Re:LibreOffice: cleaning and re-factoring a giant code-base ...
by tomelders on Sunday February 17, 2013 @08:32AM
(#5235129)
Maybe, maybe not. I've never been presented with any evidence that supports that view. I'm not saying you're wrong, I'm just saying I've got no reason to believe you're right. And vice versa.
Re:LibreOffice: cleaning and re-factoring a giant code-base ...
by jodrellblank on Sunday February 17, 2013 @08:46AM
(#5235166)
http://www.joelonsoftware.com/articles/fog0000000069.html
It is just someone else saying the same thing, or it's a real world example, or it's anecdotes-aren't-data.
Code becomes legacy pretty quickly, in my experience... Architectural decisions that were obvious before you had real customers rarely stand up to real usage.
The company that's "going to put you out of business" also doesn't currently have any working or revenue-generating code, and are unlikely to immediately understand the architectural challenges...
> I think you need to have a really good excuse not to completely rebuild your product from scratch every now and again
It pretty much just doesn't work in the real world; how's that for an excuse? As other people have said, you need to scaffold interfaces, and rebuild small parts. Any company that isn't continually reinvesting in making their codebase better will accumulate technical debt pretty fast; but the throw it away and start from scratch approach rarely works on big systems.