Re^5: Keeping, and advancing in, your job

[dragonchild]

Rewriting when you are assigned to a new script is one thing. Rewriting when you are assigned to a million-line application is another, especially if that application (like most applications) doesn't have a comprehensive test suite.

In addition, any change, no matter how small, has a certain risk associated with it. Larger changes have larger risks. A risk has a very clear dollar figure associated with it. It may be appropriate to leave an application in a crufty state, because of the dollar figures associated with changing it and not changing it. I work on an application that has that exact situation. It's crufty, horrible, and needs a complete rewrite. Except, that rewrite is being done by another group and I just need to keep it limping along for another year. It's one of my many tasks; I might spend 20 hours a month on it. Yes, I could rewrite it in that time, but that is wasted effort becaue the benefits to rewriting it are only gained by amortization - you spend $100 now to save $50/year over 5 years. You are still in the hole for two years and if you expect to turn the application off in a year, is it still worth it to rewrite?

[bwelch]

There's the difference. Given a thousand line script, I'd easily clean it up and make it work properly betting on the time saved in maintaining it. For ten-thousand line scripts, I'd question the source for having one script that long and be more hesitant to maintain it. Is that really a small script with many modules? That might be okay to maintain or it might be trouble.

A million line system and one person maintaining it? That sounds like a job (or a nightmare) all by itself. Are there many independent modules that don't need to be changed? At best, that would be difficult.