Localization as an afterthought-- screwed?

Question

So I signed on with a startup web development company as a subcontractor. They are putting together a large, complex user/product management system for a company that needs to support multiple levels of hierarchial localization. I signed a 3 month contract, and upon looking at their code, wish I hadn't.

They opted to write their own MVC framework (I guess the client company didn't want to use a prewritten one) and it's extremely poorly written. There's SQL scattered throughout almost every model view and controller (and there's no parameter-based find methods, it's all SQL) and they haven't even THOUGHT about localization yet-- something that will have an affect on nearly EVERY query.

The due date is 4 months away, and I honestly think we'd make good progress by scrapping the whole thing and going with CakePHP. Have any of you been in a similar situation, and what did you do?

PS: This is written in PHP/MySQL.

Answer 1

Rewriting from scratch is almost always the wrong idea (Read Joel's article Things You Should Never Do -- Part 1). But you have to start implementing small changes (big in your case). The sooner you get the code cleaned up the better it is for developing in... Get a sane data access layer and try to sort out the ad-hoc MVC first.

Also, while you are fighting battles, start planning out what needs to happen and get the upper brass used to the idea that some features might not make it in. I don't know your code base, or big or how messy it is... but if it is really bad then they should be prepared to ship a product minus some features rather than a half-baked solution.

Take charge on these issues, you (like the rest of us) get the pleasure of inheriting the sins of a codebase's previous owners. Is there any chance they would extend the contract or pay overtime for extra work you put in?

Answer 2

When you work for someone, you play by their rules. I've worked on a couple of sites using Expression Engine. I hated every minute of it, and I strongly recommend against it every time it comes up with my client, but what can you do, other than grind through the work until you're done?

Feel free to offer the suggestion about CakePHP, but I would drop it if nobody wants to hear it. Be happy you're only on for 3 months.