Actually, if you're even 10% serious about this, I have a suggestion to add; throwing away the obsolete concept of 'unit database'.
I saw a bit of what the GCB people did, and the code apparently got messy because they tried to cram special cases into a homogenous database.
And I do have an alternative to propose; though I don't have it in a 'written' format yet.
Quote:
|
I'll be the main programmer. Preferably the only programmer. I think I have experience enough at least for phase 1. We can develop specialist for GFX or sonar or doctrine language later, but the core should be one man work, so it fits nicely together.
|
I read the emphasized parts as
this project won't ever complete. If you can't modularize the core well enough, and provide stable interfaces between them, (which as a side effect would enable several programmers to cooperate without much problems*) you'll end up with spaghetti code whose maintainance will be a typical "Herculean Task" (tm?)
*Presuming communication, particularly in interface documentation, works properly.