Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

> Every re-write I’ve ever seen or been a part of, has either failed or taken way longer and a lot more effort than initially projected.

On the face of it what you say is true, but I've seen (and I'm sure many have seen) situations where the codebase was so bad - or the underlying technologies so ancient - that a rewrite was the only practical way to get things back on track again.

Did the rewrite take way longer than initially projected? Yes.

Was it a lot more effort than initially projected? Yes.

Was it painful medicine that nonetheless left the business in a better place? Yes.



> Did the rewrite take way longer than initially projected? Yes.

> Was it a lot more effort than initially projected? Yes.

> Was it painful medicine that nonetheless left the business in a better place? Yes.

You make it sound like this is a net positive, but that is meaningless when you are not comparing it with the alternative:

> Would refactoring it be faster? Yes

> Would it require less effort? Yes

> Would it still leave the business in a better place? Yes

I don't think one can ever justify a rewrite unless

(1) the code is trivial to implement, or

(2) it is covered with extensive unit and system testing, or

(3) the technologies used are no longer available/supported, there is no upgrade path, and a new language/framework is absolutely necessary.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: