teyc / codegen-talk Goto Github PK

Source code generation was a technique that was reasonably popular back in the 90s for the creation of N-Tier program, but has largely been subsumed by tools that generate proxies automatically on the client side, and ORMs on the server-side. Additionally, techniques like subclassing is reasonably effective in keeping the codebase tight.

Another popular use of code generation is the generation of state machines from source code. Again, improvements in programming languages with constructs like async/await in C# and Javascript mean these are less useful today.

However, they still retain some promise in the field of model driven engineering. The basic thrust of this is that large parts of a software is best written in a machine-like manner to improve readability and consistency. Code-generation ensures that consistency is maintained in large teams. As new ideas are introduced into the source code template, engineers will have to inspect the impact of these changes on all the files. While this sounds like hard work -- and it is. Over time, it ensures the entire code base is sound and consistently structured as if it were all written at the same time in the recent past.

1. Generates server.cs and client.js in dist

   $ gulp --gulpfile gulpfile-01.js

2. Finds custom sections and replaces them with custom code