4 ms·
Working with an old code base
- 10ren 17y agoJoel says don't rewrite, like the Netscape guys did http://www.joelonsoftware.com/articles/fog0000000069.html http://www.joelonsoftware.com/articles/fog0000000069.html Derek Sivers rewrote 90,000 lines of code in 12,000 lines... after 2 years on a failed rewrite. http://news.ycombinator.com/item?id=58429 http://news.ycombinator.com/item?id=58429 Maybe the lesson is that before you can renew your code, you yourself must be renewed? Or: You cannot solve a problem at the same level you created it.
- j_baker 17y agoIn my experience, a rewrite is almost always the wrong course of action. The best course of action is to just quit piling more and more unmaintainable hacks on top of it. And pay down your technical debt. Every time you find yourself thinking "boy, it sure would be easier if I could do it this way", stop wishing and make it happen. Sure, it's going to take you longer to get stuff done in the short term. But in the long run, you'll appreciate it.
- waivej 17y agoI think it just takes discipline to not change too much at once. I worked with two companies that tried to port from 16bit-32bit code. One did a rewrite. The other did it in sections. Neither survived the transition. Technology wasn't the problem. The world just didn't need their products anymore. Though my father had a rotten little shed that needed attention. He tore down and rebuilt sections. Besides a split level roof lines, it really seems like a complete redesign. I think the key is that few things were up for discussion. Also, having a sturdy base to work from made things easier. (and it stayed in operation the whole time.) With code, I see similar ideas. You probably shouldn't tear down and rewrite more than half at a time...and you should probably decide whether the world still needs the original product.