This comment was originally posted on Twitter

InfoQ: Effective API Design

InfoQ: Effective API Design

I’ve worked on a framework (hyperGIS) in flash for 7 years, and in that time re-written virtually from scratch a couple of times, from AS1 to AS2 and then to AS3. Though longevity is good, an overhaul every few years can also be good, rather than clinging on to libraries that you have invested so much sweat in, you can’t let them go.

Admitedly I’m just representing the flip side of the argument, not to be provocative, but because I would hate this article to stop people publishing what they think is cool, if not beautiful.

I’ve worked on a framework (hyperGIS) in flash for 7 years, and in that time re-written virtually from scratch a couple of times, from AS1 to AS2 and then to AS3. Though longevity is good, an overhaul every few years can also be good, rather than clinging on to libraries that you have invested so much sweat in, you can’t let them go.

Admitedly I’m just representing the flip side of the argument, not to be provocative, but because I would hate this article to stop people publishing what they think is cool, if not beautiful.

I’ve heard this often, and disagree. Hopefully I can disagree without resorting to shouty headlines?

ActionScripters tend to have a different perspective because their projects tend to be small scale and disposable. If you worked on a major project over many years, I think you’d have a different perspective. Changing requirements and coders over the life of a project means that there is a professional responsibility to maintain the highest level of readability possible. This is what R.C. Martin (Uncle Bob) calls being a ‘Craftsman’.

For me the problem is a premise to the sentence “I want to clean up the code before releasing it”. Why on earth would you write something dirty in the first place, if you are capable of something clean?

If a carpenter wanted to craft a cabinet, they wouldn’t just ‘knock one up’ then ‘sand it down’. They’d take care over every joist, every angle and every cut. When it was finished they would release it to the world, and it would be rightly admired.

Perhaps you want to write code quickly to prove a concept? If so, then to me it depends who you want to prove it to. If you want to prove to yourself something is possible, write code how you wish and don’t publish it. If you want to prove it to the world, how do I know the published code produces the result of the published SWF? Only if the general community can read the code, understand it and accept it does a ‘proof’ get close to the meaning of the word.

I’m not sure what to make of the reference to André Michelle. He’s obviously an absurdly talented developer, and I love to keep up with what he’s working on, but I can’t say I know any of his code. It may be the world’s ugliest code for all I know, in which case while I admire it, I couldn’t use it in any of my professional projects in case there were hidden errors I could do nothing about.

I’ve heard this often, and disagree. Hopefully I can disagree without resorting to shouty headlines?

ActionScripters tend to have a different perspective because their projects tend to be small scale and disposable. If you worked on a major project over many years, I think you’d have a different perspective. Changing requirements and coders over the life of a project means that there is a professional responsibility to maintain the highest level of readability possible. This is what R.C. Martin (Uncle Bob) calls being a ‘Craftsman’.

For me the problem is a premise to the sentence “I want to clean up the code before releasing it”. Why on earth would you write something dirty in the first place, if you are capable of something clean?

If a carpenter wanted to craft a cabinet, they wouldn’t just ‘knock one up’ then ‘sand it down’. They’d take care over every joist, every angle and every cut. When it was finished they would release it to the world, and it would be rightly admired.

Perhaps you want to write code quickly to prove a concept? If so, then to me it depends who you want to prove it to. If you want to prove to yourself something is possible, write code how you wish and don’t publish it. If you want to prove it to the world, how do I know the published code produces the result of the published SWF? Only if the general community can read the code, understand it and accept it does a ‘proof’ get close to the meaning of the word.

I’m not sure what to make of the reference to André Michelle. He’s obviously an absurdly talented developer, and I love to keep up with what he’s working on, but I can’t say I know any of his code. It may be the world’s ugliest code for all I know, in which case while I admire it, I couldn’t use it in any of my professional projects in case there were hidden errors I could do nothing about.

http://blog.andre-michelle.com/2005/as3-perspective-texturemapping/
(search for the word “clean” on the page)

(Good old André did go on to release the cleaned up version at a much later date, but anyone who reads his blog would have settled for the dirty version that day!)