Goodbye, Clean Code

Re:Goodbye, Clean Code by b0rsuk on Sunday January 12, 2020 @05:24AM (#22025548)

From my experience, I would start with a more compact implementation first in hopes that it would save me time. This usually ends with me having to do more complicated operations in my head to plan the code. It's tempting. It's like trying to do math in your head to save time - often leads to a mistake. Now I believe there's no shame in trying out a solution first and rewriting it later when I get a feel what the problem is really about.

John Carmack once said something like "If you're not sure which version is better, code it both ways and compare".

The only downside of later optimization is that in commercial environment they often don't let you clean up.

[ Parent | on HN ]

Re:Goodbye, Clean Code by _pmf_ on Sunday January 12, 2020 @12:12PM (#22027426)

> The only downside of later optimization is that in commercial environment they often don't let you clean up.

Exactly (speaking for custom industrial automation projects). The follow up next requirements for a project might be in 5 or 7 years or never. You would be cleaning up a 80% dead project. In these environments, it's not uncommon to no longer have a dev environment because the system is a singular testbench that costs upwards of 2 million and only exists in 2 factories, so all bugfixes must be extremely conservative and any refactoring is stricly forbidden because there's no way to locally test for regressions.

[ Parent | on HN ]