not worth it, let’s finally have balls and deprecate the circles once and forever.
so with the feeling of wind in your back, you clean up your abstractions, simplify your logic, streamline your pipelines. and you finally get around to implementing some of the features that were just too hard to do safely with the obsolete pieces in place.
and then… you see another fucking circle again and then you just die inside.
“Can you do it quicker if we assume it only has to support squares?”
“Yes but I’m not going to assume that”
“Why not?”
“Who has to do the work to support circles when a circle shows up: you or me?”
I have also tried “Okay we save 2 days by avoiding circles, but now we have to spend 1 day adding validation and messaging to make sure everyone knows circles are unsupported. And now, from here until eternity, if we spend just two 30-minute meetings with 8 people to discuss supporting circles, we will have spent more time than just supporting circles from the start.”
I’ve been around long enough,I know. It is indeed good advice to make things extensible, within reason. I have written some over engineered things that ended up being a pain to maintain.
My best advice is to learn the domain as much as possible, then you’ll be more aware of both the concepts involved and the potential pitfalls. In the OP comic, knowing that the machines are capable of making other shapes would have helped predict the problem even when management says “we’re a triangle shop”.